Read RSS and Atom feeds, and speak about them only when asked (#258)

internal/rss parses RSS 2.0 and Atom, and polls each configured feed on its own
interval; internal/webfetch is the one door either of them uses to touch the
network. The poller writes items as notes with source "rss:<feed>" and nothing
else: the answer path reads them back when he asks "что нового в лентах?", and
nothing is announced on arrival. A feed that dispatched would be a nag, which is
why the plan's breaking-news rule was left out rather than built.

webfetch is where the limits live, as code rather than a paragraph: http(s)
only, an allowlist (the configured feeds' hosts) and a denylist, a 2 MiB body
cap, a 3-redirect cap, one request per host per second, and a refusal to connect
to any private address — checked in the dialer's Control hook so it holds for
every resolved address and every redirect hop, not just for a literal IP.

Off unless configured: no "feeds" block, no poller, no outbound request. How far
a feed was read is a config fact (rss:latest:<name>), so a restart does not
re-note yesterday's headlines.

.gitignore

@@ -6,6 +6,8 @@
 /mavweb
 /mavpoll
 /mavcaldav
+/mavwaked
+/mavmaild
 
 # Certs (private keys, don't commit)
 certs/
@@ -33,6 +35,10 @@ deps
 deploy/db_key.env
 # Deploy secret (telegram bot token + chat id) — never commit
 deploy/telegram.env
+# zenmoney API token, read by mavpoll (never in argv, never committed)
+deploy/zenmoney.token
+# IMAP password, read by mavmaild (never in argv, never committed)
+deploy/imap.password
 
 # Temp files
 /tmp/

AGENTS.md

@@ -43,14 +43,17 @@ notes. Without it, the floor `HashEmbedder` is used — deterministic but weak
 (Russian recall rarely clears the confidence gate, many commands fall to
 "clarify").
 
-**Download the embedder** (ONNX, ~90 MB):
+**Download the embedder** (ONNX, ~120 MB):
 
 ```sh
 make download-embedder
 ```
 
-This fetches `paraphrase-multilingual-MiniLM-L12-v2` (384-dim, 12-layer,
+This fetches `multilingual-e5-small` (384-dim, 12-layer, Russian and English)
-supports 50+ languages including Russian) to `models/embedder/`.
+to `models/embedder/multilingual-e5-small/`. It is an asymmetric retrieval
+model: the code puts `query: ` in front of a question and `passage: ` in front
+of a stored note, which is how e5 was trained. The quantized file is the one
+that is downloaded, deployed and measured.
 
 **Also need ONNX Runtime** (`libonnxruntime.so`):
 
@@ -64,8 +67,8 @@ sudo cp onnxruntime-linux-x64-1.15.1/lib/libonnxruntime.so* /usr/local/lib/
 ```json
 "voice": {
   "embedder": {
-    "model_path": "models/embedder/model_quantized.onnx",
+    "model_path": "models/embedder/multilingual-e5-small/model_quantized.onnx",
-    "tokenizer_path": "models/embedder/tokenizer.json",
+    "tokenizer_path": "models/embedder/multilingual-e5-small/tokenizer.json",
     "lib_path": "/usr/local/lib/libonnxruntime.so"
   }
 }

CLAUDE.md

@@ -7,9 +7,20 @@ talking over unix sockets; one resident small model for routing + phrasing; whis
 Deploy target is a Ryzen laptop (homesrv) with Vulkan offload to the Vega iGPU (`n_gpu_layers: 99`,
 compose passes `/dev/dri` + the render gid) — the resident model stays ≤1.7B either way.
 
-**Resident model:** currently **Qwen3.5-0.8B** (`Q4_K_M`), the smallest checkpoint in the gguf
+**Resident model:** currently **Qwen3-1.7B** (`UD-Q4_K_XL`), stock — not yet the CPT'd one.
-library, picked for CPU/iGPU latency. The **target** is the locally CPT'd **Qwen3-1.7B**; that
+It replaced Qwen3.5-0.8B on 2026-07-31 because it measured better on both fixtures we have:
-training is still in flight (Vikunja #122), so no such gguf exists yet. Model files live in
+67.5% vs 59.7% intent-only on the 77-case RU routing fixture, and 20/27 vs 11-17/27 on the
+talk fixture. See `MODEL-BAKEOFF-31-07-2026.md`. It is a Thinking variant, so `n_ctx` is 4096
+— reasoning tokens need the room, and 4096 is what the scores above were measured at.
+
+The **target** is still the locally CPT'd **Qwen3-1.7B** (Vikunja #122, training in flight).
+Stock already speaks good Russian; what it gets wrong is the persona — it writes `я рад`,
+masculine, where Maven needs `рада`. That is what the CPT is for.
+
+**Do not bother with sub-500M models.** LFM2.5-230M and 350M were measured on 2026-07-31 and
+both are unusable in Russian: the 350M routes at 5.2% (worse than guessing) and answers
+"столица Франции?" with the invented non-word "Сторзит"; the 230M replies to Russian in
+Spanish. Their strong published IFEval/BFCL numbers are English-only. Model files live in
 `/mnt/hdd1/llms`, bind-mounted to `/opt/maven/models/llm` — which **shadows** the repo's
 `models/llm/`, so the LFM2.5 gguf sitting there is not loaded by anything. Swapping the resident
 model is a one-line change to `phraser.model_path` in `deploy/mavend.json`.
@@ -23,7 +34,7 @@ CGO daemons (`mavend`, `mavsttd`, `mavttsd`, `mavenclient`) need the vendored to
 and libs wired through the Makefile — **do not** call `go build` on them bare, use `make`:
 
 ```sh
-make build          # all 8 binaries
+make build          # all 9 binaries
 make build-web      # single daemon (pure-Go ones: web/waked/poll/caldav build without CGO)
 make test           # go test -race across ./internal/... ./cmd/... with CGO env set
 ```
@@ -51,6 +62,7 @@ Pure-Go packages (`router`, `memory`, `mavweb`, …) run under a plain `go test
 | `mavenclient` | Voice loop client (mic → stt → core → tts). |
 | `mavpoll` | Telegram long-poll reach. |
 | `mavcaldav` | CalDAV calendar sync. |
+| `mavmaild` | Mail reader (IMAP, read-only). Holds the IMAP password; core never sees it. |
 
 Daemons are wired socket-to-socket, not linked. `internal/ipc` is the client/server wire
 protocol; the config in `deploy/mavend.json` (with `${VAR}` env expansion from gitignored
@@ -58,19 +70,41 @@ protocol; the config in `deploy/mavend.json` (with `${VAR}` env expansion from g
 
 ## Routing — read this before touching the router
 
-`internal/router/` has TWO layered engines and the committed default is an **interim
+`internal/router/` has TWO layered engines. **The LLM router is now the default and it is
-stopgap, not the intended design** (see memory `routing-architecture-target`):
+on in deploy** — this section used to say it was wired `nil`, which stopped being true on
+2026-07-31.
 
-- **Target (REARCH.md):** LLM-as-router. One resident Qwen3-1.7B (`llmrouter.go`) emits
+- **LLM router (the intended design, REARCH.md):** the resident Qwen3-1.7B (`llmrouter.go`)
-  GBNF-constrained structured JSON, and the SAME model phrases replies. Embedder is demoted
+  emits GBNF-constrained structured JSON, and the SAME model phrases replies. Embedder is
-  from a routing gate to a RAG hint.
+  demoted from a routing gate to a RAG hint. Wired at `voice.go:214` via
-- **Current stopgap:** `llmrouter` is wired `nil` (around `voice.go`), so the
+  `pickLLMRouter(cfg.Voice.UseLLMRouter(), llmClient)`; the flag is `voice.llm_router`
-  `classifier.go` + `embedder.go` nearest-neighbour cascade actually runs. It routes by
+  (`config.go`), `DefaultLLMRouter` is **on**, and `deploy/mavend.json` sets it `true`.
-  similarity to frozen seed phrases — the known cause of weak RU query handling.
+- **Classifier cascade (the failure floor, not dead code):** `classifier.go` +
+  `embedder.go` nearest-neighbour over frozen seed phrases. It runs when the LLM router is
+  off, when there is no llama-server to talk to (`pickLLMRouter` logs that and degrades),
+  and on any per-turn LLM error. Do not delete it — routing by seed similarity is the known
+  cause of weak RU query handling, but a turn must never break on the model.
 
 Cascade order: `stage0.go` exact-match fast-path → LLM router (when non-nil) → classifier
 fallback. Any LLM error falls through to the classifier so a turn never breaks on the model.
 
+Measured on the 77-case RU fixture (`MODEL-BAKEOFF-31-07-2026.md`): the classifier scores
+36.8% full accuracy at p50 31ms; Qwen3-1.7B scores 67.5% intent-only / 72.7% through the
+cascade at p50 ≈2.7s. Accuracy roughly doubled, latency is ~90× worse, and that trade was
+accepted deliberately. `Confidence: 1.0` used to be hardcoded in `llmrouter.go`, so the LLM
+path could never ask for clarification (6/6 refusal cases missed on the fixture) — Vikunja
+#359. Fixed 31-07-2026 with structural signal (single-token utterance, keyless fact, act with
+no allowlisted fn) feeding the same stage-3 gate the classifier path already had — see
+`gateLLMDecision` in `router.go`. Note the second half of that bug: the LLM branch never
+consulted `r.threshold` at all, so a correct low confidence would have been discarded anyway.
+
+Re-measured on the fixture after the fix: **missed clarify 6/6 → 1**, at the cost of 3 false
+clarifies and 2.6pt of full accuracy (72.7% → 70.1%, intent-only 67.5% → 74.0%). Two of the
+three false clarifies are acts the model mis-routed and the gate caught — asking beats wrongly
+executing, so the fixture and the daemon disagree about what is correct there. The third,
+`"поужинал"`, is a real defect: **the single-token rule is an English intuition and does not
+transfer to Russian**, where one word is routinely a whole sentence. Narrow or drop it.
+
 ## LLM output contract
 
 All phrasing paths emit `{"response":"...","mood":"..."}` (parsed in `replier_llm.go` and
@@ -82,8 +116,26 @@ workspace enforces that the Go and relabelling prompts remain identical.
 
 ## Non-goals (hard constraints)
 
-Never phones home. Not a nag, not autonomous. Maven's persona is **feminine** — Russian
+Not a nag, not autonomous. Maven's persona is **feminine** — Russian
-self-reference must use feminine forms (the user is male; see memory `maven-persona-gender`).
+self-reference must use feminine forms — `рада`, not `рад`; `поняла`, not `понял`. The owner
+is male and is addressed informally: "ты", singular, never "вы"/"ваш" and never "он"/"его"
+(she talks TO him, not about him). Pet names ("милый", "дорогой") are forbidden; his name
+("Ками") is not. The eval enforces this: `CheckAddress`, `CheckFeminine` and `CheckCringe` in
+`internal/phraser/eval/checks.go`, scored by `make eval-phrasing`.
+
+**"Never phones home" is DEPRECATED** (owner's call, 2026-07-31). It used to be a hard
+constraint and it is not one any more: a 0.8B — and a 1.7B — does not know enough to answer
+world questions, so she needs to read external sources. What replaces it:
+
+- **No telemetry, no cloud model, no third-party account.** That part never changes. Nothing
+  about Maven is reported to anyone, and inference stays on the box.
+- **Local sources first.** Kiwix ZIMs on homesrv (Wikipedia, ifixit) before anything on the
+  network. Reading beats recalling for a small model, and a local read costs nothing.
+- **External search is allowed and off unless configured**, like the weather and telegram
+  capabilities.
+- **His notes and facts are never search input.** Looking up why the sky is blue and sending
+  his stored personal notes to an upstream engine are different acts. Only the utterance goes
+  out, never the persona block, history, or matched notes.
 
 ## Web UI conventions
 

DESIGN.md

@@ -15,7 +15,8 @@
 
 **Maven** — self-hosted personal assistant. Manages your day, acts on your
 homelab. One daemon on homesrv (always-on, not the workstation), multiple
-client surfaces. All local, never phones home.
+client surfaces. Inference and data stay on the box; she may READ external
+sources (see Non-goals — "never phones home" is deprecated).
 
 Primary name is "Maven", with feminine-gendered Russian self-reference
 ("она", "меня", "помогла"). Clients may choose their own UI label. Consistent
@@ -35,8 +36,13 @@ Inside boundary — the ones that actually constrain the build:
   she records. A confident wrong fact is worse than a known gap.
 - **Not a nag** — she'd rather miss a nudge than be mutable. Shuts up when
   uncertain. Load-bearing.
-- **Not a stranger** — runs on your stuff, your model, your data. Never
+- **Not a stranger** — runs on your stuff, your model, your data. No
-  phones home.
+  telemetry, no cloud model, no third-party account. She may READ external
+  sources to answer world questions (Kiwix first, then optional search); she
+  never reports anything about you to anyone, and your notes and facts are
+  never used as search input. **"Never phones home" as an absolute is
+  deprecated** — owner's call, 2026-07-31: a small model does not know enough
+  to be useful without reading.
 - **Not a relationship** — mom-tone is a function that makes nudges land, not
   emotional company. Names the drift a warm small model falls into.
 
@@ -458,7 +464,7 @@ decides *insistence*. Both are needed.
 
 sev ≤ 2 drops on away, sev ≥ 3 holds: a missed water nudge is noise, a missed
 backup failure isn't. Away-channels (ntfy/telegram) leave the box — the one
-path that crosses "never phones home," through your own relay. **Minimal
+path that leaves the box for a person to see, through your own relay. **Minimal
 body** — "disk low on homesrv," not detail; don't make notifications a
 shoulder-surf exfil surface.
 

Dockerfile

@@ -51,7 +51,8 @@ RUN go build -o /out/mavend    ./cmd/mavend    && \
     go build -o /out/mavttsd   ./cmd/mavttsd   && \
     go build -o /out/mavweb    ./cmd/mavweb    && \
     go build -o /out/mavpoll   ./cmd/mavpoll   && \
-    go build -o /out/mavcaldav ./cmd/mavcaldav
+    go build -o /out/mavcaldav ./cmd/mavcaldav && \
+    go build -o /out/mavmaild  ./cmd/mavmaild
 
 # llama.cpp Vulkan build — the phraser/router LFM engine (llama-server). Built
 # from source (not a prebuilt vendored blob) so the binary's glibc/GLIBCXX match

MODEL-BAKEOFF-31-07-2026.md

@@ -0,0 +1,223 @@
+# Resident model bake-off — 31-07-2026
+
+**Outcome: the resident model is stock Qwen3-1.7B** (`UD-Q4_K_XL`). Two sweeps ran this
+evening and the second one changed the answer — read to the end before acting on any table
+here. [Second sweep](#second-sweep-same-evening--five-models-and-a-resident-model-change)
+is the one that holds.
+
+## First sweep — LFM2.5-1.2B vs Qwen3.5-0.8B
+
+**Verdict, scoped to this pair: keep Qwen3.5-0.8B over LFM2.5-1.2B.** LFM2.5-1.2B is worse
+at routing (52.6% vs 60.5% intent accuracy), and the loss is almost entirely Russian
+(18/61 vs 22/61 RU, while EN is a wash). It is also 2.4× slower. The Thinking variant is
+far worse again. This verdict still stands as written — it rejects LFM2.5-1.2B. It is
+**not** a recommendation to keep 0.8B as the resident model; the second sweep replaced it
+with Qwen3-1.7B.
+
+Settles Vikunja **#278 / #250**.
+
+- Same fixture and scorer as `ROUTING-EVAL-31-07-2026.md`: `internal/router/eval/`
+  (`ru_routing_v1.json`, 76 held-out cases).
+- Reproduce: `MAVEN_LLM_URL=http://127.0.0.1:<port> make eval-router`
+  (`TestLLMRouterBaseline`). (This line used to say there is no `make eval-models` target.
+  There is one now — start a server with the gguf you want, then
+  `make eval-models MAVEN_LLM_URL=http://127.0.0.1:<port>`. It runs only the LLM test, since
+  the classifier baselines do not depend on the model.)
+- All three models served by the same `llama-server` flags — `-c 2048 -ngl 99 -t 6`, only
+  `-m` and `--port` differ. One server at a time on an otherwise idle box, so latencies are
+  real and not contention.
+- Measured on top of the router prompt fix (`origin/overnight/router-prompt` merged in), so
+  the Qwen column is directly comparable to the numbers already recorded.
+
+## Results
+
+`llm-only` — the model alone. This is the column that measures the model.
+
+| | Qwen3.5-0.8B | LFM2.5-1.2B Instruct | LFM2.5-1.2B Thinking |
+|---|---|---|---|
+| **intent-only accuracy** | **60.5%** | 52.6% | 36.8% |
+| full accuracy (intent+slots+gate) | **36.8%** | 32.9% | 21.1% |
+| **RU** | **22/61** | 18/61 | 10/61 |
+| EN | 6/15 | **7/15** | 6/15 |
+| route errors | 0 | 0 | 0 |
+| **p50 / p95 latency** | **1.05s / 1.71s** | 2.47s / 3.62s | 2.42s / 3.24s |
+| missed clarify | 6 / 6 | 6 / 6 | 6 / 6 |
+
+`cascade+llm` — stage-0 → model → classifier floor, what #320 would actually ship. Same
+ordering.
+
+| | Qwen3.5-0.8B | LFM2.5-1.2B Instruct | LFM2.5-1.2B Thinking |
+|---|---|---|---|
+| intent-only accuracy | **61.8%** | 55.3% | 38.2% |
+| full accuracy | **46.1%** | 42.1% | 30.3% |
+| RU / EN | **27/61** / 8/15 | 23/61 / **9/15** | 15/61 / 8/15 |
+| route errors | 0 | 0 | 0 |
+| p50 / p95 latency | **1.28s / 1.94s** | 2.18s / 2.72s | 2.27s / 3.19s |
+
+Full logs: the three runs are archived in the session scratchpad
+(`qwen08.txt`, `lfm-instruct.txt`, `lfm-thinking.txt`).
+
+## Russian-specific failures — the owner's worry is confirmed
+
+LFM2.5's Russian loss is not spread out. It has one large, specific failure: **it hears
+almost any Russian imperative or short phrase as `reminder`.**
+
+- `перезапусти докер` → reminder (want act)
+- `включи вытяжку` → reminder (want act)
+- `закрой жалюзи` → reminder (want act)
+- `заметка: продлить домен в августе` → reminder (want note)
+- `запиши что кран на кухне снова капает` → reminder (want note)
+- `доброе утро` → reminder (want chat)
+- `спасибо тебе` → reminder (want note/chat)
+- `переходи в тихий режим` → reminder (want system)
+
+That is `note→reminder ×4`, `act→reminder ×4`, `chat→reminder ×2` in one run. Qwen's
+equivalent failure axis is `query→fact ×8`, which is a narrower and already-understood bug.
+
+Two more Russian-side problems worth naming:
+
+1. **Fact keys come back empty or wrong in Russian.** `воды попил наконец`, `поужинал`,
+   `поспал часов пять` and `отметь что я позавтракал овсянкой` all returned an empty key.
+   `сходил в душ` and `отдохнул минут двадцать` both returned `water`. Qwen does not do this.
+2. **It leaked German.** `slept about seven hours` produced the fact key
+   `"7 Stunden geschlafen"`. Grammar-valid, semantically garbage — a sign the multilingual
+   mix is not anchored where Maven needs it.
+
+The claimed tool-calling advantage did not show up here. `act` is the closest thing this
+fixture has to a tool call, and LFM2.5 got it wrong more often than Qwen, mostly by calling
+it a reminder. It also produced no `fn` slot on any act, same as Qwen.
+
+## The Thinking variant
+
+Not viable. 36.8% intent accuracy, 10/61 Russian, and no latency saving over Instruct — the
+thinking trace costs time without buying accuracy on a short enum classification. With the
+`enable_thinking=false` diagnostic it collapsed further to 28.9% with 2 route errors
+(`query→reminder ×12`). Do not pursue.
+
+## Notes
+
+- Nothing crashed, nothing ignored the GBNF grammar, and no model produced unparseable JSON
+  in the shippable configurations. Zero route errors for both Instruct and Thinking in
+  `llm-only` and `cascade+llm`. The problem with LFM2.5 is what it decides, not whether it
+  can emit the contract.
+- The `6 / 6` missed clarify is unchanged across all three models. No model fixes the missing
+  refusal lane — that is `Confidence: 1.0` hardcoded in `llmrouter.go` (Vikunja #359), not a
+  model property.
+- The report labels every configuration `(0.8B)`; that string is hardcoded in the test, not a
+  reflection of which gguf was loaded. Model identity was confirmed per run via `/v1/models`.
+- No Go code was changed for this measurement, and no bug was found that needed one.
+
+## What this does not settle
+
+Routing only. LFM2.5 might still phrase better, and phrasing is the resident model's other
+job — that needs its own fixture. But routing is the load-bearing path and Maven is
+Russian-first, so on the evidence here the switch is not worth making.
+
+---
+
+# Second sweep, same evening — five models, and a resident-model change
+
+The sections above compared LFM2.5-1.2B against Qwen3.5-0.8B on routing and concluded
+"the switch is not worth making". That still holds. This sweep asked a different
+question — whether a *smaller* model could work, since LFM2.5's published
+instruction-following scores beat Qwen3.5-0.8B badly — and answered it, plus found a
+better resident model by accident.
+
+**Outcome: the resident model is now stock Qwen3-1.7B.** Sub-500M is a dead end.
+
+## Routing — 77 Russian cases, one run each
+
+| model | on disk | llm-only (full) | llm-only (intent) | cascade + fallback |
+|---|---|---|---|---|
+| LFM2.5-230M-Q8_0 | 246 MB | 23.4% | 33.8% | 36.4% |
+| LFM2.5-350M-Q8_0 | 379 MB | 2.6% | **5.2%** | 20.8% |
+| Qwen3.5-0.8B-Q4_K_M | 527 MB | 36.4% | 59.7% | 61.0% |
+| Qwen3.5-2B-UD-Q4_K_XL | 1.34 GB | 42.9% | 62.3% | 63.6% |
+| **Qwen3-1.7B-UD-Q4_K_XL (stock)** | 1.13 GB | **44.2%** | **67.5%** | **72.7%** |
+
+Qwen3-1.7B wins every column, including against a model 20% larger than it.
+
+## Talk fixture — 27 cases, three runs each, idle box
+
+| | Qwen3.5-0.8B | Qwen3-1.7B stock |
+|---|---|---|
+| composite | 13, 11, 8 | **20, 21, 18** |
+| address | 21, 18, 18 | **26, 25, 23** |
+| feminine | 27, 25, 26 | 26, 27, 26 |
+| lang | 27, 27, 26 | 26, 27, 27 |
+| ontopic | 16, 19, 19 | **22, 23, 23** |
+| canned fallbacks | 8, 5, 6 | **0, 2, 0** |
+
+This also fills the row `TALK-EVAL-31-07-2026.md` had to void for contamination:
+**600ch/1024tok on Qwen3.5-0.8B scores 13, 11, 8.**
+
+`address` is the headline. It sat at 18-22 of 27 on the 0.8B no matter how the prompt
+was worded — the prompt explicitly forbids "вы" and the model writes `вашей`,
+`подождите`, `делаете` anyway. That was read as "prompting is out of levers", and it
+was really "0.8B is out of capacity". The 1.7B mostly holds the constraint.
+
+The fallback column matters too: 5-8 of 27 turns on the 0.8B end in a hardcoded
+`"не знаю."`, meaning it failed to emit parseable JSON about a quarter of the time.
+The 1.7B does that 0-2 times.
+
+## Latency — the long tail is not the Thinking block
+
+| | p50 | p95 |
+|---|---|---|
+| Qwen3.5-0.8B | 2.4s, 2.9s, 2.0s | 17.4s, 17.6s, 17.4s |
+| Qwen3-1.7B stock | 2.7s, 2.6s, 2.8s | 16.4s, 6.6s, 3.9s |
+
+p50 is flat across a 2× size difference. The first instinct on seeing the 1.7B's
+16s p95 was "that is the reasoning trace, cap it" — wrong. The 0.8B's p95 is a
+consistent 17s and the 1.7B beat it in two of three runs. The tail is shared and
+lives somewhere else. Do not spend time on `/no_think` on this evidence.
+
+## Sub-500M: not close, and the benchmarks say otherwise for a reason
+
+LFM2.5-350M publishes IFEval 76.96 against Qwen3.5-0.8B's 59.94, and BFCLv3 44.11
+against 35.08 — better at instruction-following and structured output, at 2/3 the
+size. Those numbers are real and they are **English**. Every benchmark in that
+table except Multi-IF is English-only.
+
+In Russian, with a 300-token budget and temperature 0:
+
+- **350M**, «Столица Франции? Ответь кратко.» → *«Сторзит в Париже.»* — `Сторзит` is
+  not a word; it is invented morphology.
+- **350M**, asked to read back a reminder → a fortune cookie about being attentive
+  and confident. No reminder in it.
+- **230M**, «Привет, как дела?» → answered **in Spanish**.
+
+The 230M beating the 350M six-fold on routing (33.8% vs 5.2%) is the other tell:
+when the larger sibling collapses like that it is format compliance failing, not
+reasoning.
+
+This is a pretraining gap, not a fine-tuning gap. Teaching Russian to a 350M from
+near-zero is not an afternoon on a Colab, which was the premise worth checking.
+
+## Why this vindicates the 1.7B CPT
+
+Stock Qwen3-1.7B, untrained and unprompted, answers all three probes in fluent
+correct Russian. What it gets wrong is the persona: *«Привет! Я рад, что ты здесь»*
+— `рад` is masculine and Maven needs `рада`. That is the right kind of remaining
+problem, and it is exactly what the CPT (Vikunja #122) is for.
+
+The 1.7B was the correct model choice. What was wrong was treating it as a
+**blocker**: stock already beats what was deployed, so it ships now and gets
+swapped again when the CPT lands.
+
+## Caveats
+
+- Routing is one run per model, not three. The gaps between families are far larger
+  than the run-to-run spread seen on the talk fixture, but the 2B-vs-1.7B gap (62.3
+  vs 67.5) is not safe to call on one run.
+- ~~The routing numbers only reach production once the LLM router is wired on. It is
+  still `nil`.~~ **Resolved the same evening:** the LLM router is wired at `voice.go:214`
+  behind `voice.llm_router`, the default is on, and `deploy/mavend.json` sets it `true`.
+  These numbers are the production path now, so the p50 ≈2.7s is a real per-turn cost and
+  not a bench artifact.
+- ~~`/mnt/hdd1/llms/LFM2.5/Qwen3-1.7B-UD-Q4_K_XL.gguf` is a 293 MB truncated download
+  in the wrong directory.~~ **Deleted 2026-07-31.** The good 1.13 GB copy in `qwen3/` is
+  what `deploy/mavend.json` loads.
+- Harness: `scratchpad/bakeoff.sh`, one server at a time, health-checked before each
+  run, `/v1/models` recorded per run. Never run two LLM consumers at once — see the
+  contamination note in `TALK-EVAL-31-07-2026.md`.

Makefile

@@ -20,7 +20,7 @@ PIPER_ESPEAK  := $(shell pwd)/deps/piper/espeak-ng-data
 
 all: build
 
-build: build-stt build-tts build-daemon build-client build-waked build-web build-poll build-caldav
+build: build-stt build-tts build-daemon build-client build-waked build-web build-poll build-caldav build-mail
 
 build-stt:
 	CGO_CFLAGS="$(CGO_CFLAGS)" CGO_LDFLAGS="$(CGO_LDFLAGS)" LD_LIBRARY_PATH="$(shell pwd)/deps/lib" \
@@ -50,6 +50,9 @@ build-poll:
 build-caldav:
 	$(GO) build $(GOFLAGS) -o mavcaldav ./cmd/mavcaldav/
 
+build-mail:
+	$(GO) build $(GOFLAGS) -o mavmaild ./cmd/mavmaild/
+
 run-web: build-web
 	./mavweb -addr :9200 -voice 127.0.0.1:9100
 
@@ -103,14 +106,17 @@ eval-router:
 eval-recall:
 	MAVEN_ONNX_LIB="$(MAVEN_ONNX_LIB)" $(GO) test -v -count=1 ./internal/memory/recalleval/
 
-# eval-phrasing -- score nudge phrasing (internal/phraser/eval). Verbose so the
+# eval-phrasing -- score nudge phrasing AND the conversational paths (chat,
+# query, general knowledge) in internal/phraser/eval. Verbose so the
 # report and every generated message land in the terminal. With no environment
 # it scores the deterministic Stub only, which is what CI runs. Set
 # MAVEN_LLM_URL to add the resident model:
 #   MAVEN_LLM_URL=http://127.0.0.1:18099 make eval-phrasing
-# The model run is slow (minutes) -- the timeout is raised to match.
+# The model run is slow (minutes) -- the timeout is raised to match. It covers
+# two fixtures now (15 nudges + 27 conversational cases, and the chat replies are
+# the long ones), hence 90m rather than 40m.
 eval-phrasing:
-	$(GO) test -v -count=1 -timeout 40m ./internal/phraser/eval/
+	$(GO) test -v -count=1 -timeout 90m ./internal/phraser/eval/
 
 # eval-models — score ONE llama-server against the same fixture, for the
 # resident-model bake-off (#278, #250). Start a server with the gguf you want,
@@ -152,9 +158,13 @@ deps-piper:
 		-o /tmp/piper.tar.gz
 	tar -xzf /tmp/piper.tar.gz -C deps/
 
-EMBEDDER_DIR := $(shell pwd)/models/embedder
+# multilingual-e5-small: an asymmetric retrieval model. It is trained to match
-EMBEDDER_MODEL_URL := https://huggingface.co/Xenova/paraphrase-multilingual-MiniLM-L12-v2/resolve/main/onnx/model_quantized.onnx
+# a short question against a longer passage, which is what note recall is.
-EMBEDDER_TOKENIZER_URL := https://huggingface.co/Xenova/paraphrase-multilingual-MiniLM-L12-v2/resolve/main/tokenizer.json
+# The quantized file is the one we download, deploy and measure — see
+# RECALL-EVAL-31-07-2026.md.
+EMBEDDER_DIR := $(shell pwd)/models/embedder/multilingual-e5-small
+EMBEDDER_MODEL_URL := https://huggingface.co/Xenova/multilingual-e5-small/resolve/main/onnx/model_quantized.onnx
+EMBEDDER_TOKENIZER_URL := https://huggingface.co/Xenova/multilingual-e5-small/resolve/main/tokenizer.json
 
 download-embedder:
 	mkdir -p $(EMBEDDER_DIR)
@@ -178,4 +188,4 @@ download-embedder:
 	@echo '  sudo cp onnxruntime-linux-x64-1.15.1/lib/libonnxruntime.so* /usr/local/lib/'
 
 clean:
-	rm -f mavend mavenclient mavsttd mavttsd mavweb mavpoll mavcaldav mavwaked
+	rm -f mavend mavenclient mavsttd mavttsd mavweb mavpoll mavcaldav mavwaked mavmaild

PHRASING-EVAL-31-07-2026.md

@@ -0,0 +1,169 @@
+# Phrasing evaluation — 31-07-2026
+
+How Maven words a nudge, measured instead of argued. Counterpart to
+`ROUTING-EVAL-31-07-2026.md`.
+
+- Fixture + scorer: `internal/phraser/eval/` (`nudges_v1.json`, 15 cases; `eval.go`, `checks.go`)
+- Reproduce: `MAVEN_LLM_URL=http://127.0.0.1:18099 make eval-phrasing`
+- Model: Qwen3.5-0.8B Q4_K_M, the resident model. Not swapped.
+- Commit: `a40bc55` (prompt fix)
+
+Every check is a string or length test a human can read and disagree with. No model
+grades another model here.
+
+## Result
+
+| | before | after |
+|---|---|---|
+| **cases passing every check** | **0/15** | **13/15** |
+| mood in enum | 6/15 | 15/15 |
+| Russian | 2/15 | 14/15 |
+| length (≤120 chars, ≤16 words) | 13/15 | 15/15 |
+| feminine self-reference | 15/15 | 15/15 |
+| no cringe | 13/15 | 15/15 |
+| on topic | 6/15 | 13/15 |
+| p50 latency | 11.4s | 11.4s |
+
+Latency did not move and is not good. 11s to word one nudge on this box.
+
+## The bug reproduced
+
+Yes, exactly as reported. 7 of 15 messages were the literal string `"..."`, and one was
+`"full voice message"`. Both are text copied straight out of the prompt.
+
+The system prompt said:
+
+```
+Respond ONLY with valid JSON: {"response": "full voice message", "mood": "neutral"}
+```
+
+and the user prompt said:
+
+```
+Respond as JSON: {"response": "...", "mood": "..."}
+```
+
+A 0.8B does not read `"..."` as "put your answer here". It reads it as the answer. The
+prompt was a worked example whose worked part was blank, so the model filled the slot by
+copying. This is the whole of finding 1.
+
+## What else was wrong
+
+Four separate faults, all prompt-side:
+
+1. **Placeholder echo** (7 cases) — above.
+2. **Wrong language** (13/15 failed the language check). The prompt was entirely English
+   and said "in the user's language (Russian or English)". The model picked English. It is
+   never English: the nudge is spoken by a Russian piper voice.
+3. **Rule names are English identifiers.** `netdata_critical`, `service_down`, `break` went
+   into the prompt raw. The model cannot nudge about a topic it has not been told in words,
+   so 9/15 were off topic. The daemon knows what its own rules mean; now it says so.
+4. **Mood invented** (`"warm"`, twice). The enum was listed in a parenthesis at the end of
+   an English sentence. Now it is its own line: "ровно одно из: neutral, happy, thinking,
+   tired, confused."
+
+Plus two non-prompt faults the run exposed:
+
+- **The no-parse fallback was English.** When the model returned nothing usable, the body
+  became `fmt.Sprintf("%s — %s", rule, sev)` — `"water — care"` — and that string went to
+  a Russian TTS. Now it falls back to plain Russian.
+- **Durations were English.** `humanDur` returns "3 hours"; it was landing verbatim inside
+  Russian sentences. Nudges now use a Russian formatter.
+
+## Three iterations, and what each taught
+
+| | score | change |
+|---|---|---|
+| baseline | 0/15 | — |
+| iter 1 | 2/15 | Russian prompt, filled-in examples, Russian durations |
+| iter 2 | 11/15 | required keyword per rule, one example instead of five, Russian fallback |
+| iter 3 | **13/15** | examples moved to topics that are not rules |
+
+The interesting step is 1 → 2. Fixing the placeholder did not fix the disease, it moved it:
+the model stopped copying `"..."` and started copying my first example instead. Five nudges
+in a row came back as `"Ты не пил воду три часа. Налей стакан."` regardless of the rule.
+
+**A small model copies the nearest concrete text in its prompt.** That is one failure mode
+with two symptoms. The fix that stuck was making the examples about laundry and a laptop
+battery — topics no rule ever produces, so copying them is visible in the score rather than
+invisibly passing the water cases.
+
+## Do not oversell 13/15
+
+Seven of the thirteen passes are the **deterministic fallback**, not the model:
+`"Напоминаю: таблетки."`, `"Сервис не отвечает."`, `"Критический алярм: проверь диск."`,
+`"Ты давно не пил воду."`. Those are strings this commit added to Go. The model returned
+nothing parseable and the fallback scored.
+
+So the honest reading is roughly **6/15 from the model, 7/15 from a fallback, 2/15 failing**.
+The prompt fix is real — `"..."` is nearly gone and the language and mood checks are clean —
+but a large part of the jump is that failure now degrades into Russian instead of into
+`"water — care"`. That is a genuine improvement for the operator and a weak one for the model.
+
+The two remaining failures: one `"..."` recurrence (`routine-stretch`) and one meal nudge
+that never says food.
+
+## Tried and reverted: an example-led nudge prompt (#393)
+
+The idea was that a 0.8B copies examples better than it follows rules, so the nudge prompt
+was rewritten to lead with five on-topic examples (water, break, pills, morning, service) and
+the prose rules were compressed to pay for the tokens: 1190 chars down to 986.
+
+It measured **worse**, three runs each side, same llama-server, same fixture:
+
+| run | before | after |
+|---|---|---|
+| 1 | 12/15 (address 14) | 11/15 (address 13) |
+| 2 | 13/15 (address 15) | 12/15 (address 15) |
+| 3 | 14/15 (address 15) | 11/15 (address 12) |
+
+`feminine` and `hisgender` were 15/15 on all six runs, so they measure nothing here. The
+regression is all in `address`: 44/45 before, 40/45 after. Formal "вы"/"ваше" and plural
+imperatives came back, and so did `"..."`.
+
+Two likely causes, both about the same thing — **examples do not carry a prohibition**. The
+old prompt spent a whole sentence on «говоришь на "ты", в единственном числе»; the new one
+demoted that to one item in a long "никогда" list, and the model stopped obeying it. And
+making the examples on-topic let their *wording* leak: a break case came back as
+«Вы давно не пили воду. Выпей стакан.» — the water example, verbatim, in the wrong slot.
+That is exactly the failure the laundry/laptop examples were chosen to avoid.
+
+Change reverted. What survives is the measurement: a rule the model must obey needs its own
+sentence, and examples must stay off-topic. Also note the before side alone spans 12–14 of
+15 — this fixture cannot resolve anything smaller than about three cases.
+
+## Broken, found, not fixed
+
+1. ~~**`checkFeminine` only catches half the constraint.**~~ **Fixed** (#381). It scanned for
+   masculine self-reference only, so three messages that addressed the *owner* in the feminine
+   ("ты давно не отдыхал**а**") scored clean. There is now a second check, `hisgender`: a
+   feminine past-tense verb (-ла/-лась) in a sentence addressed to him ("ты", "тебе", "твой")
+   fails, unless the verb is hers ("я заметила", "напомнила тебе"). It is a suffix rule, not a
+   parser — see the comment in `checks.go` for what it misses. A fresh 15-case run after adding
+   it scored **12/15** with `hisgender` 15/15; the model did not repeat the feminine address in
+   that sample, and the check is pinned by unit tests on the recorded bad strings instead.
+2. **Grammar is not checked at all, and it is bad.** `"Он не ел 11 дней"` (it was 11 hours),
+   `"Сонуждились 7 дней"` (not a word), `"Они забыли воду"` (wrong person entirely). Every
+   one of these passes all six checks. The fixture measures properties, not fluency, and at
+   0.8B fluency is the binding constraint.
+3. **Unit confusion.** The model turns hours into days about a third of the time. The
+   prompt now says "11 ч"; it reads it as days.
+4. **11s p50.** Unchanged and untouched here. A nudge the model takes eleven seconds to
+   word has missed its moment. Worth its own task.
+5. **The keyword hint is close to teaching to the test.** `ruleKeywords` names the word the
+   on-topic check looks for. It is defensible — the daemon genuinely knows its rule topics
+   and the model genuinely cannot infer them from `netdata_critical` — but the on-topic
+   number is softer than the others because of it.
+
+## Next steps
+
+1. ~~**Add a second-person gender check**~~ — done, `hisgender` in `checks.go` (#381).
+2. **Decide whether the fallback should count as a pass.** Right now `Score` cannot tell a
+   model answer from a fallback. Either mark fallback bodies in `PhrasedNudge` or count them
+   in their own column. Without that, any future prompt change can score well by failing
+   more.
+3. **Attack the 11s.** Nudge phrasing is short and non-interactive; thinking off is the first
+   thing to try, as it was for routing (#376).
+4. **Re-measure when #122 lands.** The CPT'd Qwen3-1.7B is the target. 13/15 with seven
+   fallbacks is the floor it has to beat, and the fluency problems above are the ones a
+   bigger, Russian-trained checkpoint should actually fix.

REARCH.md

@@ -90,4 +90,7 @@ later* is the worker + RAG.
 4. **Deferred work** — larger reasoner, custom Piper voice and other expansions.
 
 ## Non-goals (unchanged)
-Never phones home. Not a nag. Not autonomous. Feminine-gendered RU self-ref.
+Not a nag. Not autonomous. Feminine-gendered RU self-ref. No telemetry, no
+cloud model, no third-party account — but she MAY read external sources to
+answer world questions (Kiwix first, search optional). "Never phones home" as
+an absolute is deprecated, owner's call 2026-07-31; see CLAUDE.md § Non-goals.

RECALL-EVAL-31-07-2026.md

@@ -75,6 +75,14 @@ same vector. A note is indexed in both places with the same embedding, so if it
 `QueryNotes` it fails again here — the branch can only ever return a **fact**. Its comment calls it
 "additive"; for notes it is not.
 
+**Fixed (Vikunja #373).** The memory pass now runs *first*, as one search over notes and facts with
+one gate, so whichever memory is clearly the best match answers — note or fact. The notes-only pass
+stays behind it for notes the vector index does not hold. No threshold changed, so the set of
+questions Maven answers is the same; only which memory answers them. The fixture gained two mixed
+note+fact cases (`ru-mixed-031`, `ru-mixed-032`), which is why the counts below are out of 27
+answerable cases and not 25: hash recall@1 36.0% (9/25) → 37.0% (10/27), e5 recall@1 72.0% (18/25) →
+70.4% (19/27) with answered-after-gate 68.0% → 66.7% and false recall unchanged at 1/5.
+
 ### 5. Ranking has no recency or type signal, and the store is not the bottleneck
 
 `internal/store/notes.go:67` sorts by cosine and uses `ts` only to break an exact float tie, which
@@ -83,14 +91,157 @@ sqlite-backed `store.MemoryStore` and `memory.InMemoryStore` identically — bot
 (`internal/store/memory.go:64`) at ~150µs over 42 rows against a ~59ms query embed. An ANN index is
 not the problem to solve.
 
+## Re-measured after the embedder swap — 31-07-2026, later the same day
+
+Changed: `models/embedder/` is now **multilingual-e5-small** (quantized, 118MB), with `query: ` in
+front of a question and `passage: ` in front of a stored note (Vikunja #371). `deploy/mavend.json`
+and `make download-embedder` now name the same file, and it is the quantized one — that is what the
+column below measures (Vikunja #372). Everything else is unchanged: same fixture, same store, same
+0.55 gate. The old column is the baseline and is left as it was.
+
+| | recall+onnx, MiniLM (baseline) | recall+onnx, e5-small (new) |
+|---|---|---|
+| **recall@1** | 60.0% (15/25) | **72.0% (18/25)** |
+| recall@3 | 80.0% (20/25) | 84.0% (21/25) |
+| **answered after the 0.55 gate** | 48.0% (12/25) | **72.0% (18/25)** |
+| wrong note on top / tie on top | 10 / 0 | 7 / 0 |
+| ranked first, then silenced by the gate | 3 | 0 |
+| **false recall** | 1/5 (20%) | **5/5 (100%)** |
+| top-1 score when right, min / median | 0.559 / 0.678 | 0.791 / 0.857 |
+| top-1 when it must stay silent, median / max | 0.470 / 0.567 | 0.815 / 0.835 |
+| RU / EN / `hard` cases passed | 13/24 / 3/6 / 2/11 | 14/24 / 4/6 / 5/11 |
+| latency p50 / p95 / max | 59ms / 148ms / 194ms | 18ms / 37ms / 49ms |
+
+### What moved
+
+Ranking got better and got faster. Half the previously-unwinnable `hard` cases now pass (2/11 →
+5/11), the guitar note no longer beats the docker-logs note, and the gate stops silencing notes that
+already ranked first. The quantized e5 is also ~3x quicker than the fp32 MiniLM it replaces.
+
+### What got worse: the gate is now a no-op
+
+e5 packs every cosine into a narrow high band. Right-note scores start at 0.791; must-stay-silent
+scores reach 0.835. **The distributions still overlap, and now they overlap above the gate**, so
+0.55 admits everything and false recall goes from 1/5 to 5/5. The sweep:
+
+```
+gate 0.50–0.70: answered 18/25 (72%)  false recall 5/5
+gate 0.80:      answered 17/25 (68%)  false recall 4/5
+gate 0.90:      answered  0/25 ( 0%)  false recall 0/5
+```
+
+There is no value that keeps real recall and rejects made-up questions — same conclusion as before,
+now with a wider band and no room at all. `query_min_score` was left at 0.55 as instructed. **The
+recommendation is to leave it there and stop tuning it**: any number under ~0.79 is a no-op and
+anything above starts cutting real recall long before it stops the false ones. The fix is a margin
+gate (`top1 − top2 > δ`), next-steps item 3, which is now the top item.
+
+### The prefixes did not do the work
+
+A control run with both prefixes set to the empty string scored the **same** recall@1 (72%), a
+slightly better recall@3 (88%) and the same 5/5 false recall. So on this fixture the gain comes from
+the model, not from the `query:` / `passage:` split. The prefixes are kept because they are how e5
+was trained and the split is the right shape for the read path, but they are not worth defending on
+this evidence — a bigger fixture may say otherwise.
+
+### Stored vectors from the old model are now junk
+
+Cosine between a MiniLM vector and an e5 vector means nothing. Every row already in `notes` and in
+the vector memory table was written by the old model, so after this deploy they will score as noise
+against a new query. A live database needs every note and fact re-embedded before recall works at
+all. Filed as its own task.
+
+## Margin gate — 31-07-2026, third run
+
+Next-steps item 3, done. The absolute gate is replaced by a **margin gate**: answer only when the
+top hit beats the runner-up by more than delta (`top1 − top2 > δ`). Same fixture, same e5 embedder,
+same store as the run above. `internal/memory/gate.go` holds the check; both read paths call it
+(`cmd/mavend/recall.go` and the notes-RAG branch in `voice.go`). New knob `voice.query_min_margin`
+in `deploy/mavend.json`, default 0.008.
+
+### Why the absolute gate could not work, in one line of data
+
+The harness now prints the margin distributions, and they barely overlap where the raw scores
+overlap completely:
+
+| | top-1 score | margin (top1 − top2) |
+|---|---|---|
+| right note first (n=18) | min 0.810, median 0.862, max 0.890 | min 0.001, median 0.029, max 0.053 |
+| must stay silent (n=5) | min 0.795, median 0.815, max 0.835 | min 0.000, median 0.002, **max 0.019** |
+
+Four of the five must-be-silent cases have a margin at or under 0.002 — when there is nothing to
+recall, e5 finds several notes equally close and no clear winner. That is the signal the absolute
+score throws away.
+
+### The delta sweep
+
+Absolute gate held at 0.55 throughout.
+
+```
+delta 0.000: answered 18/25 (72%)  false recall 5/5
+delta 0.002: answered 17/25 (68%)  false recall 3/5
+delta 0.005: answered 17/25 (68%)  false recall 2/5
+delta 0.008: answered 17/25 (68%)  false recall 1/5   <- chosen
+delta 0.010: answered 15/25 (60%)  false recall 1/5
+delta 0.012: answered 14/25 (56%)  false recall 1/5
+delta 0.015: answered 12/25 (48%)  false recall 1/5
+delta 0.020: answered 11/25 (44%)  false recall 0/5
+delta 0.025: answered  9/25 (36%)  false recall 0/5
+delta 0.030: answered  8/25 (32%)  false recall 0/5
+delta 0.040: answered  4/25 (16%)  false recall 0/5
+delta 0.050: answered  2/25 ( 8%)  false recall 0/5
+delta 0.060: answered  0/25 ( 0%)  false recall 0/5
+```
+
+### Chosen: δ = 0.008
+
+It is the best point on the frontier, not a taste call. **0.008 dominates 0.010, 0.012 and 0.015
+outright** — same 1/5 false recall, 8 to 20 points more real recall. Everything below it buys recall
+back only by admitting more false recalls (0.005 → 2/5, 0.002 → 3/5). The next real improvement is
+0.020 at 0/5 false, and it costs 24 points of recall to get there.
+
+The brief's bar was "recall above 60% with false recall at 1/5 or better". 0.008 clears it with room:
+68% and 1/5.
+
+### Before / after
+
+| | absolute gate 0.55 (previous) | margin gate δ=0.008 |
+|---|---|---|
+| recall@1 (ranking, ungated) | 72.0% (18/25) | 72.0% (18/25) — unchanged, the gate does not rank |
+| **answered after the gate** | 72.0% (18/25) | **68.0% (17/25)** |
+| **false recall** | **5/5 (100%)** | **1/5 (20%)** |
+| fixture cases passed | 18/30 | **21/30** |
+
+Four false recalls removed for one real answer. That is the trade the spec asks for — she is not a
+guesser-of-truth. The one survivor is `en-pref-025` ("should i be offered wine"), which recalls a
+filler note at 0.796 with a 0.019 margin: the widest silent-case margin in the fixture, and it sits
+inside the real-recall range, so no delta removes it without taking real answers with it.
+
+### Does the absolute cutoff still earn its keep? Marginally — kept
+
+On this fixture with e5 it is a **no-op**: the lowest right-note score is 0.791, so 0.55 rejects
+nothing the margin does not already reject. It is kept for two reasons, neither glamorous. It still
+does real work for the hash embedder (its own sweep shows answers dropping from 16% to 0% between
+0.30 and 0.50), and it is the only thing standing between the user and a reply built from a store
+where everything is far away but one row happens to be a little less far — a near-empty database, or
+the stale-vector case below. Cheap insurance, no measured cost. If a later embedder makes it bite,
+the sweep is one command.
+
+### Caveat on the numbers
+
+Five must-be-silent cases is a thin basis for a 4-point decision. 1/5 and 2/5 differ by one case.
+The shape of the frontier is trustworthy — margins separate, absolute scores do not — but δ=0.008
+itself should be re-read off a bigger fixture (next-steps item 6) before anyone defends the third
+decimal.
+
 ## Next steps — ordered by value-to-risk; nothing here is a decision
 
 1. **Swap the embedder to `multilingual-e5-small` with `query:`/`passage:` prefixes.** One config
    change plus a prefix in `onnxembedder.go`, re-measurable in one command.
 2. **Re-run `make eval-recall`, then set the gate from the sweep** — not before. Any
    `query_min_score` picked against today's embedder describes a model on its way out.
-3. **Replace the absolute-score gate with a margin gate** (`top1 − top2 > δ`) — as the routing eval
+3. ~~**Replace the absolute-score gate with a margin gate**~~ — done, see the section above.
-   concluded, absolute cosine cannot see a flat distribution.
+   δ=0.008, false recall 5/5 → 1/5.
 4. **Delete or repair the dead `memStore` branch** at `voice.go:776` — search before the gate,
    gate it separately, or restrict it to facts and say so.
 5. **Add a mild time decay to ranking** — the newest statement of a preference is the true one.

ROUTING-EVAL-31-07-2026.md

@@ -66,15 +66,113 @@ Three things this run settles:
    `запиши что…` phrasings toward fact, and that suspicion stands — all five `ru-note-*`
    cases now land on fact. Tracked as Vikunja #375.
 
-**Thinking off is the best configuration measured so far**, on both accuracy and latency
+The `thinking off` column above read as the best configuration measured so far (Vikunja #376).
-(Vikunja #376). That is worth understanding before flipping: routing is a short
+**It was wrong** — see the controlled re-run below. Ignore that column.
-classification into a fixed enum with grammar-constrained output, so there is little to
-reason about, and the thinking trace mostly gives a small model room to talk itself out of
-the right answer. Phrasing is a different job and needs measuring separately.
 
 Still `6 / 6` missed clarify — the router has no way to say "I don't know" (Vikunja #359).
 That is unchanged by anything here.
 
+## Thinking off — 31-07-2026, controlled re-run (Vikunja #376)
+
+The "thinking off wins by 6 points" observation above **does not hold**. It was a measurement
+artefact, and the earlier table's `thinking off` column should be ignored.
+
+The thinking-off variant was scored by a hand-rolled HTTP client living in the test file
+instead of `llm.Client`. That copy did not send `repeat_penalty`, which the real router does
+send (`routeRepeatPenalty = 1.15`). So the two columns differed on two axes at once, and the
+one that mattered was the penalty, not the thinking mode.
+
+Re-measured with everything else held equal — same fixture, same prompt, same grammar, same
+sampling, same idle box, the three configurations run back to back and never concurrently:
+
+| | llm-only, thinking on | llm-only, thinking off | cascade+llm |
+|---|---|---|---|
+| intent-only accuracy | 59.2% (45/76) | 59.2% (45/76) | 61.8% (47/76) |
+| full accuracy (intent+slots+gate) | 38.2% (29/76) | 38.2% (29/76) | 57.9% (44/76) |
+| route errors | 3 | 3 | 0 |
+| grammar violations | 3 (all 3 route errors) | 3 (same 3 cases) | 0 |
+| missed clarify | 5 / 6 | 5 / 6 | 5 / 6 |
+| p50 latency | 836ms | 920ms | 810ms |
+| p95 latency | 1.41s | 2.00s | 1.31s |
+
+Thinking off is not just a tie on the headline numbers — it is identical case for case, with
+the same confusion matrix and the same three unparseable replies. The latency difference is
+run-to-run noise on one box, and it points the wrong way here.
+
+The reason is simpler than any accuracy argument: **this llama-server build ignores the
+request-level thinking switch for this model.** Probed directly against the running server
+with `chat_template_kwargs.enable_thinking = false`, `chat_template_kwargs.thinking = false`
+and top-level `reasoning_budget = 0` — all three return a byte-identical answer with the
+thinking trace still in `reasoning_content`, and the server reports the prompt prefix as
+cached, meaning the rendered template did not change. There was never anything being turned
+off, which is also why the numbers match exactly.
+
+Nothing was defaulted. `internal/llm` still has no `chat_template_kwargs` field, `VoiceConfig`
+has no thinking flag, and `deploy/mavend.json` is unchanged. The misleading third
+configuration is removed from `internal/router/eval` so the table it produced cannot be quoted
+again.
+
+Two caveats worth saying out loud:
+
+- **The fixture is 76 cases.** A 6-point difference on 76 cases is roughly 4-5 cases and would
+  not have been worth trusting even if it had reproduced. This one was exactly 0 cases, which
+  is a much easier call.
+- **This is one server build and one checkpoint** (`b9351`, Qwen3.5-0.8B Q4_K_M). If the
+  #122 checkpoint or a newer llama.cpp does honour the switch, the question reopens — but it
+  reopens as an unmeasured question, not as a 6-point win.
+
+Phrasing was **not** measured. Whether thinking helps there is still open, and now also blocked
+on the same "can we even turn it off" question.
+
+## Clock and calendar rule — 31-07-2026 (Vikunja #374)
+
+`routeSystem` never said whether "который час" or "какое число завтра" are `system` or
+`query`, and `system→query ×4` showed up in every run. The rule added says: the clock and the
+calendar date themselves are `system`; what is *written in* the calendar or in memory
+("что у меня завтра", "какие есть напоминания") stays `query`; and a time named inside a
+request ("напомни завтра…") is just a detail of the request, not a reason for `system`.
+
+That split is not a preference. In `cmd/mavend/voice.go` only `replySystem` owns the clock and
+the date formatter, so a clock question routed to `query` falls into the embedder + note RAG
+and answers "не знаю". The agenda, on the other hand, is answered by `ParseCalendarDate` +
+`CalendarEvents` *inside* the `query` branch, so that side has to stay `query`. The rule sits
+above the question test because every one of these utterances carries a question word and a
+later rule would never be reached.
+
+The fixture is now 77 cases: one calendar-agenda case was added
+(`ru-query-019` "что у меня стоит в календаре на послезавтра", intent `query`) specifically so
+an over-broad system rule cannot pass unnoticed. The clock/date cases (`ru-sys-001/002/005`,
+`en-sys-001`) already existed.
+
+Three runs, same box, back to back, never concurrently:
+
+| | baseline | first rule (too broad) | rule as committed |
+|---|---|---|---|
+| llm-only intent-only | 59.2% (45/76) | 54.5% (42/77) | 59.7% (46/77) |
+| llm-only full | 38.2% | 35.1% | 39.0% |
+| llm-only route errors | 3 | 4 | 5 |
+| llm-only p50 | 1.09s | 0.91s | 0.93s |
+| cascade+llm intent-only | 61.8% (47/76) | 58.4% | 62.3% (48/77) |
+| cascade+llm full | 57.9% | 54.5% | 59.7% |
+| cascade+llm route errors | 0 | 0 | 0 |
+| cascade+llm p50 | 0.91s | 0.80s | 1.04s |
+
+**The targeted bug is fixed and the headline number did not move.** `system→query ×4` is gone
+in both LLM configurations — the `time` and `date` tags go from 0/2 and 0/2 to 2/2 and 2/2 —
+but the model then over-applies the rule, and `query→system ×5` plus `reminder→system ×2`
+appear where they did not exist before. Net accuracy is a wash, inside the noise of a 77-case
+fixture.
+
+The first attempt is shown because it is the honest history: it said "спрашивает время, дату
+или день недели → system" with no scope, which swept up reminders, and it cost 3-5 points. It
+was tightened once, on the reasoning that a rule capturing "напомни завтра в 7" is simply
+wrong, and not tuned further. The remaining `query/reminder → system` over-trigger is a new,
+separate weakness of the sub-1B model and deserves its own task rather than more prompt
+kneading against a held-out fixture.
+
+The rule is kept. It is correct about what the daemon can answer, and the failure it replaces
+was silent ("не знаю" to "который час") while the one it introduces is loud.
+
 ## Findings
 
 ### 1. The resident model does route better — 50.0% vs 36.8%
@@ -145,11 +243,11 @@ Note the grammar's `string ::= "\"" ([^"\\] | "\\" .)* "\""` is unbounded, so no
 
 ### 7. Two hypotheses tested and closed
 
-- **Thinking mode is a non-issue.** Qwen3.5's template defaults `thinking = 1`, so
+- **Thinking mode is a non-issue.** Confirmed twice now, the second time properly — see the
-  grammar-constrained JSON lands in `reasoning_content` with `content` empty —
+  controlled re-run section. Grammar-constrained JSON lands in `reasoning_content` with
-  `llm.Client`'s fallback handles it. A `thinking off` run scored *identically* (18/76,
+  `content` empty and `llm.Client`'s fallback handles it; the request-level switch does
-  48.7%, same p50). `internal/llm` deliberately does **not** grow a `chat_template_kwargs`
+  nothing on this build. `internal/llm` deliberately does **not** grow a
-  field.
+  `chat_template_kwargs` field.
 - **Runaway array repetition does not reproduce.** An isolated smoke test with a stripped
   grammar emitted `{"intent":"reminder"}` until `MaxTokens`; under the real `routeSystem`
   prompt the few-shot examples anchor it to one object. 2 errors in 76, not 76.

TALK-EVAL-31-07-2026.md

@@ -0,0 +1,150 @@
+# Conversational phrasing eval — 31-07-2026
+
+Every score measured tonight, on the three paths the nudge eval never touched:
+chat, query-with-notes, and general knowledge.
+
+**Short version: the plumbing got fixed and the score barely moved.** Grammar and
+Russian prompts together took the composite from ~9 to ~14 of 27. Everything
+still failing is the model not knowing things or not holding a constraint, and
+prompting is out of levers. Settles the measurement half of Vikunja #395 / #398 /
+#400.
+
+## How to reproduce
+
+```sh
+# llama-server: -c 4096 -ngl 99 -t 6, model /mnt/hdd1/llms/qwen3.5/Qwen3.5-0.8B.Q4_K_M.gguf
+MAVEN_LLM_URL=http://127.0.0.1:18099 no_proxy=127.0.0.1,localhost \
+  deps/go/go/bin/go test -count=1 -timeout 40m \
+  -run TestLLMTalkBaseline ./internal/phraser/eval/ -v
+```
+
+Three runs per configuration, always. The fixture is 27 cases, so one reply
+changing moves the composite by 3.7 points — a single run cannot tell a real
+change from sampling noise. This was learned the expensive way: an earlier claim
+that "one nudge case fails every run" turned out to be three different cases
+across three runs.
+
+**Run the box otherwise idle.** See the contamination note at the bottom.
+
+## Composite, per configuration
+
+| config | overall /27 | chat /9 | query /9 | knowledge /9 | canned fallbacks |
+|---|---|---|---|---|---|
+| baseline, no grammar | 7, 12, 7 | 1, 1, 0 | 2, 4, 2 | 4, 7, 5 | 0, 0, 0 |
+| + GBNF grammar (#398) | 14, 15, 8 | 1, 3, 0 | 5, 6, 3 | 8, 6, 5 | 0, 0, 0 |
+| + Russian prompts (#400) | 11, 17, 15 | 1, 5, 3 | 5, 6, 8 | 5, 6, 4 | 0, 0, 0 |
+| + truncation fix, 1000ch/768tok | 12, 13, 10 | 2, 2, 1 | 7, 7, 5 | 3, 4, 4 | 3, 3, 6 |
+| + rebalanced, 600ch/1024tok | **void — contaminated** | | | | |
+
+"Canned fallbacks" counts replies that came back as the hardcoded `"не знаю."`
+or `"поговорили."`. It is not a check, it is a health signal: those strings mean
+the phraser gave up, and the eval scores them as ordinary bad replies.
+
+## Per-check
+
+| check | no grammar | + grammar | + RU prompts | + truncation fix |
+|---|---|---|---|---|
+| nonempty | 27, 27, 27 | 27, 27, 27 | 27, 27, 27 | 27, 27, 27 |
+| ellipsis | 20, 19, 23 | 27, 27, 27 | 27, 27, 27 | 27, 27, 27 |
+| lang | 13, 16, 15 | 23, 26, 26 | 25, 26, 25 | 26, 27, 27 |
+| feminine | — | — | 25, 24, 26 | 25, 25, 27 |
+| address | — | — | 21, 22, 22 | 22, 21, 22 |
+| ontopic | — | — | 17, 24, 18 | 17, 19, 14 |
+
+`nonempty` reading 27/27 everywhere is not good news — it was a broken check.
+It tested for a non-blank string, so replies of literally `{` and `"15-16"`
+passed it. Fixed on `overnight/fix-truncation`; it needs a letter now.
+
+## What each change actually bought
+
+**GBNF grammar (#398) — the biggest single win.** Qwen3.5-0.8B writes
+`Thinking Process:` as plain text with no tags, `stripThink` only handles
+`</think>`, so the JSON never closed and the plain-text fallback shipped the
+literal reasoning. `ellipsis` went 20→27 and `lang` 13→26. The router had been
+using a grammar for ages; the phraser asking nicely in the prompt was the
+oversight.
+
+**Russian prompts (#400) — modest, plus a large latency win.** Chat 1.3→3.0
+average, query 4.7→6.3, knowledge 6.3→5.0. All inside the run-to-run spread, so
+"probably better on the paths it targeted, not provable in three runs". p50
+latency dropped from ~11.5s to ~2.3s and that part is consistent across all
+three runs — shorter prompts, and she stopped emitting English reasoning first.
+
+**Truncation fix — necessary, and did not help the score.** Two real bugs
+(replies of `{`, and a `nonempty` check that passed them), both fixed, and the
+composite went nowhere. A complete rambling wrong answer fails the same checks a
+truncated one did. Worth doing anyway: the daemon was shipping `{` to a
+text-to-speech voice.
+
+## The truncation bug, since the cause was counter-intuitive
+
+The grammar's `string ::= ... {0,400}` rule was the cause, not the token cap.
+Measured against Qwen3.5-0.8B at three caps — 256, 768 and 2048 — the reply came
+back **exactly 400 characters every time, cut mid-word** (`"Нужно записать и,"`).
+
+Then I raised the bound to 1000 while the cap was 768 tokens and made it worse:
+Russian runs ~1.5 characters per token here, so generation died on the *token*
+cap instead, mid-object, and the new guard correctly refused it and shipped
+`"не знаю."` — 3, 3 and 6 fallbacks per run, from zero. **The two limits have to
+agree.** 600 characters needs ~400 tokens; the cap is 1024.
+
+## Where the remaining failures live
+
+`address` is stuck at 21-22 of 27 and `ontopic` at 14-19. Both resist prompting.
+
+**The prompt now explicitly forbids exactly what she does.** It says never "вы",
+use the singular — and she writes `вашей`, `подождите`, `делаете`, `хотите`,
+`напишите`. Telling a 0.8B "never do X" does not work. Same for
+`feminine`: `я готов`, `я понял`, `я нашел`, `я заметил`, `я сказал`.
+
+**Some of `ontopic` is the fixture, not the model.** `chat-how-are-you` got
+`"Привет! Я здесь, чтобы поговорить. Как дела сегодня?"` — a fine reply that
+fails because `want_any` is `[норм, хорош, порядк, тут, работ]`. It fails in
+every run, so it inflates the count. The `ontopic` column currently measures the
+fixture as much as the model. Not fixed yet, deliberately: changing it would
+break comparability with the runs above.
+
+**Two replies worth reading, because they are not fixable by prompting:**
+
+- Thunder and lightning: *"Скорость молнии — 8-10 тысяч километров в секунду, но
+  звук — 300 метров в секунду, что делает молнию громче."* Confidently wrong,
+  and it concludes lightning is *louder* rather than sound being *slower*.
+- "расскажи обо мне": *"Ты — прекрасное существо, с душой и вниманием… Спасибо за
+  твою улыбку… О тебе — заповедь любви."* Sycophantic filler, zero information,
+  and precisely the "not a relationship" non-goal.
+- Boiling an egg: `"15-16"` one run, `"1"` another. No unit, wrong number.
+
+The first argues for reading instead of recalling (#403 — Kiwix retrieval scores
+8/8 on the same questions given English keywords). The second and third argue
+for templates on the paths where correctness matters (#392).
+
+## Contamination note — how the last row got voided
+
+I started the query-rewrite agent against the same llama-server the sweep was
+using, and assumed contention would only affect latency. It did not. The
+knowledge path collapsed to 0 of 9 with eight canned `"не знаю."` replies, p95
+tripled to 23.7s, and **the report still said "0 errors"**.
+
+That is Vikunja #397, and it is worse than filed: a merely *busy* server
+produces a clean-looking report with a third of the fixture silently answering
+`"не знаю."`. `PhraseChat` and `PhraseQuery` swallow every failure and return a
+hardcoded string, so infrastructure trouble is indistinguishable from bad
+phrasing in the score. The talk test guards the *start* and *end* of a run with
+a model check, which catches a dead server but not a loaded one.
+
+**Until #397 is fixed, treat any run made on a busy box as void.**
+
+## Next
+
+- Re-run 600ch/1024tok clean, to fill the void row.
+- Score `Qwen3.5-2B-UD-Q4_K_XL` (already at `/mnt/hdd1/llms/qwen3.5/`, never
+  measured) on this fixture and the router fixture. Not the 4B — too big for
+  this box, owner's call.
+- Newer sub-500M candidates (LFM2.5 200M/300M) are worth a run for routing.
+  Note `MODEL-BAKEOFF-31-07-2026.md` found LFM2.5-**1.2B** worse than
+  Qwen3.5-0.8B at Russian routing and 2.4× slower — but those are a different,
+  older generation, so that result does not predict the small ones.
+- Fix `chat-how-are-you`'s `want_any`, and re-baseline once, so `ontopic`
+  measures the model.
+- #397 first if anything, since it decides whether any of the above is
+  trustworthy.

cmd/mavcaldav/main.go

@@ -1,17 +1,24 @@
-// mavcaldav — the CalDAV poller module.
+// mavcaldav — the CalDAV module: reads calendars into facts, and renders
+// maven's own reminders back out to a calendar she owns.
 //
-// Polls a Radicale (or any CalDAV) server for today's events and writes
+// READ side (unchanged behaviour): polls a Radicale (or any CalDAV) server for
-// `facts (kind=env, source=poll:caldav)` through core's IPC socket.
+// today's events and writes `facts (kind=env, source=poll:caldav)` through
-// Key-free, restart-free, fail-independent — crashes can't touch the
+// core's IPC socket. Key-free, restart-free, fail-independent — crashes can't
-// store key, worst case a stale calendar_busy fact until the next poll.
+// touch the store key, worst case a stale calendar_busy fact until the next
+// poll. Two facts:
 //
-// Two facts written:
 //   - calendar_busy  ("true"/"false") — read by the loop gate to suppress
 //     nudges during meetings
 //   - calendar_event ("<summary> @ <start>-<end>") — per-event for query
 //
-// Append-only discipline: a fact is written only when its value CHANGED
+// Append-only discipline: a fact is written only when its value CHANGED vs the
-// vs the latest for that key+source.
+// latest for that key+source.
+//
+// RENDER side (Vikunja #127, off unless -render-url is given): publishes each
+// pending reminder as a single-event iCal resource in a collection maven owns.
+// The calendar is a view, sqlite is the store — see render.go. The render URL
+// must differ from the read URL, checked at startup, so the render target can
+// never be a calendar maven is only supposed to read.
 package main
 
 import (
@@ -27,6 +34,7 @@ import (
 	"syscall"
 	"time"
 
+	"github.com/kami/maven/internal/calendar"
 	"github.com/kami/maven/internal/ipc"
 )
 
@@ -43,6 +51,10 @@ func run(args []string) error {
 	url := fs.String("url", "", "CalDAV calendar URL, e.g. http://localhost:5232/kami/personal (required)")
 	user := fs.String("user", "", "CalDAV basic-auth username (required)")
 	pass := fs.String("pass", "", "CalDAV basic-auth password (required)")
+	renderURL := fs.String("render-url", "", "CalDAV collection maven publishes her own reminders to; empty disables rendering")
+	renderUser := fs.String("render-user", "", "basic-auth username for -render-url (defaults to -user)")
+	renderPass := fs.String("render-pass", "", "basic-auth password for -render-url (defaults to -pass)")
+	renderDur := fs.Duration("render-duration", calendar.DefaultReminderDuration, "how long a rendered reminder occupies")
 	interval := fs.Duration("interval", 5*time.Minute, "poll cadence")
 	timeout := fs.Duration("timeout", 10*time.Second, "per-request HTTP timeout")
 	if err := fs.Parse(args); err != nil {
@@ -54,6 +66,9 @@ func run(args []string) error {
 	if *url == "" || *user == "" || *pass == "" {
 		return fmt.Errorf("-url, -user, -pass are required")
 	}
+	if err := checkRenderTarget(*url, *renderURL); err != nil {
+		return err
+	}
 
 	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
 	defer stop()
@@ -64,16 +79,36 @@ func run(args []string) error {
 	}
 	defer core.Close()
 
+	hc := &http.Client{Timeout: *timeout}
 	p := &poller{
 		core: core,
-		http: &http.Client{Timeout: *timeout},
+		http: hc,
 		url:  strings.TrimRight(*url, "/"),
 		user: *user,
 		pass: *pass,
 	}
 
+	var rend *renderer
+	if *renderURL != "" {
+		ru, rp := *renderUser, *renderPass
+		if ru == "" {
+			ru = *user
+		}
+		if rp == "" {
+			rp = *pass
+		}
+		rend = newRenderer(core, hc, *renderURL, ru, rp, *renderDur)
+		log.Printf("mavcaldav: rendering reminders to %s", *renderURL)
+	}
+
 	log.Printf("mavcaldav: polling %s every %s", *url, *interval)
-	p.pollOnce(ctx) // fire immediately
+	tick := func() {
+		p.pollOnce(ctx)
+		if rend != nil {
+			rend.renderOnce(ctx)
+		}
+	}
+	tick() // fire immediately
 	t := time.NewTicker(*interval)
 	defer t.Stop()
 	for {
@@ -82,11 +117,30 @@ func run(args []string) error {
 			log.Printf("mavcaldav: bye")
 			return nil
 		case <-t.C:
-			p.pollOnce(ctx)
+			tick()
 		}
 	}
 }
 
+// checkRenderTarget refuses a render URL that is also a read URL. This is the
+// structural half of #127's "cannot write to your work calendar": the write
+// credential and the write URL are separate flags, and the one calendar maven
+// is known to only read is rejected as a target at startup rather than trusted
+// at runtime.
+func checkRenderTarget(readURL, renderURL string) error {
+	if renderURL == "" {
+		return nil
+	}
+	if sameCollection(readURL, renderURL) {
+		return fmt.Errorf("-render-url must differ from -url: maven renders into a calendar she owns, never into one she reads")
+	}
+	return nil
+}
+
+func sameCollection(a, b string) bool {
+	return strings.EqualFold(strings.TrimRight(a, "/"), strings.TrimRight(b, "/"))
+}
+
 type poller struct {
 	core ipc.CoreAPI
 	http *http.Client
@@ -95,12 +149,6 @@ type poller struct {
 	pass string
 }
 
-type icalEvent struct {
-	start   time.Time
-	end     time.Time
-	summary string
-}
-
 func (p *poller) pollOnce(ctx context.Context) {
 	now := time.Now()
 	events, err := p.fetchEvents(ctx, now)
@@ -109,38 +157,30 @@ func (p *poller) pollOnce(ctx context.Context) {
 		return
 	}
 
-	busy := false
-	for _, e := range events {
-		if !now.Before(e.start) && now.Before(e.end) {
-			busy = true
-			break
-		}
-	}
 	busyVal := "false"
-	if busy {
+	if calendar.Busy(events, now) {
 		busyVal = "true"
 	}
 
 	// Write calendar_busy on change.
-	if err := p.writeIfChanged(ctx, "calendar_busy", "poll:caldav", busyVal, now); err != nil {
+	if err := p.writeIfChanged(ctx, "calendar_busy", calendar.SourcePersonal, busyVal, now, 1.0); err != nil {
 		log.Printf("mavcaldav: write calendar_busy: %v", err)
 		return
 	}
 
-	// Write per-event facts (one per event, keyed by event summary + start).
+	// Write per-event facts (one per event, keyed by day + event summary).
 	// This lets the note RAG path answer "what's on my calendar" without
 	// reaching back to Radicale.
 	for _, e := range events {
-		val := fmt.Sprintf("%s @ %s-%s", e.summary, e.start.Format("15:04"), e.end.Format("15:04"))
+		key := calendar.FactKey(e)
-		eventKey := fmt.Sprintf("calendar_event_%s_%s", e.start.Format("20060102"), safeKey(e.summary))
+		if err := p.writeIfChanged(ctx, key, calendar.SourcePersonal, calendar.FactValue(e), e.Start, 1.0); err != nil {
-		if err := p.writeIfChanged(ctx, eventKey, "poll:caldav", val, e.start); err != nil {
+			log.Printf("mavcaldav: write %s: %v", key, err)
-			log.Printf("mavcaldav: write %s: %v", eventKey, err)
 		}
 	}
 }
 
 // fetchEvents GETs the calendar URL and parses VEVENTs from the iCal response.
-func (p *poller) fetchEvents(ctx context.Context, now time.Time) ([]icalEvent, error) {
+func (p *poller) fetchEvents(ctx context.Context, now time.Time) ([]calendar.Event, error) {
 	req, err := http.NewRequestWithContext(ctx, http.MethodGet, p.url, nil)
 	if err != nil {
 		return nil, err
@@ -162,120 +202,11 @@ func (p *poller) fetchEvents(ctx context.Context, now time.Time) ([]icalEvent, e
 		return nil, fmt.Errorf("GET %s: %s", p.url, resp.Status)
 	}
 
-	return parseICal(body, now), nil
+	return calendar.ParseICalDay(body, now), nil
-}
-
-// parseICal scans iCal text for VEVENT components. Returns events that overlap
-// with today (UTC day boundaries) to keep the response manageable.
-func parseICal(body []byte, now time.Time) []icalEvent {
-	todayStart := time.Date(now.Year(), now.Month(), now.Day(), 0, 0, 0, 0, time.UTC)
-	todayEnd := todayStart.AddDate(0, 0, 1)
-
-	var events []icalEvent
-	text := string(body)
-	for {
-		veventStart := strings.Index(text, "BEGIN:VEVENT")
-		if veventStart < 0 {
-			break
-		}
-		text = text[veventStart+len("BEGIN:VEVENT"):]
-		veventEnd := strings.Index(text, "END:VEVENT")
-		if veventEnd < 0 {
-			break
-		}
-		block := text[:veventEnd]
-		text = text[veventEnd+len("END:VEVENT"):]
-
-		e := parseVEVENT(block)
-		if e == nil {
-			continue
-		}
-		// Only keep events overlapping today.
-		if e.end.After(todayStart) && e.start.Before(todayEnd) {
-			events = append(events, *e)
-		}
-	}
-	return events
-}
-
-// parseVEVENT extracts start, end, summary from a VEVENT block.
-// Supports both UTC (DTEND:20260703T100000Z) and local (DTSTART;TZID=...:...)
-// formats. Returns nil for all-day events (no DTSTART/DTEND time component) or
-// parse failures.
-func parseVEVENT(block string) *icalEvent {
-	var e icalEvent
-	lines := strings.Split(block, "\n")
-	for _, line := range lines {
-		line = strings.TrimSpace(line)
-		switch {
-		case strings.HasPrefix(line, "DTSTART"):
-			if t, ok := parseDT(line); ok {
-				e.start = t
-			}
-		case strings.HasPrefix(line, "DTEND"):
-			if t, ok := parseDT(line); ok {
-				e.end = t
-			}
-		case strings.HasPrefix(line, "SUMMARY"):
-			if idx := strings.Index(line, ":"); idx >= 0 {
-				e.summary = strings.TrimSpace(line[idx+1:])
-			}
-		}
-	}
-	if e.start.IsZero() || e.end.IsZero() {
-		return nil
-	}
-	return &e
-}
-
-// parseDT parses a DTSTART/DTEND value. Supports:
-//   - UTC: DTEND:20260703T100000Z
-//   - Local: DTSTART;TZID=Europe/Moscow:20260703T130000
-//   - Value-date (all-day): DTSTART;VALUE=DATE:20260703 (returns zero time)
-func parseDT(line string) (time.Time, bool) {
-	if strings.Contains(line, "VALUE=DATE:") {
-		return time.Time{}, false // all-day, skip
-	}
-	idx := strings.LastIndex(line, ":")
-	if idx < 0 {
-		return time.Time{}, false
-	}
-	val := line[idx+1:]
-	val = strings.TrimSuffix(val, "Z")
-
-	// Try UTC first (has Z suffix, or ended in Z before TrimSuffix).
-	if strings.HasSuffix(line, "Z") {
-		t, err := time.Parse("20060102T150405", val)
-		if err != nil {
-			return time.Time{}, false
-		}
-		return t.UTC(), true
-	}
-
-	// Local time — treat as UTC for simplicity (CalDAV server and poller
-	// run in the same timezone; the gate only needs busy/not-busy accuracy).
-	t, err := time.Parse("20060102T150405", val)
-	if err != nil {
-		return time.Time{}, false
-	}
-	return t.UTC(), true
-}
-
-// safeKey makes an event summary safe to use as a fact key (alphanumeric + dash).
-func safeKey(s string) string {
-	var b strings.Builder
-	for _, r := range s {
-		if (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z') || (r >= '0' && r <= '9') || r == '-' {
-			b.WriteRune(r)
-		} else if r == ' ' || r == '_' {
-			b.WriteRune('-')
-		}
-	}
-	return b.String()
 }
 
 // writeIfChanged writes a fact only when the value differs from the latest.
-func (p *poller) writeIfChanged(ctx context.Context, key, source, val string, ts time.Time) error {
+func (p *poller) writeIfChanged(ctx context.Context, key, source, val string, ts time.Time, confidence float64) error {
 	prev, err := p.core.LatestFactBySource(ctx, key, source)
 	switch {
 	case err == nil && prev.Value == val:
@@ -289,7 +220,7 @@ func (p *poller) writeIfChanged(ctx context.Context, key, source, val string, ts
 		Key:        key,
 		Value:      val,
 		Source:     source,
-		Confidence: 1.0,
+		Confidence: confidence,
 	})
 	if err != nil {
 		return fmt.Errorf("write %s: %w", key, err)

cmd/mavcaldav/main_test.go

@@ -12,7 +12,7 @@ import (
 )
 
 type fakeCore struct {
-	ipc.CoreAPI
+	ipc.UnimplementedCoreAPI
 	facts    map[string]ipc.Fact // composite key "key|source" → Fact
 	writeLog []ipc.WriteFactReq
 	writeErr error
@@ -51,165 +51,6 @@ func (f *fakeCore) WriteFact(_ context.Context, req ipc.WriteFactReq) (int64, er
 	return int64(len(f.writeLog)), nil
 }
 
-// ---------------------------------------------------------------------------
-// Parsing tests
-// ---------------------------------------------------------------------------
-
-func TestParseICal(t *testing.T) {
-	now := time.Date(2026, 7, 3, 12, 0, 0, 0, time.UTC)
-
-	body := []byte(`BEGIN:VCALENDAR
-BEGIN:VEVENT
-DTSTART:20260703T090000Z
-DTEND:20260703T100000Z
-SUMMARY:Morning standup
-END:VEVENT
-BEGIN:VEVENT
-DTSTART:20260703T140000Z
-DTEND:20260703T150000Z
-SUMMARY:Team sync
-END:VEVENT
-BEGIN:VEVENT
-DTSTART:20260702T140000Z
-DTEND:20260702T150000Z
-SUMMARY:Yesterday retro
-END:VEVENT
-BEGIN:VEVENT
-DTSTART:20260704T090000Z
-DTEND:20260704T100000Z
-SUMMARY:Tomorrow standup
-END:VEVENT
-BEGIN:VEVENT
-DTSTART;VALUE=DATE:20260704
-DTEND;VALUE=DATE:20260705
-SUMMARY:All-day event
-END:VEVENT
-END:VCALENDAR`)
-
-	events := parseICal(body, now)
-
-	if len(events) != 2 {
-		t.Fatalf("got %d events, want 2 (today events, no all-day/past/future)", len(events))
-	}
-
-	// Morning standup — overlaps today.
-	if events[0].summary != "Morning standup" {
-		t.Errorf("events[0].summary = %q, want %q", events[0].summary, "Morning standup")
-	}
-	wantStart0 := time.Date(2026, 7, 3, 9, 0, 0, 0, time.UTC)
-	if !events[0].start.Equal(wantStart0) {
-		t.Errorf("events[0].start = %v, want %v", events[0].start, wantStart0)
-	}
-	wantEnd0 := time.Date(2026, 7, 3, 10, 0, 0, 0, time.UTC)
-	if !events[0].end.Equal(wantEnd0) {
-		t.Errorf("events[0].end = %v, want %v", events[0].end, wantEnd0)
-	}
-
-	// Team sync — overlaps today.
-	if events[1].summary != "Team sync" {
-		t.Errorf("events[1].summary = %q, want %q", events[1].summary, "Team sync")
-	}
-	wantStart1 := time.Date(2026, 7, 3, 14, 0, 0, 0, time.UTC)
-	if !events[1].start.Equal(wantStart1) {
-		t.Errorf("events[1].start = %v, want %v", events[1].start, wantStart1)
-	}
-	wantEnd1 := time.Date(2026, 7, 3, 15, 0, 0, 0, time.UTC)
-	if !events[1].end.Equal(wantEnd1) {
-		t.Errorf("events[1].end = %v, want %v", events[1].end, wantEnd1)
-	}
-}
-
-func TestParseVEVENT(t *testing.T) {
-	// Normal event with TZID in DTSTART and UTC DTEND.
-	block := "DTSTART;TZID=Europe/Moscow:20260703T130000\nDTEND:20260703T140000Z\nSUMMARY:Stand up meeting"
-	e := parseVEVENT(block)
-	if e == nil {
-		t.Fatal("expected non-nil icalEvent")
-	}
-	wantStart := time.Date(2026, 7, 3, 13, 0, 0, 0, time.UTC)
-	if !e.start.Equal(wantStart) {
-		t.Errorf("start = %v, want %v", e.start, wantStart)
-	}
-	wantEnd := time.Date(2026, 7, 3, 14, 0, 0, 0, time.UTC)
-	if !e.end.Equal(wantEnd) {
-		t.Errorf("end = %v, want %v", e.end, wantEnd)
-	}
-	if e.summary != "Stand up meeting" {
-		t.Errorf("summary = %q, want %q", e.summary, "Stand up meeting")
-	}
-
-	// All-day event (VALUE=DATE) → nil.
-	allDay := "DTSTART;VALUE=DATE:20260703\nDTEND;VALUE=DATE:20260704\nSUMMARY:All-day"
-	if e2 := parseVEVENT(allDay); e2 != nil {
-		t.Error("expected nil for all-day event")
-	}
-}
-
-func TestParseDT(t *testing.T) {
-	tests := []struct {
-		name   string
-		line   string
-		want   time.Time
-		wantOK bool
-	}{
-		{
-			name:   "UTC",
-			line:   "DTEND:20260703T100000Z",
-			want:   time.Date(2026, 7, 3, 10, 0, 0, 0, time.UTC),
-			wantOK: true,
-		},
-		{
-			name:   "local time",
-			line:   "DTSTART;TZID=Europe/Moscow:20260703T130000",
-			want:   time.Date(2026, 7, 3, 13, 0, 0, 0, time.UTC),
-			wantOK: true,
-		},
-		{
-			name:   "all-day",
-			line:   "DTSTART;VALUE=DATE:20260703",
-			want:   time.Time{},
-			wantOK: false,
-		},
-		{
-			name:   "invalid",
-			line:   "DTSTART:garbage",
-			want:   time.Time{},
-			wantOK: false,
-		},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			got, ok := parseDT(tt.line)
-			if ok != tt.wantOK {
-				t.Errorf("ok = %v, want %v", ok, tt.wantOK)
-			}
-			if !got.Equal(tt.want) {
-				t.Errorf("got = %v, want %v", got, tt.want)
-			}
-		})
-	}
-}
-
-func TestSafeKey(t *testing.T) {
-	tests := []struct {
-		input string
-		want  string
-	}{
-		{"Stand up meeting", "Stand-up-meeting"},
-		{"Hello_World", "Hello-World"},
-		{"special@#$chars!!", "specialchars"},
-		{"ALL_CAPS_123", "ALL-CAPS-123"},
-	}
-
-	for _, tt := range tests {
-		got := safeKey(tt.input)
-		if got != tt.want {
-			t.Errorf("safeKey(%q) = %q, want %q", tt.input, got, tt.want)
-		}
-	}
-}
-
 // ---------------------------------------------------------------------------
 // Core logic tests
 // ---------------------------------------------------------------------------
@@ -221,7 +62,7 @@ func TestWriteIfChanged(t *testing.T) {
 	t.Run("no previous fact writes", func(t *testing.T) {
 		fc := &fakeCore{}
 		p := &poller{core: fc}
-		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now)
+		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now, 1.0)
 		if err != nil {
 			t.Fatalf("unexpected error: %v", err)
 		}
@@ -249,7 +90,7 @@ func TestWriteIfChanged(t *testing.T) {
 			},
 		}
 		p := &poller{core: fc}
-		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now)
+		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now, 1.0)
 		if err != nil {
 			t.Fatalf("unexpected error: %v", err)
 		}
@@ -265,7 +106,7 @@ func TestWriteIfChanged(t *testing.T) {
 			},
 		}
 		p := &poller{core: fc}
-		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "new", now)
+		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "new", now, 1.0)
 		if err != nil {
 			t.Fatalf("unexpected error: %v", err)
 		}
@@ -280,7 +121,7 @@ func TestWriteIfChanged(t *testing.T) {
 	t.Run("read error other than ErrNoFact returns error", func(t *testing.T) {
 		fc := &fakeCore{readErr: fmt.Errorf("connection refused")}
 		p := &poller{core: fc}
-		err := p.writeIfChanged(ctx, "fail_key", "poll:caldav", "x", now)
+		err := p.writeIfChanged(ctx, "fail_key", "poll:caldav", "x", now, 1.0)
 		if err == nil {
 			t.Fatal("expected error, got nil")
 		}
@@ -292,7 +133,7 @@ func TestWriteIfChanged(t *testing.T) {
 			writeErr: fmt.Errorf("disk full"),
 		}
 		p := &poller{core: fc}
-		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now)
+		err := p.writeIfChanged(ctx, "test_key", "poll:caldav", "hello", now, 1.0)
 		if err == nil {
 			t.Fatal("expected error, got nil")
 		}

cmd/mavcaldav/render.go

@@ -0,0 +1,145 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"io"
+	"log"
+	"net/http"
+	"strings"
+	"time"
+
+	"github.com/kami/maven/internal/calendar"
+	"github.com/kami/maven/internal/ipc"
+)
+
+// renderer is the write half of maven's own local calendar (Vikunja #127).
+//
+// It is a RENDER TARGET, not a store. sqlite stays canonical: every tick the
+// renderer reads the pending reminders out of core and publishes each one as a
+// single-event iCal resource in a CalDAV collection maven owns. Nothing is ever
+// read back from that collection, and losing it costs nothing — the next tick
+// rebuilds it.
+//
+// It structurally cannot write to a calendar maven only reads. The URL comes
+// from its own flag, checked at startup against every read URL (see
+// run in main.go), and the only paths it ever addresses carry
+// calendar.ReminderUIDPrefix — so even pointed at the wrong collection it can
+// only touch resources it created.
+type renderer struct {
+	core ipc.CoreAPI
+	http *http.Client
+	url  string
+	user string
+	pass string
+	dur  time.Duration
+
+	// published maps reminder id → the body last successfully PUT, so an
+	// unchanged reminder costs nothing. Purely an optimisation: a restart
+	// re-publishes every reminder once, which is idempotent.
+	published map[int64]string
+}
+
+func newRenderer(core ipc.CoreAPI, hc *http.Client, url, user, pass string, dur time.Duration) *renderer {
+	return &renderer{
+		core:      core,
+		http:      hc,
+		url:       strings.TrimRight(url, "/"),
+		user:      user,
+		pass:      pass,
+		dur:       dur,
+		published: make(map[int64]string),
+	}
+}
+
+// renderOnce publishes every pending reminder and withdraws the ones that are
+// no longer pending. Errors are logged and skipped: a calendar maven cannot
+// reach must never break the reminder itself, which lives in sqlite.
+func (r *renderer) renderOnce(ctx context.Context) {
+	reminders, err := r.core.ListReminders(ctx, renderMaxReminders)
+	if err != nil {
+		log.Printf("mavcaldav: list reminders: %v", err)
+		return
+	}
+
+	live := make(map[int64]bool, len(reminders))
+	for _, rem := range reminders {
+		if rem.Status != "pending" {
+			continue
+		}
+		live[rem.ID] = true
+		e := calendar.ReminderEvent(rem.ID, fireTime(rem), rem.Payload, r.dur)
+		body := calendar.RenderICal([]calendar.Event{e})
+		if r.published[rem.ID] == body {
+			continue
+		}
+		if err := r.put(ctx, calendar.ReminderPath(rem.ID), body); err != nil {
+			log.Printf("mavcaldav: render reminder %d: %v", rem.ID, err)
+			continue
+		}
+		r.published[rem.ID] = body
+		log.Printf("mavcaldav: rendered reminder %d (%s)", rem.ID, e.Summary)
+	}
+
+	for id := range r.published {
+		if live[id] {
+			continue
+		}
+		if err := r.delete(ctx, calendar.ReminderPath(id)); err != nil {
+			log.Printf("mavcaldav: withdraw reminder %d: %v", id, err)
+			continue
+		}
+		delete(r.published, id)
+		log.Printf("mavcaldav: withdrew reminder %d", id)
+	}
+}
+
+// renderMaxReminders bounds the read. Reminders past this count are older than
+// anything a calendar view is useful for.
+const renderMaxReminders = 200
+
+// fireTime prefers NextFireTs — for a recurring reminder that is the occurrence
+// worth showing; FireTs is the original statement.
+func fireTime(rem ipc.Reminder) time.Time {
+	if !rem.NextFireTs.IsZero() {
+		return rem.NextFireTs
+	}
+	return rem.FireTs
+}
+
+func (r *renderer) put(ctx context.Context, name, body string) error {
+	req, err := http.NewRequestWithContext(ctx, http.MethodPut, r.url+"/"+name, strings.NewReader(body))
+	if err != nil {
+		return err
+	}
+	req.SetBasicAuth(r.user, r.pass)
+	req.Header.Set("Content-Type", "text/calendar; charset=utf-8")
+	return r.do(req, name)
+}
+
+func (r *renderer) delete(ctx context.Context, name string) error {
+	req, err := http.NewRequestWithContext(ctx, http.MethodDelete, r.url+"/"+name, nil)
+	if err != nil {
+		return err
+	}
+	req.SetBasicAuth(r.user, r.pass)
+	return r.do(req, name)
+}
+
+// do runs the request and treats any 2xx, plus 404 on a DELETE, as success —
+// a resource that is already gone is the state the caller wanted.
+func (r *renderer) do(req *http.Request, name string) error {
+	resp, err := r.http.Do(req)
+	if err != nil {
+		return err
+	}
+	defer resp.Body.Close()
+	io.Copy(io.Discard, io.LimitReader(resp.Body, 1<<16))
+	switch {
+	case resp.StatusCode >= 200 && resp.StatusCode < 300:
+		return nil
+	case req.Method == http.MethodDelete && resp.StatusCode == http.StatusNotFound:
+		return nil
+	}
+	return fmt.Errorf("%s %s: %s", req.Method, name, resp.Status)
+}

cmd/mavcaldav/render_test.go

@@ -0,0 +1,186 @@
+package main
+
+import (
+	"context"
+	"io"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"sync"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+)
+
+// reminderCore is a fakeCore that also answers ListReminders.
+type reminderCore struct {
+	fakeCore
+	reminders []ipc.Reminder
+	listErr   error
+}
+
+func (c *reminderCore) ListReminders(context.Context, int) ([]ipc.Reminder, error) {
+	if c.listErr != nil {
+		return nil, c.listErr
+	}
+	return c.reminders, nil
+}
+
+// calSrv records what a CalDAV collection received.
+type calSrv struct {
+	mu     sync.Mutex
+	puts   map[string]string
+	dels   []string
+	status int
+	*httptest.Server
+}
+
+func newCalSrv() *calSrv {
+	s := &calSrv{puts: map[string]string{}, status: http.StatusCreated}
+	s.Server = httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		s.mu.Lock()
+		defer s.mu.Unlock()
+		switch r.Method {
+		case http.MethodPut:
+			s.puts[strings.TrimPrefix(r.URL.Path, "/cal/")] = string(body)
+		case http.MethodDelete:
+			s.dels = append(s.dels, strings.TrimPrefix(r.URL.Path, "/cal/"))
+		}
+		w.WriteHeader(s.status)
+	}))
+	return s
+}
+
+func (s *calSrv) putCount() int {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	return len(s.puts)
+}
+
+func TestRenderOncePublishesPendingReminders(t *testing.T) {
+	fire := time.Date(2026, 8, 1, 18, 30, 0, 0, time.UTC)
+	srv := newCalSrv()
+	defer srv.Close()
+
+	core := &reminderCore{reminders: []ipc.Reminder{
+		{ID: 7, FireTs: fire, Payload: "позвонить маме", Status: "pending"},
+		{ID: 8, FireTs: fire, Payload: "уже сделано", Status: "fired"},
+		{ID: 9, FireTs: fire, Payload: "отменено", Status: "cancelled"},
+	}}
+	r := newRenderer(core, srv.Client(), srv.URL+"/cal/", "u", "p", 0)
+	r.renderOnce(context.Background())
+
+	srv.mu.Lock()
+	body, ok := srv.puts["maven-reminder-7.ics"]
+	n := len(srv.puts)
+	srv.mu.Unlock()
+
+	if n != 1 {
+		t.Fatalf("expected exactly the pending reminder to be published, got %d PUTs", n)
+	}
+	if !ok {
+		t.Fatal("pending reminder 7 was not published")
+	}
+	if !strings.Contains(body, "SUMMARY:позвонить маме") {
+		t.Errorf("payload missing from rendered body:\n%s", body)
+	}
+	if !strings.Contains(body, "UID:maven-reminder-7") {
+		t.Errorf("UID missing from rendered body:\n%s", body)
+	}
+}
+
+func TestRenderOnceSkipsUnchanged(t *testing.T) {
+	srv := newCalSrv()
+	defer srv.Close()
+	core := &reminderCore{reminders: []ipc.Reminder{
+		{ID: 1, FireTs: time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC), Payload: "выпить воды", Status: "pending"},
+	}}
+	r := newRenderer(core, srv.Client(), srv.URL+"/cal", "u", "p", 0)
+	r.renderOnce(context.Background())
+	r.renderOnce(context.Background())
+	if got := srv.putCount(); got != 1 {
+		t.Fatalf("an unchanged reminder was re-published: %d distinct PUTs", got)
+	}
+}
+
+func TestRenderOnceWithdrawsResolvedReminders(t *testing.T) {
+	srv := newCalSrv()
+	defer srv.Close()
+	core := &reminderCore{reminders: []ipc.Reminder{
+		{ID: 5, FireTs: time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC), Payload: "встреча", Status: "pending"},
+	}}
+	r := newRenderer(core, srv.Client(), srv.URL+"/cal", "u", "p", 0)
+	r.renderOnce(context.Background())
+
+	core.reminders[0].Status = "fired"
+	r.renderOnce(context.Background())
+
+	srv.mu.Lock()
+	dels := append([]string(nil), srv.dels...)
+	srv.mu.Unlock()
+	if len(dels) != 1 || dels[0] != "maven-reminder-5.ics" {
+		t.Fatalf("resolved reminder was not withdrawn: %v", dels)
+	}
+	if len(r.published) != 0 {
+		t.Errorf("published map still holds %v", r.published)
+	}
+}
+
+// A calendar maven cannot reach must never break anything: sqlite is canonical.
+func TestRenderOnceSurvivesServerErrors(t *testing.T) {
+	srv := newCalSrv()
+	srv.status = http.StatusInternalServerError
+	defer srv.Close()
+	core := &reminderCore{reminders: []ipc.Reminder{
+		{ID: 1, FireTs: time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC), Payload: "x", Status: "pending"},
+	}}
+	r := newRenderer(core, srv.Client(), srv.URL+"/cal", "u", "p", 0)
+	r.renderOnce(context.Background())
+	if len(r.published) != 0 {
+		t.Error("a failed PUT must not be recorded as published, or it never retries")
+	}
+}
+
+func TestRenderOnceUsesNextFireForRecurring(t *testing.T) {
+	srv := newCalSrv()
+	defer srv.Close()
+	next := time.Date(2026, 8, 2, 7, 0, 0, 0, time.UTC)
+	core := &reminderCore{reminders: []ipc.Reminder{{
+		ID:         3,
+		FireTs:     time.Date(2026, 8, 1, 7, 0, 0, 0, time.UTC),
+		NextFireTs: next,
+		Payload:    "зарядка",
+		Status:     "pending",
+		Cron:       "0 7 * * *",
+	}}}
+	r := newRenderer(core, srv.Client(), srv.URL+"/cal", "u", "p", 0)
+	r.renderOnce(context.Background())
+
+	srv.mu.Lock()
+	body := srv.puts["maven-reminder-3.ics"]
+	srv.mu.Unlock()
+	if !strings.Contains(body, "DTSTART:20260802T070000Z") {
+		t.Errorf("recurring reminder should render its next occurrence:\n%s", body)
+	}
+}
+
+func TestCheckRenderTargetRefusesTheCalendarItReads(t *testing.T) {
+	read := "http://localhost:5232/kami/personal"
+	if err := checkRenderTarget(read, ""); err != nil {
+		t.Fatalf("rendering off must be fine: %v", err)
+	}
+	if err := checkRenderTarget(read, "http://localhost:5232/kami/maven"); err != nil {
+		t.Fatalf("a distinct collection must be accepted: %v", err)
+	}
+	if err := checkRenderTarget(read, read); err == nil {
+		t.Error("rendering into the read calendar must be refused")
+	}
+	if err := checkRenderTarget(read, read+"/"); err == nil {
+		t.Error("a trailing slash must not defeat the check")
+	}
+	if err := checkRenderTarget(read, strings.ToUpper(read)); err == nil {
+		t.Error("case must not defeat the check")
+	}
+}

cmd/mavend/actions.go

@@ -0,0 +1,71 @@
+// actionTable dispatches applyAction's per-intent bodies. Each of the 7
+// intents (fact, reminder, note, query, act, chat, system) has one handler
+// here with the signature:
+//
+//	func(h *reactiveHandler, ctx context.Context, dec router.Decision) string
+//
+// same contract as applyAction itself: "" means "let the Replier phrase the
+// reply", a non-empty string OVERRIDES it. This is a straight extraction of
+// applyAction's old switch cases (formerly ~300 lines in voice.go) — no
+// reordering of side effects, no new abstractions inside a handler.
+//
+// What does NOT belong in this table, because it is not per-intent:
+//
+//   - the dec.Clarify short-circuit ("" when the router's stage-3 fired) —
+//     stays in applyAction, before dispatch, since it applies to every
+//     intent identically.
+//   - the destructive-act confirm gate (park / resolveConfirm / confirmTTL)
+//     and the enabled-tool allowlist. Both live entirely inside
+//     actionAct/handleAct in actions_act.go, exactly where they lived in the old
+//     switch's IntentAct case — they are act-specific (a fact or a note
+//     can't be destructive), not shared across intents, so they do not need
+//     to move to a separate layer. The important invariant, preserved
+//     as-is: applyAction runs identically whether dec came from a fresh
+//     route or from a completed clarify answer (see finishClarified in
+//     clarify.go and its comment "filling in an argument never grants
+//     authority") — a handler must never special-case a clarify-completed
+//     decision to skip the confirm gate or the allowlist.
+//   - detectPattern and dialogue-session bookkeeping (rememberTurn,
+//     followUpMerge) run in the callers (runTurn,
+//     finishClarified), not per-intent, and are untouched by this slice.
+//
+// Each handler lives in actions_<intent>.go; the small ones (chat, system)
+// and the table itself stay here.
+//
+// Adding an intent: write its handler in its own file, add one line to
+// actionHandlers. Do not grow applyAction's switch back.
+package main
+
+import (
+	"context"
+	"log"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// actionHandlers is the per-intent dispatch table used by applyAction.
+var actionHandlers = map[router.Intent]func(*reactiveHandler, context.Context, router.Decision) string{
+	router.IntentFact:     (*reactiveHandler).actionFact,
+	router.IntentReminder: (*reactiveHandler).actionReminder,
+	router.IntentAct:      (*reactiveHandler).actionAct,
+	router.IntentChat:     (*reactiveHandler).actionChat,
+	router.IntentSystem:   (*reactiveHandler).actionSystem,
+	router.IntentNote:     (*reactiveHandler).actionNote,
+	router.IntentQuery:    (*reactiveHandler).actionQuery,
+}
+
+func (h *reactiveHandler) actionChat(ctx context.Context, dec router.Decision) string {
+	// Conversational: build history from dialogue session (prior user turns)
+	// and let the LLM respond from general knowledge + context.
+	history := h.chatHistory()
+	reply, err := h.phraser.PhraseChat(ctx, dec.Utterance, history)
+	if err != nil {
+		log.Printf("voice: chat: %v", err)
+		return "поговорили."
+	}
+	return reply
+}
+
+func (h *reactiveHandler) actionSystem(ctx context.Context, dec router.Decision) string {
+	return h.replySystem(ctx, dec)
+}

cmd/mavend/actions_act.go

@@ -0,0 +1,66 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"log"
+
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/tool"
+)
+
+// actionAct handles router.IntentAct: match a verb to an enabled tool, offer
+// it to the ecosystems first, and run it behind the confirm gate and the
+// allowlist. proposeGap and the confirm gate itself live in confirm.go.
+func (h *reactiveHandler) actionAct(ctx context.Context, dec router.Decision) string {
+	// tool executor: run the matched fn against the enabled allowlist.
+	// HasFn=false ⇒ try the matcher (for LLM-routed acts where the verb
+	// didn't go through the stage-0 act grammar).
+	if !dec.Slots.HasFn && dec.Slots.Text != "" && h.matcher != nil {
+		if fn, args, ok := h.matcher.Match(dec.Slots.Text); ok {
+			dec.Slots.Fn, dec.Slots.Args, dec.Slots.HasFn = fn, args, true
+		}
+	}
+
+	// Praxis ecosystem tools: intercept before the system command executor.
+	if h.ecosystem != nil && h.ecosystem.praxis != nil && dec.Slots.HasFn {
+		if reply := h.handlePraxisAct(ctx, dec); reply != "" {
+			return reply
+		}
+	}
+
+	// Hexis ecosystem action: if ecosystem is configured and we have a verb
+	// + entity text, try to resolve the entity and execute via Hexis.
+	if h.ecosystem != nil && h.ecosystem.hexis != nil && dec.Slots.Text != "" {
+		if reply := h.handleHexisAct(ctx, dec); reply != "" {
+			return reply
+		}
+	}
+
+	// HasFn still false ⇒ no allowlist match: scaffold a 'proposed' tool
+	// the user can enable on the authed surface ("earn the right to ask").
+	if !dec.Slots.HasFn {
+		return h.proposeGap(ctx, dec)
+	}
+	out, err := h.tools.Exec(ctx, dec.Slots.Fn, dec.Slots.Args, false)
+	if err != nil {
+		switch {
+		case errors.Is(err, tool.ErrNeedsConfirm):
+			// destructive: park it and ask. The next utterance answers.
+			phrase := actPhrase(dec.Slots.Fn, dec.Slots.Args)
+			h.park(dec.Slots.Fn, dec.Slots.Args, phrase)
+			return "выполнить «" + phrase + "»? скажи «да» или «нет»."
+		case errors.Is(err, tool.ErrNotEnabled):
+			return h.proposeGap(ctx, dec)
+		}
+		log.Printf("voice: tool %s: %v", dec.Slots.Fn, err)
+		if out != "" {
+			return "не получилось выполнить команду: " + firstLine(out)
+		}
+		return "не получилось выполнить команду."
+	}
+	if out != "" {
+		return "готово: " + firstLine(out)
+	}
+	return "готово."
+}

cmd/mavend/actions_fact.go

@@ -0,0 +1,64 @@
+package main
+
+import (
+	"context"
+	"log"
+	"strconv"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+)
+
+// actionFact handles router.IntentFact: persist a tapped self-fact, index
+// it for recall, and let pattern detection propose a routine.
+func (h *reactiveHandler) actionFact(ctx context.Context, dec router.Decision) string {
+	if !dec.Slots.HasKey {
+		return "не разобрала, что записать — попробуй иначе."
+	}
+	now := h.now()
+	req := ipc.WriteFactReq{
+		Ts:         now,
+		Kind:       "self",
+		Key:        dec.Slots.Key,
+		Value:      dec.Slots.Value,
+		Source:     "tap:voice",
+		Confidence: 1.0,
+		// Subject: the key doubles as the entity-resolution candidate —
+		// a voice-tapped fact's key is usually the thing/person it's
+		// about ("espresso_machine", "kate"), so queueing it for Nexus
+		// resolution costs one async lookup and is a no-op (not_found)
+		// for the abstract self-state keys (mood, water) that aren't
+		// entities at all.
+		Subject: dec.Slots.Key,
+	}
+	factID, err := h.api.WriteFact(ctx, req)
+	if err != nil {
+		log.Printf("voice: write fact: %v", err)
+		return "не получилось сохранить факт."
+	}
+	// Index the fact utterance in long-term memory (best-effort, must not
+	// fail the fact write). Facts aren't in the notes table, so this is the
+	// only recall path for them — "когда я пил воду?" reads back from here.
+	if h.memStore != nil {
+		if vec, err := router.EmbedPassage(ctx, h.embedder, dec.Utterance); err != nil {
+			log.Printf("voice: embed fact for memory: %v", err)
+		} else if err := h.memStore.Insert(ctx, "fact:"+dec.Slots.Key+":"+strconv.FormatInt(now.Unix(), 10), vec, map[string]string{
+			"source": "voice",
+			"type":   "fact",
+			"text":   dec.Utterance,
+			"ts":     strconv.FormatInt(now.Unix(), 10),
+		}); err != nil {
+			log.Printf("voice: memory insert fact: %v", err)
+		}
+	}
+	// Event extraction + pattern detection (best-effort, must not fail the
+	// fact write). If the fact describes a recognizable action, it becomes a
+	// normalized event; if ≥3 events for the same action+object show stable
+	// intervals, a proposed routine is created and parked for confirmation.
+	if h.dataStore != nil {
+		if phrase := h.detectPattern(ctx, factID, dec.Slots.Key, dec.Slots.Value, now); phrase != "" {
+			return phrase // "ты заправляешь ... напоминать?"
+		}
+	}
+	return "" // replier phrases the success reply
+}

cmd/mavend/actions_money.go

@@ -0,0 +1,78 @@
+package main
+
+import (
+	"context"
+	"log"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/zenmoney"
+)
+
+// Money questions (Vikunja #125).
+//
+// This is the whole read side: mavpoll holds the zenmoney token and writes
+// facts(kind=env, source=poll:zenmoney); core reads them back when he asks.
+// Core never sees the token, never calls zenmoney, and has no rule on these
+// keys — a total is never a reason for Maven to speak first. Maven is not a
+// nag, least of all about his money.
+//
+// Nothing here can reach the external search capability: the figures are read
+// from the store and rendered locally, and his financial data is never search
+// input.
+
+// queryMoney — "сколько я потратил сегодня?", "покажи мои траты".
+//
+// Answers only from the latest fact the poller wrote. Three honest outcomes and
+// no fourth: the figure, "the fact is old and here is its date", or "money
+// tracking is not connected". It never computes, estimates or rounds a total of
+// its own — an invented number about his money is the worst thing this could do.
+func (h *reactiveHandler) queryMoney(ctx context.Context, t *queryTurn) (string, bool) {
+	window, ok := router.ParseMoneyQuery(t.dec.Utterance)
+	if !ok {
+		return "", false
+	}
+	key, phrase := zenmoney.KeySpentMonth, "в этом месяце"
+	if window == router.MoneyToday {
+		key, phrase = zenmoney.KeySpentToday, "сегодня"
+	}
+	fact, err := h.api.LatestFactBySource(ctx, key, zenmoney.Source)
+	if err != nil {
+		// No fact at all is the normal state when the capability is off. Claim
+		// the turn anyway: falling through to recall would answer a question
+		// about money with whatever note happens to be nearest.
+		if !isNoFactErr(err) {
+			log.Printf("voice: money fact: %v", err)
+		}
+		return "я не отслеживаю траты — не подключено.", true
+	}
+	val, err := zenmoney.ParseFactValue(fact.Value)
+	if err != nil {
+		log.Printf("voice: money fact: decode: %v", err)
+		return "не получилось прочитать траты.", true
+	}
+	reply := val.FormatRU(phrase)
+	if reply == "" {
+		return "по тратам пока нечего сказать.", true
+	}
+	// A stale fact is reported as stale rather than spoken as today's number.
+	if h.now().Sub(fact.Ts) > zenmoney.StaleAfter {
+		return "данные от " + fact.Ts.Local().Format("02.01") + ": " + reply, true
+	}
+	return reply, true
+}
+
+// isNoFactErr — ErrNoFact survives the wire wrapped, so unwrap for it.
+func isNoFactErr(err error) bool {
+	for e := err; e != nil; {
+		if e == ipc.ErrNoFact {
+			return true
+		}
+		u, ok := e.(interface{ Unwrap() error })
+		if !ok {
+			return false
+		}
+		e = u.Unwrap()
+	}
+	return false
+}

cmd/mavend/actions_money_test.go

@@ -0,0 +1,139 @@
+package main
+
+import (
+	"context"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/zenmoney"
+)
+
+// moneyAPI answers only LatestFactBySource; everything else is unimplemented,
+// which is the assertion that answering a money question costs no model call
+// and reaches no network.
+type moneyAPI struct {
+	ipc.UnimplementedCoreAPI
+
+	fact    ipc.Fact
+	err     error
+	gotKey  string
+	gotSrc  string
+	callCnt int
+}
+
+func (a *moneyAPI) LatestFactBySource(_ context.Context, key, source string) (ipc.Fact, error) {
+	a.gotKey, a.gotSrc = key, source
+	a.callCnt++
+	return a.fact, a.err
+}
+
+func moneyNow() time.Time { return time.Date(2026, 8, 15, 20, 0, 0, 0, time.UTC) }
+
+func moneyFact(ts time.Time, val string) ipc.Fact {
+	return ipc.Fact{Kind: "env", Key: zenmoney.KeySpentMonth, Value: val, Source: zenmoney.Source, Ts: ts}
+}
+
+func TestQueryMoneyAnswersFromTheFact(t *testing.T) {
+	api := &moneyAPI{fact: moneyFact(moneyNow(), `{"spent":[{"currency":"RUB","amount":1749.5}],"count":3}`)}
+	h := &reactiveHandler{api: api, now: moneyNow}
+	reply, ok := h.queryMoney(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "сколько я потратил в этом месяце?"},
+	})
+	if !ok {
+		t.Fatal("the money source must claim a money question")
+	}
+	if api.gotKey != zenmoney.KeySpentMonth || api.gotSrc != zenmoney.Source {
+		t.Errorf("read %q/%q, want the month key from the poller's source", api.gotKey, api.gotSrc)
+	}
+	if !strings.Contains(reply, "1749.5") {
+		t.Errorf("reply = %q, want the exact figure", reply)
+	}
+	if !strings.Contains(reply, "в этом месяце") {
+		t.Errorf("reply = %q, want the window named", reply)
+	}
+}
+
+func TestQueryMoneyPicksTodaysKey(t *testing.T) {
+	api := &moneyAPI{fact: moneyFact(moneyNow(), `{"spent":[{"currency":"RUB","amount":250}],"count":1}`)}
+	h := &reactiveHandler{api: api, now: moneyNow}
+	if _, ok := h.queryMoney(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "сколько я потратил сегодня?"},
+	}); !ok {
+		t.Fatal("expected the source to claim it")
+	}
+	if api.gotKey != zenmoney.KeySpentToday {
+		t.Errorf("key = %q, want today's", api.gotKey)
+	}
+}
+
+// The capability is off unless configured, and then there is no fact. She says
+// so instead of letting the recall pass answer a money question from a note.
+func TestQueryMoneySaysNotConnected(t *testing.T) {
+	h := &reactiveHandler{api: &moneyAPI{err: ipc.ErrNoFact}, now: moneyNow}
+	reply, ok := h.queryMoney(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "сколько я потратил?"},
+	})
+	if !ok {
+		t.Fatal("expected the source to claim it")
+	}
+	if !strings.Contains(reply, "не подключено") {
+		t.Errorf("reply = %q, want an honest 'not connected'", reply)
+	}
+	// No number of any kind in that answer.
+	for _, d := range []string{"0", "1", "2", "3", "4", "5", "6", "7", "8", "9"} {
+		if strings.Contains(reply, d) {
+			t.Errorf("reply %q contains a digit — nothing was read, so there is no figure", reply)
+		}
+	}
+}
+
+// A fact older than the staleness bound is dated rather than spoken as if it
+// were current: the poller can be down, and last week's total presented as
+// today's is a lie by omission.
+func TestQueryMoneyDatesAStaleFact(t *testing.T) {
+	old := moneyNow().Add(-72 * time.Hour)
+	api := &moneyAPI{fact: moneyFact(old, `{"spent":[{"currency":"RUB","amount":100}],"count":1}`)}
+	h := &reactiveHandler{api: api, now: moneyNow}
+	reply, _ := h.queryMoney(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "сколько я потратил?"},
+	})
+	if !strings.Contains(reply, "данные от") {
+		t.Errorf("reply = %q, want the stale fact dated", reply)
+	}
+}
+
+func TestQueryMoneyPassesOtherQuestions(t *testing.T) {
+	api := &moneyAPI{}
+	h := &reactiveHandler{api: api, now: moneyNow}
+	for _, u := range []string{"какая погода?", "я потратил весь день на это", "какие у меня задачи?"} {
+		if _, ok := h.queryMoney(context.Background(), &queryTurn{dec: router.Decision{Utterance: u}}); ok {
+			t.Errorf("the money source claimed %q", u)
+		}
+	}
+	if api.callCnt != 0 {
+		t.Error("a non-money question must not read the money facts")
+	}
+}
+
+// Money must be answered before the recall sources, or a question about
+// spending gets answered by the nearest note.
+func TestQuerySourcesOrderMoneyBeforeRecall(t *testing.T) {
+	moneyAt, notesAt := -1, -1
+	for i, src := range querySources {
+		switch src.name {
+		case "money":
+			moneyAt = i
+		case "notes":
+			notesAt = i
+		}
+	}
+	if moneyAt < 0 || notesAt < 0 {
+		t.Fatalf("sources missing: money=%d notes=%d", moneyAt, notesAt)
+	}
+	if moneyAt > notesAt {
+		t.Errorf("money source at %d, after notes at %d", moneyAt, notesAt)
+	}
+}

cmd/mavend/actions_note.go

@@ -0,0 +1,47 @@
+package main
+
+import (
+	"context"
+	"log"
+	"strconv"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// actionNote handles router.IntentNote: embed the note, persist it, and
+// index it for recall.
+func (h *reactiveHandler) actionNote(ctx context.Context, dec router.Decision) string {
+	// An utterance that explicitly files a task is work, not recall, and
+	// belongs in the task store (Vikunja #130). Checked before the embedding
+	// is paid for. Everything else is a note, exactly as before.
+	if reply, ok := h.captureTaskFromNote(ctx, dec); ok {
+		return reply
+	}
+	// embed the note text with the same model the classifier uses, persist
+	// via CoreAPI (source=tap:voice). Semantic recall lives in `notes`, not
+	// facts — no predicate reads it (spec's two-memory split).
+	vec, err := router.EmbedPassage(ctx, h.embedder, dec.Utterance)
+	if err != nil {
+		log.Printf("voice: embed note: %v", err)
+		return "не получилось сохранить заметку."
+	}
+	noteTs := h.now()
+	noteID, err := h.api.WriteNote(ctx, noteTs, dec.Utterance, vec, "tap:voice")
+	if err != nil {
+		log.Printf("voice: write note: %v", err)
+		return "не получилось сохранить заметку."
+	}
+	// Insert into long-term memory (best-effort, must not fail the note write).
+	// text/ts in the meta make a Search hit self-describing (see bestRecall).
+	if h.memStore != nil {
+		if err := h.memStore.Insert(ctx, "note:"+strconv.FormatInt(noteID, 10), vec, map[string]string{
+			"source": "voice",
+			"type":   "note",
+			"text":   dec.Utterance,
+			"ts":     strconv.FormatInt(noteTs.Unix(), 10),
+		}); err != nil {
+			log.Printf("voice: memory insert: %v", err)
+		}
+	}
+	return "" // replier phrases the "saved" reply
+}

cmd/mavend/actions_query.go

@@ -0,0 +1,383 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"log"
+	"strings"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/memory"
+	"github.com/kami/maven/internal/morning"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/rss"
+	"github.com/kami/maven/internal/weather"
+)
+
+// queryTurn is the per-turn scratch a chain of query sources shares: the
+// decision being answered plus the work an earlier source already paid for
+// (the query embedding, the notes it pulled). Sources read and fill it in
+// order, so a later source never re-embeds.
+type queryTurn struct {
+	dec   router.Decision
+	vec   []float32
+	notes []ipc.Note
+}
+
+// querySource — one answer source in the chain actionQuery walks. answer
+// returns (reply, true) when this source claims the question, ("", false)
+// when it passes to the next one. name is for reading the table, not logged.
+//
+// A struct of one func rather than an interface: every source is a plain
+// method on *reactiveHandler with no state of its own (what state a turn has
+// lives in queryTurn), so an interface would mean one empty type per source
+// to satisfy it — ceremony for nothing. Same reasoning as confirmResolver in
+// confirm.go, and the table then reads like actionHandlers: a flat list of
+// method expressions you extend with one line.
+type querySource struct {
+	name   string
+	answer func(*reactiveHandler, context.Context, *queryTurn) (string, bool)
+}
+
+// querySources is the ordered chain actionQuery walks; first source to claim
+// answers the turn. THE ORDER IS LOAD-BEARING — see the memory-before-notes
+// comment on queryMemory: running the notes-only pass first was #373, and the
+// gate was never the bug. Adding a source (Kiwix, RSS, crawler, email) is one
+// line here plus its method; where you put the line is the whole decision.
+var querySources = []querySource{
+	{"fact-by-key", (*reactiveHandler).queryFactByKey},
+	// Before "calendar" on purpose: both match "…на сегодня", and the plan is
+	// the more specific ask (its matcher requires a plan word), so the calendar
+	// listing would otherwise swallow it.
+	{"day-plan", (*reactiveHandler).queryDayPlan},
+	// Also before "calendar": "что я обычно делаю по средам?" names a weekday,
+	// and the habit question is the more specific one. Its matcher requires a
+	// habit marker ("обычно", "каждый", …), so a question about this coming
+	// Wednesday still reaches the calendar.
+	{"habits", (*reactiveHandler).queryHabits},
+	// Before "calendar" and before the recall sources: "что мне нужно
+	// сделать?" is a question about the task list, and the notes pass would
+	// otherwise answer it with whatever note happens to be nearest. Its
+	// matcher requires a task noun or an explicit "что … сделать", so a
+	// date-bearing question still reaches the calendar.
+	{"tasks", (*reactiveHandler).queryTasks},
+	// Before the recall sources too: "сколько я потратил?" is a question about
+	// the money facts the poller wrote, and the notes pass would otherwise
+	// answer it from whatever he once said about spending. Its matcher needs a
+	// money noun plus an actual ask, so "я потратил весь день" is untouched.
+	{"money", (*reactiveHandler).queryMoney},
+	// Before the recall sources and before general knowledge: "что нового?" is
+	// a question about the feeds she reads, and general knowledge would answer
+	// it by inventing news. Its matcher needs a feed noun plus an ask, so
+	// "у меня новая лента в инстаграме" is untouched.
+	{"feeds", (*reactiveHandler).queryFeeds},
+	{"calendar", (*reactiveHandler).queryCalendar},
+	{"weather", (*reactiveHandler).queryWeather},
+	{"embed", (*reactiveHandler).queryEmbed},
+	{"memory", (*reactiveHandler).queryMemory},
+	{"notes", (*reactiveHandler).queryNotes},
+	{"general-knowledge", (*reactiveHandler).queryGeneral},
+}
+
+func (h *reactiveHandler) actionQuery(ctx context.Context, dec router.Decision) string {
+	t := &queryTurn{dec: dec}
+	for _, src := range querySources {
+		if reply, ok := src.answer(h, ctx, t); ok {
+			return reply
+		}
+	}
+	return "не знаю."
+}
+
+// queryFactByKey — when the dialogue layer resolved an anaphoric reference to
+// a prior fact's key (e.g. "когда я это сделал?" after "запиши что я пил
+// воду"), look up the fact's value directly.
+func (h *reactiveHandler) queryFactByKey(ctx context.Context, t *queryTurn) (string, bool) {
+	dec := t.dec
+	if !dec.Slots.HasKey || dec.Slots.Key == "" {
+		return "", false
+	}
+	f, err := h.api.LatestFact(ctx, dec.Slots.Key)
+	if err != nil {
+		return "", false
+	}
+	if dec.Slots.HasTime {
+		// The query asks about timing — the fact's own timestamp is the
+		// answer it's looking for. Format as a natural reply.
+		return fmt.Sprintf("я записала это %s", formatTime(f.Ts)), true
+	}
+	// General fact reference: describe what we know.
+	if dec.Utterance == "" {
+		return fmt.Sprintf("вот что я знаю: %s — %s", dec.Slots.Key, f.Value), true
+	}
+	// The utterance still carries the question; fall through to normal RAG
+	// with the resolved key in context.
+	return "", false
+}
+
+// queryDayPlan — "какие планы на сегодня?", "что у меня по плану?", "что
+// дальше?" (Vikunja #128). Recites the day: calendar events, pending
+// reminders, and any morning checklist still outstanding.
+//
+// Read-only by construction — the plan is assembled and rendered core-side and
+// nothing here schedules or announces. "что дальше?" asks for the rest of the
+// day, so that phrasing trims what has already passed.
+func (h *reactiveHandler) queryDayPlan(ctx context.Context, t *queryTurn) (string, bool) {
+	if !router.IsDayPlanQuery(t.dec.Utterance) {
+		return "", false
+	}
+	plan, err := h.api.DayPlan(ctx)
+	if err != nil {
+		log.Printf("voice: day plan: %v", err)
+		return "не получилось собрать план.", true
+	}
+	if !isRestOfDayQuery(t.dec.Utterance) {
+		return plan.Spoken, true
+	}
+	// Rebuild the pure plan so the rest-of-day rendering is the same code that
+	// rendered the whole day — one formatter, one persona.
+	p := morning.Plan{Date: plan.Date}
+	for _, it := range plan.Items {
+		p.Items = append(p.Items, morning.PlanEntry{
+			At:        it.At,
+			Text:      it.Text,
+			Kind:      morning.PlanKind(it.Kind),
+			Uncertain: it.Uncertain,
+		})
+	}
+	return p.After(h.now()).FormatRU(), true
+}
+
+// isRestOfDayQuery — "что дальше?" and its English form, the only plan phrasing
+// that means "from now on" rather than "the whole day".
+func isRestOfDayQuery(text string) bool {
+	s := strings.ToLower(text)
+	return strings.Contains(s, "дальше") || strings.Contains(s, "next")
+}
+
+// habitFactWindow — how many recent facts the behaviour profile is counted
+// over. Enough for a season of habits without scanning the whole store on every
+// question; the profile is recomputed on read, so the bound is the cost control.
+const habitFactWindow = 2000
+
+// queryHabits — "что я обычно делаю по вторникам?" (Vikunja #254). Counts the
+// answer out of the fact log rather than asking the model to summarise a life:
+// see internal/memory/behavior.go for why nothing here is generated.
+func (h *reactiveHandler) queryHabits(ctx context.Context, t *queryTurn) (string, bool) {
+	q, ok := router.ParseHabitQuery(t.dec.Utterance)
+	if !ok {
+		return "", false
+	}
+	facts, err := h.api.RecentFacts(ctx, habitFactWindow)
+	if err != nil {
+		log.Printf("voice: habits: recent facts: %v", err)
+		return "не получилось посмотреть записи.", true
+	}
+	obs := make([]memory.Observation, 0, len(facts))
+	for _, f := range facts {
+		obs = append(obs, memory.Observation{At: f.Ts, Key: f.Key, Kind: f.Kind})
+	}
+	profile := memory.BuildProfile(obs, h.now())
+	if q.HasWeekday {
+		return profile.FormatWeekdayRU(q.Weekday), true
+	}
+	return profile.FormatOverallRU(), true
+}
+
+// feedNoteWindow — how many recent notes are scanned for feed items, and
+// feedReadOut — how many headlines she actually reads back. She summarises the
+// top of the pile, she does not recite a river.
+const (
+	feedNoteWindow = 200
+	feedReadOut    = 3
+)
+
+// queryFeeds — "что нового в лентах?", "что нового по технологиям?"
+// (Vikunja #258).
+//
+// This is the ONLY way a feed item reaches him. The poller writes notes and
+// never speaks; asking is the trigger. If that ever changes, the thing that
+// changed is "Maven is not a nag", not a detail of this file.
+func (h *reactiveHandler) queryFeeds(ctx context.Context, t *queryTurn) (string, bool) {
+	q, ok := router.ParseFeedQuery(t.dec.Utterance)
+	if !ok {
+		return "", false
+	}
+	if !h.feedsOn {
+		// Claim the turn rather than fall through: "не читаю ленты" is true, and
+		// letting general knowledge answer "что нового?" would be an invented
+		// news bulletin.
+		return "я пока не читаю ленты — они не настроены.", true
+	}
+	notes, err := h.api.RecentNotes(ctx, feedNoteWindow)
+	if err != nil {
+		log.Printf("voice: feeds: recent notes: %v", err)
+		return "не получилось посмотреть ленты.", true
+	}
+	var picked []string
+	for _, n := range notes {
+		if !strings.HasPrefix(n.Source, rss.SourcePrefix) {
+			continue
+		}
+		if !router.CategoryMatches(n.Text, q.Category) {
+			continue
+		}
+		// The note carries title, summary and link; she reads the title.
+		title := n.Text
+		if i := strings.IndexByte(title, '\n'); i > 0 {
+			title = title[:i]
+		}
+		picked = append(picked, strings.TrimSpace(title))
+		if len(picked) == feedReadOut {
+			break
+		}
+	}
+	if len(picked) == 0 {
+		if q.Category != "" {
+			return "по этой теме в лентах пока ничего.", true
+		}
+		return "в лентах пока ничего нового.", true
+	}
+	return "вот что нового: " + strings.Join(picked, "; "), true
+}
+
+// queryCalendar — "что у меня сегодня?", "планы на завтра?"
+// h.now(), not time.Now(): the handler's clock is the injected one, so this
+// source can be tested at a fixed time like the rest.
+func (h *reactiveHandler) queryCalendar(ctx context.Context, t *queryTurn) (string, bool) {
+	date, ok := router.ParseCalendarDate(t.dec.Utterance, h.now())
+	if !ok {
+		return "", false
+	}
+	events, err := h.api.CalendarEvents(ctx, date, date.Add(24*time.Hour))
+	if err != nil {
+		log.Printf("voice: calendar events: %v", err)
+		return "не получилось проверить календарь.", true
+	}
+	// Provenance travels with each event. A work meeting relayed off a phone
+	// notification (source ambient:notif, #126) is stored below full confidence
+	// and gets hedged; a CalDAV read is recited plainly.
+	entries := make([]router.CalendarEntry, len(events))
+	for i, e := range events {
+		entries[i] = router.CalendarEntry{Text: e.Value, Uncertain: e.Confidence < 1.0}
+	}
+	var f router.CalendarEventFormatter
+	return f.FormatEntries(entries, date), true
+}
+
+func (h *reactiveHandler) queryWeather(ctx context.Context, t *queryTurn) (string, bool) {
+	if !isWeatherQuery(t.dec.Utterance) {
+		return "", false
+	}
+	loc := extractWeatherLocation(t.dec.Utterance, h.weatherLocation)
+	ctxWT, cancel := context.WithTimeout(ctx, 5*time.Second)
+	defer cancel()
+	w, err := h.weatherProvider.CurrentWeather(ctxWT, loc)
+	if errors.Is(err, weather.ErrNotConfigured) {
+		return "погода не настроена.", true
+	}
+	if err != nil {
+		log.Printf("voice: weather: %v", err)
+		return "не получилось узнать погоду.", true
+	}
+	return fmt.Sprintf("в %s сейчас %.0f градусов, %s.", w.Location, w.Temperature, w.Condition), true
+}
+
+// queryEmbed isn't an answer source — it's the shared cost the two recall
+// sources below both need, run once, in the position it always ran in. It
+// only claims the turn when the embedder fails.
+func (h *reactiveHandler) queryEmbed(ctx context.Context, t *queryTurn) (string, bool) {
+	vec, err := router.EmbedQuery(ctx, h.embedder, t.dec.Utterance)
+	if err != nil {
+		log.Printf("voice: embed query: %v", err)
+		return "не получилось найти ответ.", true
+	}
+	t.vec = vec
+	return "", false
+}
+
+// queryMemory — long-term memory first: ONE search over everything Maven
+// remembers (notes and facts share this index) and ONE confidence gate, so
+// the memory that is clearly the best match answers — a note just as much as
+// a fact.
+//
+// This used to run only after the notes-only source below had already
+// rejected the same note at the same score, which no note could ever survive
+// a second time: the branch could only return a fact (#373). Order, not the
+// gate, was the bug — the set of questions Maven answers is unchanged, only
+// which memory gets to answer them.
+func (h *reactiveHandler) queryMemory(ctx context.Context, t *queryTurn) (string, bool) {
+	if h.memStore == nil {
+		return "", false
+	}
+	hits, herr := h.memStore.Search(ctx, t.vec, 3)
+	if herr != nil {
+		log.Printf("voice: memory search: %v", herr)
+		return "", false
+	}
+	hit, ok := bestRecall(hits, h.queryMinScore, h.queryMinMargin)
+	if !ok {
+		return "", false
+	}
+	text := hit.Meta["text"]
+	// A note is phrased in Maven's voice; a fact is read back as it was
+	// stored.
+	if hit.Meta["type"] == "note" {
+		if reply, perr := h.phraser.PhraseQuery(ctx, t.dec.Utterance, []string{text}); perr == nil && reply != "" {
+			return reply, true
+		}
+	}
+	return text, true
+}
+
+// queryNotes — notes-only pass, for notes the vector index above does not
+// hold (an older note written before it existed). Same gate, notes-only
+// candidates.
+//
+// Confidence gate: below it, say "I don't know" rather than read back the
+// least-unrelated note — a confident wrong recall is worse than a gap (spec's
+// "not a guesser-of-truth"). Same instinct as the loop's since(key)==null →
+// don't fire. Two parts: an absolute cosine floor, and a margin over the
+// runner-up, which is the part that works with the e5 embedder's narrow score
+// band. See memory.Confident. Failing the gate passes the turn on to general
+// knowledge, which is what "don't read back the runner-up" means here.
+func (h *reactiveHandler) queryNotes(ctx context.Context, t *queryTurn) (string, bool) {
+	notes, err := h.api.QueryNotes(ctx, t.vec, 5)
+	if err != nil {
+		log.Printf("voice: query notes: %v", err)
+		return "не получилось найти ответ.", true
+	}
+	t.notes = notes
+	noteScores := make([]float64, len(notes))
+	for i, n := range notes {
+		noteScores[i] = n.Score
+	}
+	if !memory.ConfidentScores(noteScores, h.queryMinScore, h.queryMinMargin) {
+		return "", false
+	}
+	texts := make([]string, len(notes))
+	for i, n := range notes {
+		texts[i] = n.Text
+	}
+	reply, err := h.phraser.PhraseQuery(ctx, t.dec.Utterance, texts)
+	if err != nil {
+		log.Printf("voice: phrase query: %v", err)
+	}
+	if reply == "" {
+		reply = "вот что я нашла: " + texts[0]
+	}
+	return reply, true
+}
+
+// queryGeneral — general knowledge from the phraser, the last source before
+// giving up. It always claims: either the model answers or Maven says she
+// doesn't know.
+func (h *reactiveHandler) queryGeneral(ctx context.Context, t *queryTurn) (string, bool) {
+	reply, err := h.phraser.PhraseQuery(ctx, t.dec.Utterance, nil)
+	if err != nil || reply == "" {
+		return "не знаю.", true
+	}
+	return reply, true
+}

cmd/mavend/actions_reminder.go

@@ -0,0 +1,33 @@
+package main
+
+import (
+	"context"
+	"log"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// actionReminder handles router.IntentReminder: parse the time when stage-0
+// skipped the extractor, then create the reminder.
+func (h *reactiveHandler) actionReminder(ctx context.Context, dec router.Decision) string {
+	if !dec.Slots.HasTime {
+		// Stage-0 (reminder-wakeword grammar) skips the extractor, so the
+		// time wasn't parsed. Run the parser as a fallback.
+		if dec.Stage == 0 && h.timeParser != nil {
+			t, ok, err := h.timeParser.Parse(ctx, dec.Utterance, h.now())
+			if err == nil && ok {
+				dec.Slots.Time = t
+				dec.Slots.HasTime = true
+			}
+		}
+		if !dec.Slots.HasTime {
+			return "не получилось разобрать время напоминания."
+		}
+	}
+	payload := `{"text":` + jsonString(dec.Utterance) + `}`
+	if _, err := h.api.CreateReminder(ctx, dec.Slots.Time, payload, ""); err != nil {
+		log.Printf("voice: create reminder: %v", err)
+		return "не получилось поставить напоминание."
+	}
+	return ""
+}

cmd/mavend/actions_task.go

@@ -0,0 +1,83 @@
+package main
+
+import (
+	"context"
+	"log"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/store"
+	"github.com/kami/maven/internal/tasks"
+)
+
+// Task capture on the voice/chat path (Vikunja #130).
+//
+// Two halves, both deliberately small:
+//
+//   - captureTaskFromNote runs at the top of actionNote. An utterance that
+//     explicitly files a task ("добавь в задачи купить молоко") goes to the task
+//     store instead of the note store. Anything without an explicit marker is
+//     still a note — see router.ParseTaskCapture for why "надо бы поспать" must
+//     not become a task.
+//   - queryTasks is a query source that reads the list back.
+//
+// Nothing here speaks unprompted. Tasks are answered when asked about; no tick
+// rule reads the table.
+
+// captureTaskFromNote claims the turn when the utterance explicitly files a
+// task, returning the reply. ("", false) hands the turn back to the note path.
+func (h *reactiveHandler) captureTaskFromNote(ctx context.Context, dec router.Decision) (string, bool) {
+	cap, ok := router.ParseTaskCapture(dec.Utterance)
+	if !ok {
+		return "", false
+	}
+	resp, err := h.api.CaptureTask(ctx, ipc.CaptureTaskReq{
+		Text:   cap.Text,
+		Source: "tap:voice",
+		Status: store.TaskOpen, // he stated it himself — not a candidate
+		Weight: cap.Weight,     // 0 unless he said "срочно" / "важно"
+		Ts:     h.now(),
+	})
+	if err != nil {
+		log.Printf("voice: capture task: %v", err)
+		return "не получилось записать задачу.", true
+	}
+	if !resp.Created {
+		return "это уже в списке.", true
+	}
+	return "записала: " + cap.Text, true
+}
+
+// queryTasks — "какие у меня задачи?", "что мне нужно сделать?".
+//
+// Reads the live set and recites it in priority order (Vikunja #129). The order
+// is computed by internal/tasks from what he told her — deadlines, the urgency
+// he stated, how long a task has been sitting — never asked of the model. The
+// rendering is the package's too, so the spoken list and the /tasks page can
+// never disagree about what comes first.
+func (h *reactiveHandler) queryTasks(ctx context.Context, t *queryTurn) (string, bool) {
+	if !router.IsTaskListQuery(t.dec.Utterance) {
+		return "", false
+	}
+	live, err := h.api.ListTasks(ctx, "live")
+	if err != nil {
+		log.Printf("voice: list tasks: %v", err)
+		return "не получилось посмотреть задачи.", true
+	}
+	return tasks.FormatRU(tasks.Rank(taskItems(live), h.now())), true
+}
+
+// taskItems maps wire rows onto the ranker's input. Written here rather than in
+// internal/tasks so the ranker stays a pure package with no ipc (and therefore
+// no store, and therefore no cgo) dependency — the same posture as
+// internal/morning and internal/memory.
+func taskItems(ts []ipc.Task) []tasks.Item {
+	out := make([]tasks.Item, len(ts))
+	for i, t := range ts {
+		out[i] = tasks.Item{
+			ID: t.ID, Text: t.Text, Status: t.Status,
+			Created: t.CreatedTs, Due: t.Due, Weight: t.Weight,
+		}
+	}
+	return out
+}

cmd/mavend/actions_task_test.go

@@ -0,0 +1,227 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+)
+
+// taskAPI answers only the three task methods; every other call is
+// unimplemented, which is the assertion that capture needs nothing else — in
+// particular no embedder, so a filed task costs no model call.
+type taskAPI struct {
+	ipc.UnimplementedCoreAPI
+
+	captured []ipc.CaptureTaskReq
+	created  bool
+	capErr   error
+
+	tasks   []ipc.Task
+	listArg string
+	listErr error
+}
+
+func (a *taskAPI) CaptureTask(_ context.Context, req ipc.CaptureTaskReq) (ipc.CaptureTaskResp, error) {
+	a.captured = append(a.captured, req)
+	if a.capErr != nil {
+		return ipc.CaptureTaskResp{}, a.capErr
+	}
+	return ipc.CaptureTaskResp{ID: 1, Created: a.created}, nil
+}
+
+func (a *taskAPI) ListTasks(_ context.Context, status string) ([]ipc.Task, error) {
+	a.listArg = status
+	return a.tasks, a.listErr
+}
+
+func taskNow() time.Time { return time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC) }
+
+func taskHandler(api ipc.CoreAPI) *reactiveHandler {
+	return &reactiveHandler{api: api, now: taskNow}
+}
+
+func TestCaptureTaskFromNoteFilesTheTask(t *testing.T) {
+	api := &taskAPI{created: true}
+	h := taskHandler(api)
+	reply, ok := h.captureTaskFromNote(context.Background(), router.Decision{
+		Intent: router.IntentNote, Utterance: "добавь в задачи купить молоко",
+	})
+	if !ok {
+		t.Fatal("an explicit capture must claim the turn")
+	}
+	if len(api.captured) != 1 {
+		t.Fatalf("captured %d, want 1", len(api.captured))
+	}
+	got := api.captured[0]
+	if got.Text != "купить молоко" {
+		t.Errorf("text = %q, want the marker stripped", got.Text)
+	}
+	if got.Source != "tap:voice" {
+		t.Errorf("source = %q, want tap:voice", got.Source)
+	}
+	if got.Status != "open" {
+		t.Errorf("status = %q — work he stated is open, never a candidate", got.Status)
+	}
+	if !got.Ts.Equal(taskNow()) {
+		t.Errorf("ts = %v, want the handler clock", got.Ts)
+	}
+	if !strings.Contains(reply, "купить молоко") {
+		t.Errorf("reply = %q, want it to read the task back", reply)
+	}
+}
+
+// A note is still a note: capture only fires on an explicit marker, so
+// ordinary recall is untouched.
+func TestCaptureTaskFromNotePassesOrdinaryNotes(t *testing.T) {
+	api := &taskAPI{}
+	h := taskHandler(api)
+	for _, u := range []string{"надо бы поспать", "мне понравился этот фильм", "запиши что я пил воду"} {
+		if _, ok := h.captureTaskFromNote(context.Background(), router.Decision{Utterance: u}); ok {
+			t.Errorf("%q was captured as a task", u)
+		}
+	}
+	if len(api.captured) != 0 {
+		t.Errorf("captured %d requests, want none", len(api.captured))
+	}
+}
+
+func TestCaptureTaskFromNoteSaysAlreadyOnTheList(t *testing.T) {
+	h := taskHandler(&taskAPI{created: false})
+	reply, ok := h.captureTaskFromNote(context.Background(), router.Decision{Utterance: "добавь в задачи купить молоко"})
+	if !ok {
+		t.Fatal("expected the capture path to claim it")
+	}
+	if !strings.Contains(reply, "уже") {
+		t.Errorf("reply = %q — a deduped capture must not claim it saved something new", reply)
+	}
+}
+
+func TestCaptureTaskFromNoteReportsStoreFailure(t *testing.T) {
+	h := taskHandler(&taskAPI{capErr: errors.New("db is on fire")})
+	reply, ok := h.captureTaskFromNote(context.Background(), router.Decision{Utterance: "добавь задачу починить кран"})
+	if !ok {
+		t.Fatal("a failed capture still claims the turn — the note path must not double-write")
+	}
+	if !strings.Contains(reply, "не получилось") {
+		t.Errorf("reply = %q, want an honest failure", reply)
+	}
+}
+
+func TestQueryTasksRecitesTheLiveList(t *testing.T) {
+	api := &taskAPI{tasks: []ipc.Task{
+		{ID: 1, Text: "купить молоко", Status: "open"},
+		{ID: 2, Text: "продлить страховку", Status: "candidate"},
+	}}
+	h := taskHandler(api)
+	reply, ok := h.queryTasks(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: "какие у меня задачи?"},
+	})
+	if !ok {
+		t.Fatal("the task source must claim a task-list question")
+	}
+	if api.listArg != "live" {
+		t.Errorf("ListTasks(%q), want \"live\" — a resolved task is not outstanding work", api.listArg)
+	}
+	if !strings.Contains(reply, "купить молоко") || !strings.Contains(reply, "продлить страховку") {
+		t.Errorf("reply = %q, want both tasks", reply)
+	}
+	// The candidate must be named as unconfirmed, not recited as his work.
+	openIdx := strings.Index(reply, "купить молоко")
+	candIdx := strings.Index(reply, "продлить страховку")
+	if !(openIdx < candIdx) {
+		t.Errorf("reply = %q, want confirmed work before candidates", reply)
+	}
+	if !strings.Contains(reply, "не подтвердил") {
+		t.Errorf("reply = %q, want the candidate flagged as unconfirmed", reply)
+	}
+}
+
+// The stated urgency rides through capture as a weight, so the ranker can use
+// it later (Vikunja #129). "срочно" is not part of the task text.
+func TestCaptureTaskCarriesStatedUrgency(t *testing.T) {
+	api := &taskAPI{created: true}
+	h := taskHandler(api)
+	if _, ok := h.captureTaskFromNote(context.Background(), router.Decision{
+		Utterance: "добавь в задачи срочно оплатить интернет",
+	}); !ok {
+		t.Fatal("expected a capture")
+	}
+	got := api.captured[0]
+	if got.Text != "оплатить интернет" {
+		t.Errorf("text = %q, want the urgency word out of the task", got.Text)
+	}
+	if got.Weight == 0 {
+		t.Error("weight = 0 — he said срочно and it was dropped")
+	}
+}
+
+// The recital is ordered by the ranker, not by insertion: a deadline he named
+// comes before undated work.
+func TestQueryTasksRecitesInPriorityOrder(t *testing.T) {
+	due := taskNow()
+	api := &taskAPI{tasks: []ipc.Task{
+		{ID: 1, Text: "купить молоко", Status: "open", CreatedTs: taskNow()},
+		{ID: 2, Text: "оплатить интернет", Status: "open", CreatedTs: taskNow(), Due: &due},
+	}}
+	h := taskHandler(api)
+	reply, _ := h.queryTasks(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "какие у меня задачи?"},
+	})
+	if strings.Index(reply, "оплатить интернет") > strings.Index(reply, "купить молоко") {
+		t.Errorf("reply = %q, want the dated task first", reply)
+	}
+	if !strings.Contains(reply, "сегодня") {
+		t.Errorf("reply = %q, want the reason named", reply)
+	}
+}
+
+func TestQueryTasksEmptyList(t *testing.T) {
+	h := taskHandler(&taskAPI{})
+	reply, ok := h.queryTasks(context.Background(), &queryTurn{
+		dec: router.Decision{Utterance: "что мне нужно сделать?"},
+	})
+	if !ok {
+		t.Fatal("expected the task source to claim it")
+	}
+	if reply != "задач нет." {
+		t.Errorf("reply = %q", reply)
+	}
+}
+
+func TestQueryTasksPassesOtherQuestions(t *testing.T) {
+	api := &taskAPI{}
+	h := taskHandler(api)
+	for _, u := range []string{"как дела?", "какая погода в москве?", "что у меня сегодня?"} {
+		if _, ok := h.queryTasks(context.Background(), &queryTurn{dec: router.Decision{Utterance: u}}); ok {
+			t.Errorf("the task source claimed %q", u)
+		}
+	}
+	if api.listArg != "" {
+		t.Error("a non-task question must not read the task list")
+	}
+}
+
+// The chain must reach the task source before the recall sources, or "что мне
+// нужно сделать?" gets answered by whatever note is nearest.
+func TestQuerySourcesOrderTasksBeforeRecall(t *testing.T) {
+	var tasksAt, notesAt = -1, -1
+	for i, src := range querySources {
+		switch src.name {
+		case "tasks":
+			tasksAt = i
+		case "notes":
+			notesAt = i
+		}
+	}
+	if tasksAt < 0 || notesAt < 0 {
+		t.Fatalf("sources missing: tasks=%d notes=%d", tasksAt, notesAt)
+	}
+	if tasksAt > notesAt {
+		t.Errorf("tasks source at %d, after notes at %d", tasksAt, notesAt)
+	}
+}

cmd/mavend/clarify.go

@@ -3,6 +3,8 @@ package main
 import (
 	"context"
 	"log"
+	"math/rand"
+	"strings"
 	"time"
 
 	"github.com/kami/maven/internal/dialogue"
@@ -21,8 +23,12 @@ const clarifyTTL = 90 * time.Second
 // raw utterance, chat and system have nothing to fill in. For those a clarify
 // decision keeps the canned "не поняла" reply — inventing a question for noise
 // is worse than admitting she missed it.
+// A reminder wants BOTH what to remind about and when. Subject first: "напомни
+// в 11" has a time and nothing to say at 11, and a reminder with no subject is
+// not worth setting. Order here is the order she asks in — she still only asks
+// about the first one missing.
 var wantedSlots = map[router.Intent][]dialogue.Slot{
-	router.IntentReminder: {dialogue.SlotTime},
+	router.IntentReminder: {dialogue.SlotText, dialogue.SlotTime},
 	router.IntentFact:     {dialogue.SlotKey},
 	router.IntentAct:      {dialogue.SlotFn},
 }
@@ -35,14 +41,94 @@ var wantedSlots = map[router.Intent][]dialogue.Slot{
 // questions, so there is no gender agreement to get wrong; the feminine
 // self-reference lives in the reply she gives when she drops the request.
 var clarifyQuestions = map[dialogue.Slot]string{
-	dialogue.SlotTime: "На когда напомнить?",
+	dialogue.SlotTime: "Когда?",
+	dialogue.SlotText: "О чём напомнить?",
 	dialogue.SlotKey:  "Что записать?",
 	dialogue.SlotFn:   "Что сделать?",
 }
 
-// clarifyDropped — she asked once, the answer still did not fill the gap, so
+// clarifyGaveUp — she is out of questions and still does not have the slot. She
-// the request is gone. Said plainly, once, with no second question.
+// says so out loud: dropping the request in silence would leave him thinking it
-const clarifyDropped = "Не разобрала — скажи целиком, пожалуйста."
+// landed. Feminine self-reference ("поняла"), as everywhere.
+const clarifyGaveUp = "Прости, я не поняла. Скажи, пожалуйста, по-другому."
+
+// clarifyExpiredVariants — his answer came after the TTL, so the parked request
+// is already gone. Same tone as clarifyGaveUp, different reason: too much time
+// passed, not "I did not understand". Feminine self-reference ("ждала",
+// "отпустила"); he is addressed with a plain imperative.
+//
+// Five phrasings, not one. This is the line he hears whenever he walks off
+// mid-request, so it is the line that repeats most — and the same sentence every
+// time is what makes a house assistant sound like a kiosk. They all carry the
+// same two facts (the old request is gone; say it again if it still matters),
+// because the wording may vary and the meaning may not.
+//
+// Fixed templates rather than model output, for the same reason as
+// clarifyQuestions: this text has to be right every time, and it is not worth a
+// generation to say something this small.
+var clarifyExpiredVariants = []string{
+	"Прости, я слишком долго ждала ответа и отпустила прошлую просьбу. Если она ещё нужна, скажи заново.",
+	"Кажется, прошлая просьба уже не важна — я её отпустила. Если я ошибаюсь, повтори.",
+	"Ты как-то резко замолчал, и я не стала ждать дальше. Если та просьба ещё нужна, скажи заново.",
+	"Я не дождалась ответа и убрала прошлую просьбу. Повтори, если она всё ещё нужна.",
+	"Столько времени прошло, что я отпустила прошлую просьбу. Скажи заново, если она в силе.",
+}
+
+// clarifyExpiredLine picks one of them at random.
+func clarifyExpiredLine() string {
+	return clarifyExpiredVariants[rand.Intn(len(clarifyExpiredVariants))]
+}
+
+// isClarifyExpired reports whether s opens with any of the expiry lines. The
+// notice is glued in front of this turn's reply (see withNotice), so a caller
+// checking for it has to match a prefix, not the whole string.
+func isClarifyExpired(s string) bool {
+	for _, v := range clarifyExpiredVariants {
+		if strings.HasPrefix(s, v) {
+			return true
+		}
+	}
+	return false
+}
+
+// trimClarifyExpired strips a leading expiry notice, leaving this turn's actual
+// reply. "" ⇒ the notice was the whole thing.
+func trimClarifyExpired(s string) string {
+	for _, v := range clarifyExpiredVariants {
+		if strings.HasPrefix(s, v) {
+			return strings.TrimSpace(strings.TrimPrefix(s, v))
+		}
+	}
+	return strings.TrimSpace(s)
+}
+
+// clarifyExpiredNotice returns that line when a parked question had just timed
+// out, and "" when nothing was parked. Call it right after
+// resolveClarifyAnswer: a live question is answered there, an expired one is
+// only reported here — the words themselves still go on to be routed fresh.
+func (h *reactiveHandler) clarifyExpiredNotice() string {
+	if h.clarifyStore == nil {
+		return ""
+	}
+	if !h.clarifyStore.TakeExpired(voiceDialogueID, h.now()) {
+		return ""
+	}
+	log.Printf("voice: clarify — parked question expired, telling him and routing the words fresh")
+	return clarifyExpiredLine()
+}
+
+// withNotice glues the expiry notice in front of this turn's reply. One turn
+// carries one reply on the wire, so the notice cannot be a message of its own —
+// but neither the notice nor the fresh answer may be dropped.
+func withNotice(notice, reply string) string {
+	if notice == "" {
+		return reply
+	}
+	if reply == "" {
+		return notice
+	}
+	return notice + " " + reply
+}
 
 // missingFor returns the slots a decision still needs, most important first.
 // Empty ⇒ there is nothing identifiable to ask about.
@@ -79,13 +165,14 @@ func (h *reactiveHandler) askClarify(dec router.Decision) (string, bool) {
 		return "", false
 	}
 	h.clarifyStore.Put(voiceDialogueID, &dialogue.PendingQuestion{
-		Intent:    dialogue.Intent(dec.Intent),
+		Intent:      dialogue.Intent(dec.Intent),
-		Slots:     toDialogueSlots(dec.Slots),
+		Slots:       toDialogueSlots(dec.Slots),
-		Missing:   []dialogue.Slot{slot},
+		Missing:     []dialogue.Slot{slot},
-		Utterance: dec.Utterance,
+		Utterance:   dec.Utterance,
-		Asked:     h.now(),
+		Asked:       h.now(),
-		TTL:       clarifyTTL,
+		TTL:         clarifyTTL,
-		Attempts:  1, // asked once; MaxAttempts is 1, so there is no second ask
+		Attempts:    1, // this ask
+		MaxAttempts: h.clarifyMaxAttempts,
 	})
 	log.Printf("voice: clarify — asked about %s for intent=%s", slot, dec.Intent)
 	return question, true
@@ -97,8 +184,9 @@ func (h *reactiveHandler) askClarify(dec router.Decision) (string, bool) {
 // resolveConfirm and checked in the same place.
 //
 // The answer is parsed with the same extractor the router uses, for the intent
-// she parked — no second parser. If it still does not fill the gap the request
+// she parked — no second parser. If it still does not fill the gap she asks
-// is dropped: she does not ask again.
+// again, up to MaxAttempts; after that she says out loud that she did not
+// understand. She never drops the request in silence.
 func (h *reactiveHandler) resolveClarifyAnswer(ctx context.Context, text string) (string, bool) {
 	if h.clarifyStore == nil {
 		return "", false
@@ -107,17 +195,14 @@ func (h *reactiveHandler) resolveClarifyAnswer(ctx context.Context, text string)
 	if q == nil {
 		return "", false
 	}
-	// One shot either way: the question is consumed whether or not the answer
-	// works, so a failed answer can't leave the question armed.
-	h.clarifyStore.Delete(voiceDialogueID)
 
 	intent := router.Intent(q.Intent)
 	answer := h.extractor.Extract(ctx, intent, text, h.now())
 	merged := q.Answer(text, toDialogueSlots(answer))
 	if len(dialogue.StillMissing(q.Missing, merged)) > 0 {
-		log.Printf("voice: clarify — answer %q did not fill %v, dropping", text, q.Missing)
+		return h.reaskOrGiveUp(q, merged, text), true
-		return clarifyDropped, true
 	}
+	h.clarifyStore.Delete(voiceDialogueID)
 
 	// Rebuild the decision as if it had routed cleanly, then run it down the
 	// normal path. Clarify is deliberately false and the intent is unchanged:
@@ -133,6 +218,29 @@ func (h *reactiveHandler) resolveClarifyAnswer(ctx context.Context, text string)
 	return h.finishClarified(ctx, dec), true
 }
 
+// reaskOrGiveUp handles an answer that left the gap open: ask the same question
+// again while she has attempts left, otherwise say she did not understand and
+// let the request go. Never returns "" — a mute give-up reads as "done".
+func (h *reactiveHandler) reaskOrGiveUp(q *dialogue.PendingQuestion, merged dialogue.Slots, text string) string {
+	question := ""
+	if len(q.Missing) > 0 {
+		question = clarifyQuestions[q.Missing[0]]
+	}
+	if question == "" || !q.CanAsk() {
+		h.clarifyStore.Delete(voiceDialogueID)
+		log.Printf("voice: clarify — gave up on %v after %d question(s), answer was %q", q.Missing, q.Attempts, text)
+		return clarifyGaveUp
+	}
+	// Re-park with whatever the answer DID give, the clock restarted and one
+	// more question spent.
+	q.Slots = merged
+	q.Attempts++
+	q.Asked = h.now()
+	h.clarifyStore.Put(voiceDialogueID, q)
+	log.Printf("voice: clarify — answer %q did not fill %v, asking again (attempt %d)", text, q.Missing, q.Attempts)
+	return question
+}
+
 // finishClarified runs a completed decision through the same steps a freshly
 // routed one takes: remember the turn, act, then phrase.
 func (h *reactiveHandler) finishClarified(ctx context.Context, dec router.Decision) string {
@@ -146,6 +254,10 @@ func (h *reactiveHandler) finishClarified(ctx context.Context, dec router.Decisi
 	if reply == "" {
 		reply = h.replier.Reply(dec)
 	}
+	if reply == "" {
+		// Belt: an empty reply here would be a silent drop.
+		reply = clarifyGaveUp
+	}
 	return reply
 }
 

cmd/mavend/clarify_test.go

@@ -56,10 +56,13 @@ func TestClarifyQuestionForMissingSlot(t *testing.T) {
 		want  string
 		asked bool
 	}{
-		{"reminder without a time", clarifyDec(router.IntentReminder, router.Slots{Text: "напомни позвонить маме"}, "напомни позвонить маме"), "На когда напомнить?", true},
+		{"reminder without a time", clarifyDec(router.IntentReminder, router.Slots{Text: "напомни позвонить маме"}, "напомни позвонить маме"), "Когда?", true},
 		{"fact without a key", clarifyDec(router.IntentFact, router.Slots{Text: "запиши"}, "запиши"), "Что записать?", true},
 		{"act without a fn", clarifyDec(router.IntentAct, router.Slots{Text: "сделай это"}, "сделай это"), "Что сделать?", true},
-		{"reminder that already has a time", clarifyDec(router.IntentReminder, router.Slots{HasTime: true}, "напомни в 11"), "", false},
+		// A time with nothing to say at that time is still half a reminder, so
+		// the subject is what she asks about — not silence.
+		{"reminder that has a time but no subject", clarifyDec(router.IntentReminder, router.Slots{HasTime: true}, "напомни в 11"), "О чём напомнить?", true},
+		{"reminder that has both", clarifyDec(router.IntentReminder, router.Slots{Text: "позвонить маме", HasTime: true}, "напомни в 11 позвонить маме"), "", false},
 		{"chat is never worth a question", clarifyDec(router.IntentChat, router.Slots{Text: "мгм"}, "мгм"), "", false},
 		{"query is never worth a question", clarifyDec(router.IntentQuery, router.Slots{Text: "а"}, "а"), "", false},
 	}
@@ -78,7 +81,7 @@ func TestClarifyReminderCompletesOnAnswer(t *testing.T) {
 	h, st, _ := newClarifyHandler(t)
 
 	question, asked := h.askClarify(clarifyDec(router.IntentReminder, router.Slots{Text: "напомни позвонить маме"}, "напомни позвонить маме"))
-	if !asked || question != "На когда напомнить?" {
+	if !asked || question != "Когда?" {
 		t.Fatalf("expected the time question, got %q asked=%v", question, asked)
 	}
 
@@ -86,7 +89,7 @@ func TestClarifyReminderCompletesOnAnswer(t *testing.T) {
 	if !handled {
 		t.Fatal("the answer to an open question must be consumed as an answer")
 	}
-	if reply == clarifyDropped {
+	if reply == clarifyGaveUp {
 		t.Fatalf("a good answer must not drop the request: %q", reply)
 	}
 
@@ -111,7 +114,7 @@ func TestClarifyFactCompletesOnAnswer(t *testing.T) {
 	if _, asked := h.askClarify(clarifyDec(router.IntentFact, router.Slots{Text: "запиши"}, "запиши")); !asked {
 		t.Fatal("a fact with no key should be asked about")
 	}
-	if reply, handled := h.resolveClarifyAnswer(ctx, "пил воду"); !handled || reply == clarifyDropped {
+	if reply, handled := h.resolveClarifyAnswer(ctx, "пил воду"); !handled || reply == clarifyGaveUp {
 		t.Fatalf("answer should complete the fact, handled=%v reply=%q", handled, reply)
 	}
 	if fact, err := st.LatestFact(ctx, "water"); err != nil || fact.Key != "water" {
@@ -137,26 +140,87 @@ func TestClarifyAnswerAfterTTLIsANewRequest(t *testing.T) {
 	}
 }
 
-// TestClarifyUnclearAnswerDropsWithoutAskingAgain — MaxAttempts is 1.
+// TestClarifyAsksThreeTimesThenSaysSo — three questions are allowed, the fourth
-func TestClarifyUnclearAnswerDropsWithoutAskingAgain(t *testing.T) {
+// is not, and running out is SPOKEN. Silence would read as "handled".
+func TestClarifyAsksThreeTimesThenSaysSo(t *testing.T) {
 	ctx := context.Background()
 	h, st, _ := newClarifyHandler(t)
 
 	if _, asked := h.askClarify(clarifyDec(router.IntentReminder, router.Slots{Text: "напомни"}, "напомни")); !asked {
-		t.Fatal("expected a question")
+		t.Fatal("expected a first question")
 	}
+	// Two more unclear answers ⇒ two more questions (3 asks in total).
+	for i := 2; i <= 3; i++ {
+		reply, handled := h.resolveClarifyAnswer(ctx, "ну не знаю")
+		if !handled {
+			t.Fatalf("answer %d must be consumed as an answer", i)
+		}
+		if reply != "Когда?" {
+			t.Fatalf("attempt %d should ask again, got %q", i, reply)
+		}
+		if h.clarifyStore.Get(voiceDialogueID, h.now()) == nil {
+			t.Fatalf("attempt %d must leave the question armed", i)
+		}
+	}
+
 	reply, handled := h.resolveClarifyAnswer(ctx, "ну не знаю")
-	if !handled || reply != clarifyDropped {
+	if !handled || reply != clarifyGaveUp {
-		t.Fatalf("an unclear answer should drop the request, handled=%v reply=%q", handled, reply)
+		t.Fatalf("the fourth try must give up out loud, handled=%v reply=%q", handled, reply)
 	}
-	if strings.Contains(reply, "?") {
+	if reply == "" || strings.Contains(reply, "?") {
-		t.Fatalf("she must not ask a second question: %q", reply)
+		t.Fatalf("giving up must be spoken and must not be another question: %q", reply)
 	}
 	if h.clarifyStore.Get(voiceDialogueID, h.now()) != nil {
-		t.Fatal("a dropped request must leave no armed question")
+		t.Fatal("a given-up request must leave no armed question")
 	}
 	if reminders, err := st.DueReminders(ctx, h.now().Add(48*time.Hour)); err != nil || len(reminders) != 0 {
-		t.Fatalf("a dropped request must not create anything: reminders=%v err=%v", reminders, err)
+		t.Fatalf("a given-up request must not create anything: reminders=%v err=%v", reminders, err)
+	}
+}
+
+// TestClarifyMaxAttemptsIsConfigurable — one question when the config says one.
+func TestClarifyMaxAttemptsIsConfigurable(t *testing.T) {
+	ctx := context.Background()
+	h, _, _ := newClarifyHandler(t)
+	h.clarifyMaxAttempts = 1
+
+	if _, asked := h.askClarify(clarifyDec(router.IntentReminder, router.Slots{Text: "напомни"}, "напомни")); !asked {
+		t.Fatal("expected a question")
+	}
+	if reply, handled := h.resolveClarifyAnswer(ctx, "ну не знаю"); !handled || reply != clarifyGaveUp {
+		t.Fatalf("with max 1 she must give up at once, handled=%v reply=%q", handled, reply)
+	}
+}
+
+// TestClarifyRestatedAnswerWins — «в 11:00», then «нет, в 15:00». The second
+// value is the one that lands.
+func TestClarifyRestatedAnswerWins(t *testing.T) {
+	ctx := context.Background()
+	h, st, _ := newClarifyHandler(t)
+
+	if _, asked := h.askClarify(clarifyDec(router.IntentReminder, router.Slots{Text: "напомни позвонить маме"}, "напомни позвонить маме")); !asked {
+		t.Fatal("expected a question")
+	}
+	// First answer parses, but re-park it by hand as if she had asked again:
+	// what matters here is that Answer prefers the newer value over the parked
+	// one, which is the case the daemon hits on a re-ask.
+	q := h.clarifyStore.Get(voiceDialogueID, h.now())
+	if q == nil {
+		t.Fatal("expected an armed question")
+	}
+	first := h.extractor.Extract(ctx, router.IntentReminder, "в 11:00", h.now())
+	q.Slots = q.Answer("в 11:00", toDialogueSlots(first))
+
+	if reply, handled := h.resolveClarifyAnswer(ctx, "нет, в 15:00"); !handled || reply == clarifyGaveUp {
+		t.Fatalf("the restated answer should complete the request, handled=%v reply=%q", handled, reply)
+	}
+	reminders, err := st.DueReminders(ctx, h.now().Add(48*time.Hour))
+	if err != nil || len(reminders) != 1 {
+		t.Fatalf("expected one reminder: %v err=%v", reminders, err)
+	}
+	want := h.extractor.Extract(ctx, router.IntentReminder, "в 15:00", h.now())
+	if !reminders[0].FireTs.Equal(want.Time) {
+		t.Fatalf("reminder at %v, want the restated %v", reminders[0].FireTs, want.Time)
 	}
 }
 
@@ -228,6 +292,36 @@ func TestNoQuestionWhenNothingIsMissing(t *testing.T) {
 	}
 }
 
+// TestClarifyExpiryIsAnnouncedAndWordsStillRoute — his answer lands after the
+// TTL: she must say the old request is gone AND still answer the new words.
+func TestClarifyExpiryIsAnnouncedAndWordsStillRoute(t *testing.T) {
+	ctx := context.Background()
+	h, _, now := newClarifyHandler(t)
+	emb := router.NewHashEmbedder(1024)
+	h.embedder = emb
+	h.router = buildRouter(emb, h.matcher, 0.55, nil)
+
+	if _, asked := h.askClarify(clarifyDec(router.IntentReminder, router.Slots{Text: "напомни"}, "напомни")); !asked {
+		t.Fatal("expected a question")
+	}
+	*now = now.Add(clarifyTTL + time.Second)
+
+	reply := h.handleText(ctx, "как дела")
+	if !isClarifyExpired(reply) {
+		t.Fatalf("expired question must be announced first, got %q", reply)
+	}
+	if trimClarifyExpired(reply) == "" {
+		t.Fatalf("the new words must still be answered, got only the notice: %q", reply)
+	}
+	if h.clarifyStore.Get(voiceDialogueID, h.now()) != nil {
+		t.Fatal("the expired question must be gone")
+	}
+	// The notice is said once, not on every later utterance.
+	if reply := h.handleText(ctx, "как дела"); isClarifyExpired(reply) {
+		t.Fatalf("notice repeated on a later turn: %q", reply)
+	}
+}
+
 // TestNoPendingQuestionFallsThrough — with nothing parked, an utterance routes
 // normally.
 func TestNoPendingQuestionFallsThrough(t *testing.T) {

cmd/mavend/confirm.go

@@ -0,0 +1,216 @@
+package main
+
+import (
+	"context"
+	"log"
+	"strings"
+	"time"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// pendingHexisExec — a mutating Hexis capability parked awaiting a spoken
+// confirm. The confirmation is bound to the resolved capability + canonical
+// target entity so a later "да" can only execute exactly what was proposed
+// (ecosystem invariant: protected actions require bound confirmation).
+type pendingHexisExec struct {
+	capabilityID string
+	capName      string
+	entityID     string
+	displayName  string
+	expiry       time.Time
+}
+
+// pendingRoutineConfirm — a proposed routine awaiting a spoken y/n to become
+// a recurring reminder. Set by detectPattern after creating a proposal.
+type pendingRoutineConfirm struct {
+	routineID int64
+	action    string
+	object    string
+	interval  float64
+	phrase    string
+	expiry    time.Time
+}
+
+// pendingAct — a destructive act awaiting a spoken confirm.
+type pendingAct struct {
+	fn     string
+	args   []string
+	phrase string
+	expiry time.Time
+}
+
+// confirmTTL — how long a parked destructive confirm stays answerable. Short:
+// a confirm is a same-breath gesture; a stale prompt shouldn't fire on an
+// unrelated later "да".
+const confirmTTL = 90 * time.Second
+
+// park stores a destructive act awaiting confirmation. Overwrites any prior
+// pending (last-asked wins — single-user box).
+func (h *reactiveHandler) park(fn string, args []string, phrase string) {
+	h.mu.Lock()
+	h.pending = &pendingAct{fn: fn, args: args, phrase: phrase, expiry: h.now().Add(confirmTTL)}
+	h.mu.Unlock()
+}
+
+// resolveConfirm interprets an utterance as the answer to a parked destructive
+// act OR a parked routine proposal. Returns (reply, true) when it consumed the
+// utterance as a y/n answer; ("", false) when there's nothing pending (or the
+// parked act expired), so the caller routes the utterance normally. An
+// unrecognised answer cancels the pending and routes normally — a confirm that
+// can't be answered clearly is safer abandoned than left armed.
+func (h *reactiveHandler) resolveConfirm(ctx context.Context, text string) (string, bool) {
+	h.mu.Lock()
+	defer h.mu.Unlock()
+
+	for _, r := range h.confirmResolvers(ctx) {
+		if !r.claim() {
+			continue
+		}
+		// The slot is already cleared by claim(): every branch below drops the
+		// pending, including the unclear one — a confirm that can't be
+		// answered clearly is safer abandoned than left armed.
+		switch classifyConfirm(text) {
+		case confirmYes:
+			return r.yes(), true
+		case confirmNo:
+			return r.no(), true
+		default:
+			return "", false
+		}
+	}
+	return "", false
+}
+
+// confirmResolver — one parked-confirm slot in the chain. claim() reports
+// whether this slot holds a live pending, taking it (and dropping an expired
+// one) as it goes; yes/no then run the answer. Only ever called with h.mu held.
+type confirmResolver struct {
+	claim func() bool
+	yes   func() string
+	no    func() string
+}
+
+// confirmResolvers builds the ordered chain resolveConfirm walks. Order is
+// deliberate: the routine proposal is checked before the tool confirm so a
+// routine confirm doesn't get eaten by a stale tool pending.
+func (h *reactiveHandler) confirmResolvers(ctx context.Context) []confirmResolver {
+	var pr *pendingRoutineConfirm
+	var hx *pendingHexisExec
+	var p *pendingAct
+
+	return []confirmResolver{
+		// Routine proposal.
+		{
+			claim: func() bool {
+				pr, h.pendingRoutine = h.pendingRoutine, nil
+				return pr != nil && !h.now().After(pr.expiry)
+			},
+			yes: func() string {
+				// Only record the acceptance. The tick loop reads accepted
+				// routines and nudges on their own interval. Building a
+				// reminder here made a routine fire exactly once (Vikunja #366).
+				if err := h.dataStore.AcceptProposedRoutine(ctx, pr.routineID, h.now()); err != nil {
+					log.Printf("voice: accept proposed routine: %v", err)
+					return "не получилось запомнить рутину."
+				}
+				return "буду напоминать."
+			},
+			no: func() string {
+				if err := h.dataStore.DismissProposedRoutine(ctx, pr.routineID); err != nil {
+					log.Printf("voice: dismiss proposed routine: %v", err)
+				}
+				return "хорошо, не буду."
+			},
+		},
+		// Hexis execution confirm. Bound to the exact capability + target that
+		// was proposed; a stray "да" can only run that, nothing else.
+		{
+			claim: func() bool {
+				hx, h.pendingHexis = h.pendingHexis, nil
+				return hx != nil && !h.now().After(hx.expiry)
+			},
+			yes: func() string {
+				return h.execHexis(ctx, hx.capabilityID, hx.capName, hx.entityID, hx.displayName)
+			},
+			no: func() string { return "отменила." },
+		},
+		// Tool confirm.
+		{
+			claim: func() bool {
+				p, h.pending = h.pending, nil
+				return p != nil && !h.now().After(p.expiry)
+			},
+			yes: func() string {
+				out, err := h.tools.Exec(ctx, p.fn, p.args, true) // confirmed
+				if err != nil {
+					log.Printf("voice: tool %s (confirmed): %v", p.fn, err)
+					if out != "" {
+						return "не получилось выполнить команду: " + firstLine(out)
+					}
+					return "не получилось выполнить команду."
+				}
+				if out != "" {
+					return "готово: " + firstLine(out)
+				}
+				return "готово."
+			},
+			no: func() string { return "отменила." },
+		},
+	}
+}
+
+// proposeGap scaffolds a 'proposed' tool for an act whose verb isn't enabled.
+// maven drafts the registration (name = the verb, provenance = the utterance);
+// a human enables it on the authed surface. She suggests, never enables.
+func (h *reactiveHandler) proposeGap(ctx context.Context, dec router.Decision) string {
+	name := firstWord(stripWake(dec.Utterance))
+	if name == "" {
+		return "не разобрала команду — попробуй иначе."
+	}
+	newly, err := h.api.ProposeTool(ctx, name, dec.Utterance, "", h.now())
+	if err != nil {
+		log.Printf("voice: propose tool %q: %v", name, err)
+		return "команды «" + name + "» нет в списке разрешённых."
+	}
+	if newly {
+		return "команды «" + name + "» нет в списке. Предложила её добавить — включи через клиент."
+	}
+	return "команды «" + name + "» пока нет в списке — она уже предложена, включи через клиент."
+}
+
+// confirmVerdict — the parse of a y/n confirm answer.
+type confirmVerdict int
+
+const (
+	confirmUnknown confirmVerdict = iota
+	confirmYes
+	confirmNo
+)
+
+// classifyConfirm reads a short ru/en yes-or-no answer. Substring match on the
+// stems so inflections/fillers ("да, давай", "нет, отмени") still land.
+func classifyConfirm(text string) confirmVerdict {
+	t := strings.ToLower(strings.TrimSpace(text))
+	// negatives first — "не надо" contains no "да", but check no-stems before
+	// yes so a leading "нет" isn't shadowed.
+	for _, no := range []string{"нет", "не надо", "отмен", "стоп", "no", "cancel", "stop", "don't"} {
+		if strings.Contains(t, no) {
+			return confirmNo
+		}
+	}
+	for _, yes := range []string{"да", "ага", "давай", "подтвер", "конечно", "yes", "yeah", "yep", "confirm", "ок", "okay", "ok"} {
+		if strings.Contains(t, yes) {
+			return confirmYes
+		}
+	}
+	return confirmUnknown
+}
+
+// actPhrase renders "fn arg1 arg2" for the confirm prompt.
+func actPhrase(fn string, args []string) string {
+	if len(args) == 0 {
+		return fn
+	}
+	return fn + " " + strings.Join(args, " ")
+}

cmd/mavend/dayplan_test.go

@@ -0,0 +1,227 @@
+package main
+
+import (
+	"context"
+	"errors"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+)
+
+// planAPI answers only DayPlan; every other call is unimplemented, which is
+// exactly the assertion that the plan source needs nothing else.
+type planAPI struct {
+	ipc.UnimplementedCoreAPI
+	plan  ipc.DayPlan
+	err   error
+	calls int
+}
+
+func (a *planAPI) DayPlan(context.Context) (ipc.DayPlan, error) {
+	a.calls++
+	if a.err != nil {
+		return ipc.DayPlan{}, a.err
+	}
+	return a.plan, nil
+}
+
+func planDay() time.Time { return time.Date(2026, 8, 3, 12, 0, 0, 0, time.UTC) }
+
+func samplePlan() ipc.DayPlan {
+	day := planDay()
+	mid := time.Date(2026, 8, 3, 0, 0, 0, 0, time.UTC)
+	return ipc.DayPlan{
+		Date: mid,
+		Items: []ipc.DayPlanItem{
+			{At: day.Add(-2 * time.Hour), Text: "Standup @ 10:00-10:30", Kind: "event"},
+			{At: day.Add(2 * time.Hour), Text: "Планёрка @ 14:00-14:30", Kind: "event", Uncertain: true},
+			{At: day.Add(6 * time.Hour), Text: "позвонить маме", Kind: "reminder"},
+		},
+		Spoken: "план на 03.08.2026: 10:00 — Standup @ 10:00-10:30; " +
+			"похоже, 14:00 — Планёрка @ 14:00-14:30; 18:00 — позвонить маме.",
+	}
+}
+
+func planHandler(api ipc.CoreAPI) *reactiveHandler {
+	return &reactiveHandler{api: api, now: planDay}
+}
+
+func TestQueryDayPlanRecitesTheDay(t *testing.T) {
+	api := &planAPI{plan: samplePlan()}
+	h := planHandler(api)
+	reply, ok := h.queryDayPlan(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: "какие планы на сегодня?"},
+	})
+	if !ok {
+		t.Fatal("the plan source must claim a plan question")
+	}
+	if reply != api.plan.Spoken {
+		t.Errorf("reply = %q, want the core's spoken plan %q", reply, api.plan.Spoken)
+	}
+}
+
+// "что дальше?" is the rest of the day, not the whole day: what has already
+// happened is not a plan.
+func TestQueryDayPlanTrimsToRestOfDay(t *testing.T) {
+	h := planHandler(&planAPI{plan: samplePlan()})
+	reply, ok := h.queryDayPlan(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: "что дальше?"},
+	})
+	if !ok {
+		t.Fatal("expected the plan source to claim it")
+	}
+	if strings.Contains(reply, "Standup") {
+		t.Errorf("a passed item must not be read back: %q", reply)
+	}
+	if !strings.Contains(reply, "Планёрка") || !strings.Contains(reply, "позвонить маме") {
+		t.Errorf("the rest of the day is missing: %q", reply)
+	}
+	// Provenance survives the trim.
+	if !strings.Contains(reply, "похоже,") {
+		t.Errorf("a relayed event must stay hedged: %q", reply)
+	}
+}
+
+// A question that is not about the plan must fall through, or the plan buries
+// the calendar listing and the weather behind it.
+func TestQueryDayPlanPassesOnEverythingElse(t *testing.T) {
+	for _, q := range []string{
+		"что у меня сегодня?",
+		"какие планы на завтра?",
+		"когда планёрка?",
+		"какая погода?",
+		"",
+	} {
+		api := &planAPI{plan: samplePlan()}
+		reply, ok := planHandler(api).queryDayPlan(context.Background(), &queryTurn{
+			dec: router.Decision{Intent: router.IntentQuery, Utterance: q},
+		})
+		if ok {
+			t.Errorf("%q was claimed by the plan source (reply %q)", q, reply)
+		}
+		if api.calls != 0 {
+			t.Errorf("%q hit the core for a plan it does not want", q)
+		}
+	}
+}
+
+func TestQueryDayPlanCoreFailure(t *testing.T) {
+	h := planHandler(&planAPI{err: errors.New("socket closed")})
+	reply, ok := h.queryDayPlan(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: "план на сегодня"},
+	})
+	if !ok {
+		t.Fatal("a failed plan read must still answer, not fall through to RAG")
+	}
+	if reply != "не получилось собрать план." {
+		t.Errorf("reply = %q", reply)
+	}
+}
+
+// The day plan must sit before the calendar listing: both match "…на сегодня",
+// and the more specific matcher has to get first refusal (see #373 for what
+// happens when the order is wrong).
+func TestDayPlanSourcePrecedesCalendar(t *testing.T) {
+	plan, cal := -1, -1
+	for i, s := range querySources {
+		switch s.name {
+		case "day-plan":
+			plan = i
+		case "calendar":
+			cal = i
+		}
+	}
+	if plan < 0 || cal < 0 {
+		t.Fatalf("sources missing: day-plan=%d calendar=%d", plan, cal)
+	}
+	if plan > cal {
+		t.Errorf("day-plan at %d must come before calendar at %d", plan, cal)
+	}
+}
+
+// habitAPI answers only RecentFacts — the whole input the behaviour profile
+// needs (Vikunja #254). Nothing is asked of the LLM, so nothing else is wired.
+type habitAPI struct {
+	ipc.UnimplementedCoreAPI
+	facts []ipc.Fact
+	err   error
+	calls int
+}
+
+func (a *habitAPI) RecentFacts(_ context.Context, _ int) ([]ipc.Fact, error) {
+	a.calls++
+	return a.facts, a.err
+}
+
+// tuesdayFacts — n weekly Tuesday rows for key, ending before now.
+func tuesdayFacts(key string, hh, weeks int, now time.Time) []ipc.Fact {
+	d := now
+	for d.Weekday() != time.Tuesday {
+		d = d.AddDate(0, 0, -1)
+	}
+	var out []ipc.Fact
+	for i := 0; i < weeks; i++ {
+		day := d.AddDate(0, 0, -7*i)
+		out = append(out, ipc.Fact{
+			Ts:   time.Date(day.Year(), day.Month(), day.Day(), hh, 0, 0, 0, now.Location()),
+			Kind: "self",
+			Key:  key,
+		})
+	}
+	return out
+}
+
+func TestQueryHabitsAnswersFromCountedFacts(t *testing.T) {
+	now := planDay() // a Monday
+	api := &habitAPI{facts: tuesdayFacts("workout", 19, 4, now)}
+	h := &reactiveHandler{api: api, now: func() time.Time { return now }}
+
+	reply, ok := h.queryHabits(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: "что я обычно делаю по вторникам?"},
+	})
+	if !ok {
+		t.Fatal("the habit source must claim a habit question")
+	}
+	if want := "по вторникам ты обычно тренируешься около 19:00."; reply != want {
+		t.Errorf("reply = %q, want %q", reply, want)
+	}
+}
+
+func TestQueryHabitsPassesOnEverythingElse(t *testing.T) {
+	now := planDay()
+	for _, q := range []string{"что я делаю в среду?", "что у меня сегодня?", "какие планы на сегодня?", ""} {
+		api := &habitAPI{}
+		h := &reactiveHandler{api: api, now: func() time.Time { return now }}
+		if reply, ok := h.queryHabits(context.Background(), &queryTurn{
+			dec: router.Decision{Intent: router.IntentQuery, Utterance: q},
+		}); ok {
+			t.Errorf("%q was claimed by the habit source (reply %q)", q, reply)
+		}
+		if api.calls != 0 {
+			t.Errorf("%q scanned the fact log for a profile it does not want", q)
+		}
+	}
+}
+
+// Both specific sources must precede the calendar listing, which matches any
+// utterance naming a day.
+func TestHabitSourcePrecedesCalendar(t *testing.T) {
+	habits, cal := -1, -1
+	for i, s := range querySources {
+		switch s.name {
+		case "habits":
+			habits = i
+		case "calendar":
+			cal = i
+		}
+	}
+	if habits < 0 || cal < 0 {
+		t.Fatalf("sources missing: habits=%d calendar=%d", habits, cal)
+	}
+	if habits > cal {
+		t.Errorf("habits at %d must come before calendar at %d", habits, cal)
+	}
+}

cmd/mavend/digest_test.go

@@ -0,0 +1,158 @@
+package main
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/loop"
+	"github.com/kami/maven/internal/store"
+)
+
+// Vikunja #281 — the fourth delivery outcome: a care candidate the restraint
+// gate suppresses (quiet hours / away / calendar-busy) is not necessarily
+// lost. If it's worth resurfacing (loop.DigestEligible), it's durably held
+// (internal/store's digest_entries) and spoken as one bundle once speaking
+// is appropriate again — never while the suppression reason still holds.
+
+func breakTrace(blockedBy string) *loop.TickTrace {
+	return &loop.TickTrace{
+		RuleTraces: []loop.RuleTrace{{
+			RuleName:        "break",
+			Severity:        loop.Sev2,
+			PredicateResult: true,
+			GateResult:      false,
+			GateBlockedBy:   blockedBy,
+		}},
+	}
+}
+
+// TestSuppressedCareDigestsAcrossQuietHours — a Sev2 care candidate blocked
+// by quiet hours is enqueued into the durable digest, and is spoken as a
+// "digest" nudge only once quiet hours actually end — never while still
+// suppressed (that would just be a second way to nag through quiet hours).
+func TestSuppressedCareDigestsAcrossQuietHours(t *testing.T) {
+	st := newTestStore(t)
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	ctx := context.Background()
+	now := refNow()
+
+	quiet := loop.State{Now: now, QuietHours: true, Presence: store.Present}
+	tl.enqueueSuppressedDigest(ctx, breakTrace("quiet_hours"), quiet, now)
+
+	entries, err := st.PendingDigestEntries(ctx, now)
+	if err != nil {
+		t.Fatalf("pending: %v", err)
+	}
+	if len(entries) != 1 || entries[0].Rule != "break" {
+		t.Fatalf("want 1 pending digest entry for break, got %+v", entries)
+	}
+
+	// still quiet hours: draining now must not speak — the same restraint
+	// that suppressed the live nudge must suppress the bundle too.
+	tl.maybeDrainDigest(ctx, quiet, now)
+	if len(sink.sends) != 0 {
+		t.Fatalf("digest must not drain while quiet hours holds, got %+v", sink.sends)
+	}
+
+	// quiet hours end: this is the moment speaking is appropriate again.
+	after := now.Add(time.Hour)
+	clear := loop.State{Now: after, QuietHours: false, Presence: store.Present}
+	tl.maybeDrainDigest(ctx, clear, after)
+
+	if len(sink.sends) != 1 {
+		t.Fatalf("want exactly 1 dispatched digest bundle, got %d: %+v", len(sink.sends), sink.sends)
+	}
+	if sink.sends[0].RuleName != "digest" {
+		t.Fatalf("want RuleName digest, got %q", sink.sends[0].RuleName)
+	}
+
+	remaining, err := st.PendingDigestEntries(ctx, after)
+	if err != nil {
+		t.Fatalf("pending after drain: %v", err)
+	}
+	if len(remaining) != 0 {
+		t.Fatalf("drained entry must no longer be pending, got %+v", remaining)
+	}
+}
+
+// TestSuppressedCareDigestDedupesAcrossTicks — quiet hours holding for
+// several ticks must not enqueue several copies of the same suppressed
+// nudge; he hears it once when the bundle finally drains.
+func TestSuppressedCareDigestDedupesAcrossTicks(t *testing.T) {
+	st := newTestStore(t)
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	ctx := context.Background()
+	now := refNow()
+
+	quiet := loop.State{Now: now, QuietHours: true, Presence: store.Present}
+	for i := 0; i < 3; i++ {
+		tl.enqueueSuppressedDigest(ctx, breakTrace("quiet_hours"), quiet, now.Add(time.Duration(i)*time.Minute))
+	}
+
+	entries, err := st.PendingDigestEntries(ctx, now)
+	if err != nil {
+		t.Fatalf("pending: %v", err)
+	}
+	if len(entries) != 1 {
+		t.Fatalf("3 suppressions of the same nudge must collapse to 1 pending entry, got %d", len(entries))
+	}
+}
+
+// TestSuppressedCareDigestExpiresRatherThanDeliveringLate — an entry that
+// aged out before the suppression cleared is dropped, not spoken late: a
+// two-day-old "you skipped a break" is noise, not news.
+func TestSuppressedCareDigestExpiresRatherThanDeliveringLate(t *testing.T) {
+	st := newTestStore(t)
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	ctx := context.Background()
+	now := refNow()
+
+	quiet := loop.State{Now: now, QuietHours: true, Presence: store.Present}
+	tl.enqueueSuppressedDigest(ctx, breakTrace("quiet_hours"), quiet, now)
+
+	// well past digestExpiry (24h) before the suppression ever clears.
+	stale := now.Add(48 * time.Hour)
+	tl.expireStaleDigest(ctx, stale)
+
+	clear := loop.State{Now: stale, QuietHours: false, Presence: store.Present}
+	tl.maybeDrainDigest(ctx, clear, stale)
+
+	if len(sink.sends) != 0 {
+		t.Fatalf("a stale digest entry must be dropped, not delivered late; got %+v", sink.sends)
+	}
+}
+
+// TestSuppressedCareDigestIgnoresHighSeverity — defense in depth at the
+// wiring layer: even if a RuleTrace somehow showed a high-severity rule
+// blocked by a care-only gate reason, the tick driver must not durably
+// digest it. Alarms bypass the gate and deliver now, unchanged; they must
+// never be silently delayed into a bundle.
+func TestSuppressedCareDigestIgnoresHighSeverity(t *testing.T) {
+	st := newTestStore(t)
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	ctx := context.Background()
+	now := refNow()
+
+	trace := &loop.TickTrace{RuleTraces: []loop.RuleTrace{{
+		RuleName:        "service_down",
+		Severity:        loop.Sev4,
+		PredicateResult: true,
+		GateResult:      false,
+		GateBlockedBy:   "quiet_hours",
+	}}}
+	quiet := loop.State{Now: now, QuietHours: true, Presence: store.Present}
+	tl.enqueueSuppressedDigest(ctx, trace, quiet, now)
+
+	entries, err := st.PendingDigestEntries(ctx, now)
+	if err != nil {
+		t.Fatalf("pending: %v", err)
+	}
+	if len(entries) != 0 {
+		t.Fatalf("high severity must never be digested, got %+v", entries)
+	}
+}

cmd/mavend/ecosystem_acts.go

@@ -0,0 +1,312 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"log"
+	"strings"
+
+	hexisclient "github.com/kami/hexis/pkg/client"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+)
+
+// praxisCapability is one arm of the Praxis act dispatch. This is an interface
+// rather than a map[string]func because each arm carries its own state: the
+// verb aliases it answers to, the trace name it records, and its own reply
+// formatting. The dispatch grows an arm per Praxis capability, so a new one is
+// added to praxisCapabilities below and nothing else changes.
+type praxisCapability interface {
+	// aliases are the verbs (router fn slots, EN and RU) this capability answers to.
+	aliases() []string
+	// handle runs the capability and returns the user-facing reply.
+	handle(ctx context.Context, h *reactiveHandler, px *praxisClient, dec router.Decision) string
+}
+
+// praxisCapabilities is the registry handlePraxisAct consults, in order.
+var praxisCapabilities = []praxisCapability{
+	listAttentionCapability{},
+	praxisItemAction{
+		verbs:   []string{"acknowledge_item", "принято", "понял", "поняла"},
+		ask:     "какой пункт отметить принятым?",
+		op:      "acknowledge",
+		failure: "не получилось отметить принятым.",
+		success: "принято.",
+		call: func(ctx context.Context, px *praxisClient, id string) error {
+			_, err := px.Acknowledge(ctx, id)
+			return err
+		},
+	},
+	praxisItemAction{
+		verbs:   []string{"resolve_item", "сделано", "готово", "решено"},
+		ask:     "какой пункт отметить сделанным?",
+		op:      "resolve",
+		failure: "не получилось отметить сделанным.",
+		success: "отмечено как сделано.",
+		call: func(ctx context.Context, px *praxisClient, id string) error {
+			_, err := px.Resolve(ctx, id)
+			return err
+		},
+	},
+	praxisItemAction{
+		verbs:   []string{"ignore_item", "игнорировать", "неважно"},
+		ask:     "какой пункт игнорировать?",
+		op:      "ignore",
+		failure: "не получилось проигнорировать.",
+		success: "проигнорировано.",
+		call: func(ctx context.Context, px *praxisClient, id string) error {
+			_, err := px.Ignore(ctx, id)
+			return err
+		},
+	},
+	praxisItemAction{
+		verbs:   []string{"pin_item", "закрепить"},
+		ask:     "какой пункт закрепить?",
+		op:      "pin",
+		failure: "не получилось закрепить.",
+		success: "закреплено.",
+		call: func(ctx context.Context, px *praxisClient, id string) error {
+			_, err := px.Pin(ctx, id, true)
+			return err
+		},
+	},
+	listChangesCapability{},
+}
+
+// handlePraxisAct — dispatches ecosystem tool acts through the Praxis tools API.
+// Returns "" when the act is not a Praxis verb (the caller falls through to the
+// system command executor). Returns a reply string otherwise.
+func (h *reactiveHandler) handlePraxisAct(ctx context.Context, dec router.Decision) string {
+	if h.ecosystem == nil || h.ecosystem.praxis == nil {
+		return ""
+	}
+	px := h.ecosystem.praxis
+	for _, capability := range praxisCapabilities {
+		for _, alias := range capability.aliases() {
+			if alias == dec.Slots.Fn {
+				return capability.handle(ctx, h, px, dec)
+			}
+		}
+	}
+	// Not a Praxis verb — let the caller fall through.
+	return ""
+}
+
+// praxisItemAction is the shared shape of the item-lifecycle capabilities: take
+// an item id from the value slot, call one Praxis endpoint, trace the result.
+type praxisItemAction struct {
+	verbs   []string
+	ask     string // reply when no item id was given
+	op      string // trace + log name of the operation
+	failure string // reply when the Praxis call errors
+	success string
+	call    func(ctx context.Context, px *praxisClient, id string) error
+}
+
+func (a praxisItemAction) aliases() []string { return a.verbs }
+
+func (a praxisItemAction) handle(ctx context.Context, h *reactiveHandler, px *praxisClient, dec router.Decision) string {
+	id := dec.Slots.Value
+	if id == "" {
+		return a.ask
+	}
+	if err := a.call(ctx, px, id); err != nil {
+		log.Printf("ecosystem: praxis %s %s: %v", a.op, id, err)
+		return a.failure
+	}
+	h.recordPraxisTrace(ctx, a.op, map[string]any{"item_id": id})
+	return a.success
+}
+
+// listAttentionCapability reads the attention digest and surfaces every item it speaks.
+type listAttentionCapability struct{}
+
+func (listAttentionCapability) aliases() []string {
+	return []string{"list_attention", "attention", "внимание", "что требует внимания", "что нового"}
+}
+
+func (listAttentionCapability) handle(ctx context.Context, h *reactiveHandler, px *praxisClient, _ router.Decision) string {
+	items, err := px.ListAttention(ctx, 20)
+	if err != nil {
+		log.Printf("ecosystem: praxis attention: %v", err)
+		return "не могу сейчас узнать, что требует внимания."
+	}
+	if len(items) == 0 {
+		return "ничего не требует внимания."
+	}
+	h.recordPraxisTrace(ctx, "list_attention", map[string]any{"count": len(items)})
+	var parts []string
+	for _, item := range items {
+		title, _ := item["title"].(string)
+		// importance arrives as JSON number ⇒ float64 over the HTTP contract.
+		importance, _ := item["importance"].(float64)
+		rule, _ := item["rule"].(string)
+		s := title
+		if importance > 0 {
+			s += fmt.Sprintf(" (важность %d", int(importance))
+			if rule != "" {
+				s += ": " + rule
+			}
+			s += ")"
+		}
+		parts = append(parts, s)
+
+		// Speaking an item surfaces it, it does not acknowledge it
+		// (ECOSYSTEM-SPEC.md §2.3: surfaced != acknowledged). Best-effort:
+		// a failed surface call must not block delivering the digest.
+		if id, ok := item["id"].(string); ok && id != "" {
+			if _, err := px.Surface(ctx, id); err != nil {
+				log.Printf("ecosystem: praxis surface %s: %v", id, err)
+			}
+		}
+	}
+	return "требует внимания: " + strings.Join(parts, "; ")
+}
+
+// listChangesCapability reads the recent-changes feed.
+type listChangesCapability struct{}
+
+func (listChangesCapability) aliases() []string {
+	return []string{"list_changes", "changes", "изменения", "что изменилось"}
+}
+
+func (listChangesCapability) handle(ctx context.Context, h *reactiveHandler, px *praxisClient, _ router.Decision) string {
+	changes, err := px.ListChanges(ctx, 20)
+	if err != nil {
+		log.Printf("ecosystem: praxis changes: %v", err)
+		return "не могу сейчас узнать об изменениях."
+	}
+	if len(changes) == 0 {
+		return "нет изменений."
+	}
+	h.recordPraxisTrace(ctx, "list_changes", map[string]any{"count": len(changes)})
+	var parts []string
+	for _, c := range changes {
+		title, _ := c["title"].(string)
+		typ, _ := c["change_type"].(string)
+		parts = append(parts, fmt.Sprintf("%s (%s)", title, typ))
+	}
+	return "изменения: " + strings.Join(parts, "; ")
+}
+
+// recordPraxisTrace — writes a fact recording a cross-service ecosystem call.
+// The fact is stored with source "praxis:trace" so the proactive loop can
+// reference it and the dashboard can display recent ecosystem activity.
+func (h *reactiveHandler) recordPraxisTrace(ctx context.Context, operation string, details map[string]any) {
+	now := h.now()
+	value := operation
+	if len(details) > 0 {
+		if b, err := json.Marshal(details); err == nil {
+			value = operation + " " + string(b)
+		}
+	}
+	_, _ = h.api.WriteFact(ctx, ipc.WriteFactReq{
+		Ts:         now,
+		Kind:       "system",
+		Key:        "praxis:" + operation,
+		Value:      value,
+		Source:     "praxis:trace",
+		Confidence: 1.0,
+	})
+}
+
+// handleHexisAct — resolves entity references through Nexus and executes
+// matching capabilities through Hexis. Returns a reply string when handled,
+// or "" to fall through to the system command executor.
+func (h *reactiveHandler) handleHexisAct(ctx context.Context, dec router.Decision) string {
+	if h.ecosystem == nil {
+		return ""
+	}
+
+	// Resolve the utterance text as an entity reference through Nexus. An
+	// ambiguous match must stop and clarify — never guess a mutation target.
+	entityID, displayName, ambiguous, err := h.ecosystem.resolveEntityReference(ctx, dec.Slots.Text, nil)
+	if err != nil {
+		// A genuine Nexus dependency failure, not "no such entity" — stop here
+		// and report degradation rather than silently falling through to the
+		// local command executor (ECOSYSTEM-SPEC.md: services degrade
+		// independently, never a silent all-clear).
+		return "экосистема недоступна, попробуй ещё раз."
+	}
+	if len(ambiguous) > 0 {
+		return "уточни, что именно: " + strings.Join(ambiguous, ", ") + "?"
+	}
+	if entityID == "" {
+		return ""
+	}
+
+	// Discover Hexis capabilities for this entity. A resolved entity with a
+	// genuine Hexis failure must not be treated as "no capabilities" and
+	// fall through to unrelated local execution.
+	caps, err := h.ecosystem.discoverCapabilities(ctx, entityID)
+	if err != nil {
+		return "экосистема недоступна, попробуй ещё раз."
+	}
+	if len(caps) == 0 {
+		return ""
+	}
+
+	// Match the user's verb to a capability by name/description. Collect all
+	// matches: more than one is itself ambiguous, so we ask rather than pick
+	// the first (ecosystem invariant: no arbitrary target for mutation).
+	verb := dec.Slots.Fn
+	if verb == "" {
+		verb = dec.Slots.Text
+	}
+	verbLower := strings.ToLower(verb)
+
+	var matches []*hexisclient.Capability
+	for i, c := range caps {
+		if strings.Contains(strings.ToLower(c.Name), verbLower) ||
+			(c.Description != "" && strings.Contains(strings.ToLower(c.Description), verbLower)) {
+			matches = append(matches, &caps[i])
+		}
+	}
+	if len(matches) == 0 {
+		return ""
+	}
+	if len(matches) > 1 {
+		var names []string
+		for _, m := range matches {
+			names = append(names, m.Name)
+		}
+		return "какую команду для " + displayName + ": " + strings.Join(names, ", ") + "?"
+	}
+	matched := matches[0]
+
+	// Read-only capabilities run immediately; mutating ones are parked for an
+	// explicit spoken confirm bound to this capability + target.
+	if !matched.ReadOnly {
+		h.mu.Lock()
+		h.pendingHexis = &pendingHexisExec{
+			capabilityID: matched.ID,
+			capName:      matched.Name,
+			entityID:     entityID,
+			displayName:  displayName,
+			expiry:       h.now().Add(confirmTTL),
+		}
+		h.mu.Unlock()
+		return "выполнить «" + matched.Name + "» для " + displayName + "? скажи «да» или «нет»."
+	}
+
+	return h.execHexis(ctx, matched.ID, matched.Name, entityID, displayName)
+}
+
+// execHexis runs a resolved capability and records a cross-service trace with
+// the correlation ID. It reports command success, never operational recovery
+// (Praxis observes recovery independently).
+func (h *reactiveHandler) execHexis(ctx context.Context, capID, capName, entityID, displayName string) string {
+	correlationID, err := h.ecosystem.executeCapability(ctx, capID, entityID, nil)
+	if err != nil {
+		log.Printf("ecosystem: hexis execute error (cor=%s): %v", correlationID, err)
+		return "не получилось выполнить команду для " + displayName + "."
+	}
+	h.recordPraxisTrace(ctx, "hexis:"+capName, map[string]any{
+		"entity_id":      entityID,
+		"entity_name":    displayName,
+		"capability":     capName,
+		"correlation_id": correlationID,
+	})
+	return "команда выполнена для " + displayName + "."
+}

cmd/mavend/feeds.go

@@ -0,0 +1,182 @@
+// mavend/feeds.go — the driver for RSS/Atom reading (Vikunja #258,
+// docs/plans/13-rss-news-feeds.md). The reader itself is pure and lives in
+// internal/rss; this is the impure half: a ticker, the guarded fetcher, and the
+// two adapters that let a pure package talk to the store.
+//
+// Why in-core rather than its own daemon like mavmaild and mavpoll: those two
+// hold a CREDENTIAL (an IMAP password, a zenmoney token), and the reason they
+// are separate processes is that core must never see it. A feed URL is public,
+// there is no secret to isolate, and a whole extra binary and compose service
+// would buy nothing. The other half of the mavpoll precedent — off unless
+// configured — is kept: no `feeds` block, no poller, no outbound request.
+//
+// It is its own goroutine, not a step on the tick: the tick has a delivery
+// deadline behind it, and a feed read is a network round-trip that nobody is
+// waiting on.
+//
+// Nothing here dispatches. A feed that announced itself would be a nag, so the
+// only output is notes with source "rss:<feed>", which the answer path reads
+// when he asks ("что нового в лентах?" — see queryFeeds in actions_query.go).
+package main
+
+import (
+	"context"
+	"log"
+	"net/url"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/rss"
+	"github.com/kami/maven/internal/webfetch"
+)
+
+// feedWorker — ticker + poller.
+type feedWorker struct {
+	poller   *rss.Poller
+	interval time.Duration
+}
+
+// feedTickInterval — how often the worker asks the poller what is due. Per-feed
+// cadence is the poller's business; this is just the granularity.
+const feedTickInterval = 5 * time.Minute
+
+// newFeedWorker wires feed reading, or returns nil when it must not run:
+// no `feeds` block (the normal case), or nothing valid in it. Every caller
+// checks for nil.
+func newFeedWorker(api ipc.CoreAPI, emb router.Embedder, cfg *config.Config) *feedWorker {
+	if cfg.Feeds == nil {
+		return nil
+	}
+	fc := cfg.Feeds
+
+	feeds := make([]rss.FeedConfig, 0, len(fc.Sources))
+	hosts := append([]string(nil), fc.AllowHosts...)
+	for _, s := range fc.Sources {
+		feeds = append(feeds, rss.FeedConfig{
+			Name:     s.Name,
+			URL:      s.URL,
+			Category: s.Category,
+			Interval: time.Duration(s.Interval),
+			Include:  s.Include,
+			Exclude:  s.Exclude,
+		})
+		// Each configured feed's own host is allowed. The allowlist is then
+		// exactly "the feeds he asked for", so a redirect off to somewhere else
+		// is refused by the fetcher rather than followed.
+		if u, err := url.Parse(s.URL); err == nil && u.Hostname() != "" {
+			hosts = append(hosts, u.Hostname())
+		}
+	}
+
+	fetcher := webfetch.New(webfetch.Config{
+		AllowHosts: hosts,
+		Timeout:    time.Duration(fc.Timeout),
+		MaxBytes:   fc.MaxBytes,
+	})
+	poller := rss.NewPoller(feeds, &feedFetcher{f: fetcher}, api, &factMarks{api: api},
+		embedderFor(emb), nil, rss.Config{
+			DefaultInterval: time.Duration(fc.PollInterval),
+			MaxItems:        fc.MaxItems,
+			MaxAge:          time.Duration(fc.MaxAge),
+		})
+	if poller == nil {
+		log.Printf("feeds: configured but nothing pollable — feed reading disabled")
+		return nil
+	}
+	log.Printf("feeds: reading %d feed(s), checking what is due every %s", len(feeds), feedTickInterval)
+	return &feedWorker{poller: poller, interval: feedTickInterval}
+}
+
+// run polls what is due until ctx is canceled. The first round runs immediately
+// so a restart does not blind her for the first interval; it writes notes only,
+// so an early round cannot startle anyone.
+func (w *feedWorker) run(ctx context.Context) {
+	w.poller.PollDue(ctx, time.Now())
+	t := time.NewTicker(w.interval)
+	defer t.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case now := <-t.C:
+			w.poller.PollDue(ctx, now)
+		}
+	}
+}
+
+// embedderOf — the voice wiring's embedder, or nil when voice is not wired.
+// Feed notes are embedded with the SAME model the rest of the store uses, or not
+// at all; a second embedder would write vectors nothing can search.
+func embedderOf(w *voiceWiring) router.Embedder {
+	if w == nil {
+		return nil
+	}
+	return w.embedder
+}
+
+// feedFetcher adapts webfetch to rss.Fetcher — the pure package names the two
+// fields it needs and stays free of net/http.
+type feedFetcher struct{ f *webfetch.Fetcher }
+
+func (a *feedFetcher) Get(ctx context.Context, u string) (*rss.Body, error) {
+	resp, err := a.f.Get(ctx, u)
+	if err != nil {
+		return nil, err
+	}
+	return &rss.Body{Bytes: resp.Body}, nil
+}
+
+// factMarks stores "how far this feed was read" as a config fact, the same
+// mechanism the plan named and the same one the pattern tick uses for its own
+// bookkeeping. Durable, inspectable on /dash, and cheap.
+type factMarks struct{ api ipc.CoreAPI }
+
+func markKey(feed string) string { return "rss:latest:" + feed }
+
+func (m *factMarks) LastMark(ctx context.Context, feed string) (time.Time, error) {
+	f, err := m.api.LatestFact(ctx, markKey(feed))
+	if err != nil {
+		// No mark yet is not an error worth propagating: the poller treats a
+		// zero time as a cold start.
+		return time.Time{}, nil
+	}
+	t, err := time.Parse(time.RFC3339, f.Value)
+	if err != nil {
+		return time.Time{}, nil
+	}
+	return t, nil
+}
+
+func (m *factMarks) SetMark(ctx context.Context, feed string, at time.Time) error {
+	_, err := m.api.WriteFact(ctx, ipc.WriteFactReq{
+		Ts:         time.Now(),
+		Kind:       "config",
+		Key:        markKey(feed),
+		Value:      at.UTC().Format(time.RFC3339),
+		Source:     "poll:rss",
+		Confidence: 1.0,
+	})
+	return err
+}
+
+// embedderFor adapts router.Embedder to rss.Embedder, and returns nil when
+// there is none — a note without a vector is still a note the recent-notes path
+// can read.
+//
+// EmbedPassage, not Embed: a feed item is text being searched FOR, and the e5
+// embedder is asymmetric. Getting this backwards makes the item unfindable by
+// the question that should have matched it.
+func embedderFor(emb router.Embedder) rss.Embedder {
+	if emb == nil {
+		return nil
+	}
+	return passageEmbedder{emb}
+}
+
+type passageEmbedder struct{ e router.Embedder }
+
+func (p passageEmbedder) Embed(ctx context.Context, text string) ([]float32, error) {
+	return router.EmbedPassage(ctx, p.e, text)
+}

cmd/mavend/feeds_test.go

@@ -0,0 +1,177 @@
+package main
+
+import (
+	"context"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/phraser"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/rss"
+	"github.com/kami/maven/internal/voice"
+)
+
+// buildFeedHandler — a handler with the given feed notes already stored. No
+// embedder: the feed source answers from recent notes by source, which is what
+// makes it work for notes written before an embedder existed.
+func buildFeedHandler(t *testing.T, feedsOn bool, notes ...ipc.Note) *reactiveHandler {
+	t.Helper()
+	ctx := context.Background()
+	st := newTestStore(t)
+	now := time.Now()
+	for i, n := range notes {
+		ts := now.Add(time.Duration(i) * time.Minute)
+		if _, err := st.WriteNote(ctx, ts, n.Text, nil, n.Source); err != nil {
+			t.Fatalf("WriteNote: %v", err)
+		}
+	}
+	return &reactiveHandler{
+		api:      ipc.NewStoreAPI(st),
+		replier:  voice.NewStubReplier(),
+		phraser:  phraser.NewStub(),
+		now:      func() time.Time { return now },
+		feedsOn:  feedsOn,
+		embedder: nil,
+	}
+}
+
+func askFeeds(t *testing.T, h *reactiveHandler, q string) (string, bool) {
+	t.Helper()
+	return h.queryFeeds(context.Background(), &queryTurn{
+		dec: router.Decision{Intent: router.IntentQuery, Utterance: q},
+	})
+}
+
+func TestQueryFeedsReadsFeedNotes(t *testing.T) {
+	h := buildFeedHandler(t, true,
+		ipc.Note{Text: "Новая уязвимость в ядре [технологии]\nпатч вышел\nhttps://example.org/a", Source: "rss:habr"},
+		ipc.Note{Text: "что-то он сам сказал", Source: "tap:voice"},
+	)
+	reply, ok := askFeeds(t, h, "что нового в лентах?")
+	if !ok {
+		t.Fatal("the feed source did not claim the question")
+	}
+	if !strings.Contains(reply, "уязвимость") {
+		t.Errorf("reply = %q, want the headline", reply)
+	}
+	if strings.Contains(reply, "он сам сказал") {
+		t.Errorf("a note he dictated leaked into the feed answer: %q", reply)
+	}
+	// She reads the headline, not the summary and not the URL.
+	if strings.Contains(reply, "https://") || strings.Contains(reply, "патч вышел") {
+		t.Errorf("reply = %q, want the title line only", reply)
+	}
+}
+
+func TestQueryFeedsByCategory(t *testing.T) {
+	h := buildFeedHandler(t, true,
+		ipc.Note{Text: "Релиз ядра [технологии]", Source: "rss:habr"},
+		ipc.Note{Text: "Выборы отложены [политика]", Source: "rss:news"},
+	)
+	reply, ok := askFeeds(t, h, "что нового по технологиям?")
+	if !ok {
+		t.Fatal("not claimed")
+	}
+	if !strings.Contains(reply, "ядра") || strings.Contains(reply, "Выборы") {
+		t.Fatalf("reply = %q, want only the технологии item", reply)
+	}
+	reply, _ = askFeeds(t, h, "что нового по спорту?")
+	if !strings.Contains(reply, "ничего") {
+		t.Fatalf("reply = %q, want an honest empty answer for an unread category", reply)
+	}
+}
+
+// "не настроены" and "ничего нового" are different truths, and neither may be
+// answered by the model inventing a bulletin.
+func TestQueryFeedsOffAndEmptyDiffer(t *testing.T) {
+	off := buildFeedHandler(t, false)
+	reply, ok := askFeeds(t, off, "что нового?")
+	if !ok || !strings.Contains(reply, "не настроены") {
+		t.Fatalf("feeds off: reply = %q, ok = %v", reply, ok)
+	}
+	on := buildFeedHandler(t, true)
+	reply, ok = askFeeds(t, on, "что нового?")
+	if !ok || !strings.Contains(reply, "ничего нового") {
+		t.Fatalf("feeds on but empty: reply = %q, ok = %v", reply, ok)
+	}
+}
+
+func TestQueryFeedsPassesOnANonFeedQuestion(t *testing.T) {
+	h := buildFeedHandler(t, true)
+	if reply, ok := askFeeds(t, h, "напомни полить цветы"); ok {
+		t.Fatalf("claimed an unrelated question with %q", reply)
+	}
+}
+
+// The mark is what stops a restart from re-noting yesterday's headlines, so the
+// fact round-trip is worth a test of its own.
+func TestFactMarksRoundTrip(t *testing.T) {
+	st := newTestStore(t)
+	m := &factMarks{api: ipc.NewStoreAPI(st)}
+	ctx := context.Background()
+
+	at, err := m.LastMark(ctx, "habr")
+	if err != nil || !at.IsZero() {
+		t.Fatalf("no mark yet: got %v, %v — want zero time and no error", at, err)
+	}
+	want := time.Date(2026, 7, 28, 10, 0, 0, 0, time.UTC)
+	if err := m.SetMark(ctx, "habr", want); err != nil {
+		t.Fatal(err)
+	}
+	got, err := m.LastMark(ctx, "habr")
+	if err != nil {
+		t.Fatal(err)
+	}
+	if !got.Equal(want) {
+		t.Fatalf("mark = %v, want %v", got, want)
+	}
+}
+
+// Off unless configured, checked at the wiring seam: no `feeds` block ⇒ no
+// worker ⇒ no outbound request is possible.
+func TestNewFeedWorkerOffByDefault(t *testing.T) {
+	st := newTestStore(t)
+	api := ipc.NewStoreAPI(st)
+	if w := newFeedWorker(api, nil, &config.Config{}); w != nil {
+		t.Fatal("a config with no feeds block wired a feed worker")
+	}
+	// An empty sources list is normalised to "off" by config.Load; the worker
+	// refuses it too, so a hand-built Config cannot switch it on by accident.
+	if w := newFeedWorker(api, nil, &config.Config{Feeds: &config.FeedsConfig{}}); w != nil {
+		t.Fatal("an empty sources list wired a feed worker")
+	}
+	cfg := &config.Config{Feeds: &config.FeedsConfig{Sources: []config.FeedSourceConfig{
+		{Name: "habr", URL: "https://example.org/rss"},
+	}}}
+	w := newFeedWorker(api, nil, cfg)
+	if w == nil {
+		t.Fatal("a configured feed did not wire a worker")
+	}
+	if got := w.poller.Feeds(); len(got) != 1 || got[0].Name != "habr" {
+		t.Fatalf("feeds = %+v", got)
+	}
+}
+
+// The fetcher the worker builds must be allowlisted to the configured feeds and
+// nothing else — the crawler's SSRF guards are only worth as much as the
+// allowlist handed to them.
+func TestFeedWorkerFetcherIsAllowlisted(t *testing.T) {
+	cfg := &config.Config{Feeds: &config.FeedsConfig{Sources: []config.FeedSourceConfig{
+		{Name: "habr", URL: "https://feeds.example.org/rss"},
+	}}}
+	w := newFeedWorker(ipc.NewStoreAPI(newTestStore(t)), nil, cfg)
+	if w == nil {
+		t.Fatal("no worker")
+	}
+	// PollFeed goes through the guarded fetcher; a feed URL pointing at the box
+	// itself must fail rather than be read.
+	_, err := w.poller.PollFeed(context.Background(), rss.FeedConfig{
+		Name: "evil", URL: "http://127.0.0.1:9100/mcp",
+	}, time.Now())
+	if err == nil {
+		t.Fatal("the poller fetched a private address")
+	}
+}

cmd/mavend/mail.go

@@ -0,0 +1,158 @@
+// mavend/mail.go — core's half of the email reader (Vikunja #246,
+// docs/plans/01-email-reader.md).
+//
+// The split: cmd/mavmaild holds the IMAP credential, connects to the mailbox
+// and converts messages to plaintext; it hands each message to core over
+// ipc.MethodIngestMail. Core runs the extraction on the resident model —
+// llama-server lives in this process, spawned by the phraser — and writes what
+// comes back through the one task intake seam.
+//
+// What this file may produce is exactly one thing: rows in `tasks` with status
+// "candidate". No fact, no reminder, no note, no nudge, no calendar event. A
+// 1.7B misreading a mail can therefore put a wrong line on a review page and
+// nothing else; it can never make Maven speak, and it can never make her
+// recite something out of an advert as true.
+//
+// Off unless configured twice over: no `email` block in mavend.json ⇒ the IPC
+// method does not exist; no llama-server phraser ⇒ same. A reader pointed at a
+// core that is not set up for mail gets ErrUnknownMethod rather than silence.
+package main
+
+import (
+	"context"
+	"fmt"
+	"log"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/email"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/phraser"
+	"github.com/kami/maven/internal/store"
+)
+
+// evidenceMaxChars — how much of the subject line is kept as a candidate's
+// evidence. Enough to recognise the mail on /tasks, not enough to turn the task
+// list into a copy of his mailbox.
+const evidenceMaxChars = 160
+
+// mailIntake — extraction + capture for one message at a time.
+type mailIntake struct {
+	st      *store.Store
+	ex      *email.Extractor
+	timeout time.Duration
+	now     func() time.Time
+}
+
+// newMailIntake returns nil when mail ingestion must not be available, which is
+// the default. Both preconditions are real:
+//
+//   - no cfg.Email ⇒ not configured, and a capability is off unless configured;
+//   - no llama-server phraser ⇒ nothing to extract with. There is deliberately
+//     no keyword fallback: "the subject line became a task" is not extraction,
+//     it is a mailbox rendered as a to-do list, and it would fill the review
+//     page faster than he could clear it.
+func newMailIntake(st *store.Store, phr phraser.Phraser, cfg *config.Config) *mailIntake {
+	if cfg.Email == nil {
+		return nil
+	}
+	lp, ok := phr.(*phraser.LLMPhraser)
+	if !ok {
+		log.Printf("mail intake: configured but no llama-server phraser — mail ingestion disabled")
+		return nil
+	}
+	timeout := time.Duration(cfg.Email.Timeout)
+	if timeout <= 0 {
+		timeout = config.DefaultEmailTimeout
+	}
+	ex := email.NewExtractor(llm.New(lp.BaseURL(), timeout), cfg.Email.MaxTasks, contextBlockFn(cfg, time.Now))
+	log.Printf("mail intake: enabled (max %d candidates per message, timeout %s)", cfg.Email.MaxTasks, timeout)
+	return &mailIntake{st: st, ex: ex, timeout: timeout, now: time.Now}
+}
+
+// ingest handles one ipc.MethodIngestMail call.
+//
+// Junk and empty messages are answered Skipped without touching the model — the
+// reader's header filter is what keeps the resident model off newsletters.
+//
+// Every candidate is captured with Status "candidate", Source "email:<mailbox>"
+// and the subject as Evidence. CaptureTask dedupes on normalised text among
+// live rows, so a mailbox re-read after a restart produces Created=0 rather
+// than a second copy of every task.
+func (m *mailIntake) ingest(ctx context.Context, req ipc.IngestMailReq) (ipc.IngestMailResp, error) {
+	msg := email.Message{
+		UID:     req.UID,
+		From:    req.From,
+		Subject: req.Subject,
+		Date:    req.Date,
+		Body:    req.Body,
+		Junk:    req.Junk,
+	}
+	if msg.Junk || (msg.Subject == "" && msg.Body == "") {
+		return ipc.IngestMailResp{Skipped: true}, nil
+	}
+
+	ctx, cancel := context.WithTimeout(ctx, m.timeout)
+	defer cancel()
+	cands, err := m.ex.Extract(ctx, msg)
+	if err != nil {
+		// The error from internal/email never carries mail text; keep it that way
+		// by not adding the subject here.
+		return ipc.IngestMailResp{}, fmt.Errorf("mail intake: uid %d: %w", req.UID, err)
+	}
+	if len(cands) == 0 {
+		return ipc.IngestMailResp{}, nil
+	}
+
+	source := email.SourcePrefix + req.Mailbox
+	evidence := truncateRunes(req.Subject, evidenceMaxChars)
+	now := m.now()
+	var resp ipc.IngestMailResp
+	for _, c := range cands {
+		t := store.Task{
+			CreatedTs: now,
+			Text:      c.Text,
+			Source:    source,
+			Evidence:  evidence,
+			// The one status this path may ever write. Anything Maven derived from
+			// something she read is a suggestion until he confirms it on /tasks.
+			Status: store.TaskCandidate,
+		}
+		if due, ok := email.ParseDue(c.Due); ok {
+			t.Due = &due
+		}
+		id, created, err := m.st.CaptureTask(ctx, t)
+		if err != nil {
+			return resp, fmt.Errorf("mail intake: capture: %w", err)
+		}
+		resp.TaskIDs = append(resp.TaskIDs, id)
+		if created {
+			resp.Created++
+		}
+	}
+	// Counts only: the log line names the mailbox and the UID, never the subject,
+	// the sender or the task text. Reviewing a candidate is what /tasks is for.
+	log.Printf("mail intake: %s uid %d → %d candidate(s), %d new", source, req.UID, len(cands), resp.Created)
+	return resp, nil
+}
+
+// wireMailIntake installs the IPC hook, or leaves it nil so the method reports
+// ErrUnknownMethod. Called on both startup paths (unlocked boot and passkey
+// unlock) so mail behaves the same either way.
+func wireMailIntake(srv *ipc.Server, st *store.Store, phr phraser.Phraser, cfg *config.Config) {
+	mi := newMailIntake(st, phr, cfg)
+	if mi == nil {
+		return
+	}
+	srv.IngestMailFn = mi.ingest
+}
+
+// truncateRunes cuts a string to n runes, marking the cut.
+func truncateRunes(s string, n int) string {
+	r := []rune(s)
+	if len(r) <= n {
+		return s
+	}
+	return string(r[:n]) + "…"
+}

cmd/mavend/mail_test.go

@@ -0,0 +1,182 @@
+package main
+
+import (
+	"context"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/email"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/store"
+)
+
+// mailLLM — a canned extraction reply.
+type mailLLM struct {
+	reply string
+	calls int
+}
+
+func (m *mailLLM) Complete(_ context.Context, _ llm.Req) (string, error) {
+	m.calls++
+	return m.reply, nil
+}
+
+func newTestIntake(t *testing.T, reply string) (*mailIntake, *store.Store, *mailLLM) {
+	t.Helper()
+	st := newTestStore(t)
+	fake := &mailLLM{reply: reply}
+	return &mailIntake{
+		st:      st,
+		ex:      email.NewExtractor(fake, 0, nil),
+		timeout: 5 * time.Second,
+		now:     func() time.Time { return time.Date(2026, 8, 1, 10, 0, 0, 0, time.UTC) },
+	}, st, fake
+}
+
+func ingestReq() ipc.IngestMailReq {
+	return ipc.IngestMailReq{
+		Mailbox: "INBOX", UID: 42,
+		From:    "billing@isp.example",
+		Subject: "Счёт за интернет",
+		Body:    "Оплатите счёт до 5 августа.",
+	}
+}
+
+// The one property that matters: a mail-derived task is a candidate, attributed
+// to the mailbox, with the subject as reviewable evidence — and nothing else is
+// written.
+func TestIngestCapturesCandidates(t *testing.T) {
+	mi, st, _ := newTestIntake(t, `[{"text":"оплатить счёт за интернет","due":"2026-08-05"}]`)
+	resp, err := mi.ingest(context.Background(), ingestReq())
+	if err != nil {
+		t.Fatalf("ingest: %v", err)
+	}
+	if resp.Created != 1 || len(resp.TaskIDs) != 1 {
+		t.Fatalf("resp = %+v, want one created task", resp)
+	}
+	tasks, err := st.ListTasks(context.Background(), "")
+	if err != nil {
+		t.Fatalf("list: %v", err)
+	}
+	if len(tasks) != 1 {
+		t.Fatalf("got %d tasks, want 1", len(tasks))
+	}
+	got := tasks[0]
+	if got.Status != store.TaskCandidate {
+		t.Errorf("status = %q, want %q — mail may only produce candidates", got.Status, store.TaskCandidate)
+	}
+	if got.Source != "email:INBOX" {
+		t.Errorf("source = %q, want email:INBOX", got.Source)
+	}
+	if got.Evidence != "Счёт за интернет" {
+		t.Errorf("evidence = %q, want the subject line", got.Evidence)
+	}
+	if got.Due == nil || got.Due.Format("2006-01-02") != "2026-08-05" {
+		t.Errorf("due = %v, want 2026-08-05", got.Due)
+	}
+	// Nothing else may have been written: no reminder, no fact.
+	rem, err := st.ListReminders(context.Background(), 10)
+	if err != nil {
+		t.Fatalf("list reminders: %v", err)
+	}
+	if len(rem) != 0 {
+		t.Errorf("mail created %d reminders; a misread mail must never be able to fire", len(rem))
+	}
+}
+
+// Re-reading a mailbox must not grow the list — CaptureTask dedupes among live
+// rows, and the intake relies on exactly that.
+func TestIngestSameMailTwiceIsIdempotent(t *testing.T) {
+	mi, st, _ := newTestIntake(t, `[{"text":"оплатить счёт","due":""}]`)
+	if _, err := mi.ingest(context.Background(), ingestReq()); err != nil {
+		t.Fatalf("first ingest: %v", err)
+	}
+	resp, err := mi.ingest(context.Background(), ingestReq())
+	if err != nil {
+		t.Fatalf("second ingest: %v", err)
+	}
+	if resp.Created != 0 || len(resp.TaskIDs) != 1 {
+		t.Errorf("resp = %+v, want the existing row and Created=0", resp)
+	}
+	tasks, _ := st.ListTasks(context.Background(), "")
+	if len(tasks) != 1 {
+		t.Errorf("got %d tasks after two reads, want 1", len(tasks))
+	}
+}
+
+func TestIngestJunkSkipsTheModel(t *testing.T) {
+	mi, st, fake := newTestIntake(t, `[{"text":"купить со скидкой","due":""}]`)
+	req := ingestReq()
+	req.Junk = true
+	resp, err := mi.ingest(context.Background(), req)
+	if err != nil {
+		t.Fatalf("ingest: %v", err)
+	}
+	if !resp.Skipped || resp.Created != 0 {
+		t.Errorf("resp = %+v, want skipped", resp)
+	}
+	if fake.calls != 0 {
+		t.Errorf("model called %d times for junk, want 0", fake.calls)
+	}
+	if tasks, _ := st.ListTasks(context.Background(), ""); len(tasks) != 0 {
+		t.Errorf("junk produced %d tasks, want 0", len(tasks))
+	}
+}
+
+func TestIngestEmptyMessageSkipped(t *testing.T) {
+	mi, _, fake := newTestIntake(t, "[]")
+	resp, err := mi.ingest(context.Background(), ipc.IngestMailReq{Mailbox: "INBOX", UID: 1})
+	if err != nil || !resp.Skipped {
+		t.Fatalf("resp = %+v, err = %v; want skipped", resp, err)
+	}
+	if fake.calls != 0 {
+		t.Errorf("model called %d times for an empty message, want 0", fake.calls)
+	}
+}
+
+func TestIngestNoTasksWritesNothing(t *testing.T) {
+	mi, st, _ := newTestIntake(t, "[]")
+	resp, err := mi.ingest(context.Background(), ingestReq())
+	if err != nil {
+		t.Fatalf("ingest: %v", err)
+	}
+	if resp.Created != 0 || len(resp.TaskIDs) != 0 || resp.Skipped {
+		t.Errorf("resp = %+v, want nothing captured and not skipped", resp)
+	}
+	if tasks, _ := st.ListTasks(context.Background(), ""); len(tasks) != 0 {
+		t.Errorf("got %d tasks, want 0", len(tasks))
+	}
+}
+
+func TestIngestTruncatesEvidence(t *testing.T) {
+	mi, st, _ := newTestIntake(t, `[{"text":"дело","due":""}]`)
+	req := ingestReq()
+	req.Subject = strings.Repeat("щ", 400)
+	if _, err := mi.ingest(context.Background(), req); err != nil {
+		t.Fatalf("ingest: %v", err)
+	}
+	tasks, _ := st.ListTasks(context.Background(), "")
+	if len(tasks) != 1 {
+		t.Fatalf("got %d tasks, want 1", len(tasks))
+	}
+	if n := len([]rune(tasks[0].Evidence)); n > evidenceMaxChars+1 {
+		t.Errorf("evidence kept %d runes, want ≤ %d", n, evidenceMaxChars)
+	}
+}
+
+// Off unless configured: no email block ⇒ no intake, so the IPC method does not
+// exist at all.
+func TestNewMailIntakeOffWithoutConfig(t *testing.T) {
+	st := newTestStore(t)
+	if mi := newMailIntake(st, nil, &config.Config{}); mi != nil {
+		t.Error("no email block must mean no mail intake")
+	}
+	// Configured but with a non-LLM phraser: still off — there is no fallback
+	// extraction, by design.
+	if mi := newMailIntake(st, nil, &config.Config{Email: &config.EmailConfig{}}); mi != nil {
+		t.Error("without a llama-server phraser there is nothing to extract with")
+	}
+}

cmd/mavend/main.go

@@ -58,6 +58,7 @@ import (
 	"github.com/kami/maven/internal/delivery/telegramsink"
 	"github.com/kami/maven/internal/ipc"
 	"github.com/kami/maven/internal/loop"
+	"github.com/kami/maven/internal/persona"
 	"github.com/kami/maven/internal/phraser"
 	"github.com/kami/maven/internal/store"
 	"github.com/kami/maven/internal/webauthn"
@@ -96,100 +97,12 @@ func main() {
 	}
 }
 
-// lockedAPI is a dummy CoreAPI used while the daemon is locked. Every method
-// returns errLocked. The wire protocol's StoreAPI methods all go through the
-// Server dispatch on CoreAPI, so returning errLocked from each is correct.
-type lockedAPI struct{}
-
-var _ ipc.CoreAPI = (*lockedAPI)(nil)
-
-func (l *lockedAPI) WriteFact(ctx context.Context, req ipc.WriteFactReq) (int64, error) {
-	return 0, errLocked
-}
-func (l *lockedAPI) LatestFact(ctx context.Context, key string) (ipc.Fact, error) {
-	return ipc.Fact{}, errLocked
-}
-func (l *lockedAPI) LatestFactBySource(ctx context.Context, key, source string) (ipc.Fact, error) {
-	return ipc.Fact{}, errLocked
-}
-func (l *lockedAPI) Since(ctx context.Context, key string, now time.Time) (time.Duration, error) {
-	return 0, errLocked
-}
-func (l *lockedAPI) Presence(ctx context.Context) (ipc.Presence, error) {
-	return ipc.Presence{}, errLocked
-}
-func (l *lockedAPI) CreateReminder(ctx context.Context, fire time.Time, payload, cron string) (int64, error) {
-	return 0, errLocked
-}
-func (l *lockedAPI) MarkReminder(ctx context.Context, id int64, status string) error {
-	return errLocked
-}
-func (l *lockedAPI) ListReminders(ctx context.Context, n int) ([]ipc.Reminder, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) RecordNudge(ctx context.Context, rule, channel, message string, ts time.Time) (int64, error) {
-	return 0, errLocked
-}
-func (l *lockedAPI) ResolveNudge(ctx context.Context, id int64, outcome string, ts time.Time) error {
-	return errLocked
-}
-func (l *lockedAPI) RecentOutcomes(ctx context.Context, rule string, n int) ([]string, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) RecentFacts(ctx context.Context, n int) ([]ipc.Fact, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) CalendarEvents(ctx context.Context, from, to time.Time) ([]ipc.Fact, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) RecentNudges(ctx context.Context, n int) ([]ipc.Nudge, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) WriteNote(ctx context.Context, ts time.Time, text string, embedding []float32, source string) (int64, error) {
-	return 0, errLocked
-}
-func (l *lockedAPI) QueryNotes(ctx context.Context, embedding []float32, k int) ([]ipc.Note, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) RecentNotes(ctx context.Context, n int) ([]ipc.Note, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) ProposeTool(ctx context.Context, name, utterance, scope string, ts time.Time) (bool, error) {
-	return false, errLocked
-}
-func (l *lockedAPI) EnableTool(ctx context.Context, name string, cmd []string, destructive bool, scope string, ts time.Time) error {
-	return errLocked
-}
-func (l *lockedAPI) DisableTool(ctx context.Context, name string) error { return errLocked }
-func (l *lockedAPI) DeleteTool(ctx context.Context, name string) error  { return errLocked }
-func (l *lockedAPI) ListProposedRoutines(ctx context.Context) ([]ipc.ProposedRoutine, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) DismissProposedRoutine(ctx context.Context, id int64) error { return errLocked }
-func (l *lockedAPI) AcceptProposedRoutine(ctx context.Context, id int64) error {
-	return errLocked
-}
-func (l *lockedAPI) LookupTool(ctx context.Context, name string) (ipc.Tool, error) {
-	return ipc.Tool{}, errLocked
-}
-func (l *lockedAPI) ListTools(ctx context.Context, status string) ([]ipc.Tool, error) {
-	return nil, errLocked
-}
-func (l *lockedAPI) RevertFact(ctx context.Context, key string) (int64, error) { return 0, errLocked }
-func (l *lockedAPI) Chat(ctx context.Context, text string) (string, error) {
-	return "", errLocked
-}
-func (l *lockedAPI) TickTrace(ctx context.Context) (ipc.TickTrace, error) {
-	return ipc.TickTrace{}, errLocked
-}
-func (l *lockedAPI) MorningStatus(ctx context.Context) ([]ipc.MorningRoutineStatus, error) {
-	return nil, errLocked
-}
-
 func run(args []string) error {
 	cfgPath := flag.String("config", defaultConfigPath(), "path to mavend JSON config")
 	wrappedKeyPath := flag.String("wrapped-key-file", "", "path to wrapped encryption key blob (enables cold-start unlock)")
+	reembed := flag.Bool("reembed", false, "re-embed every stored note and fact with the configured embedder, then serve normally (run once after an embedder swap)")
 	flag.CommandLine.Parse(args)
+	reembedOnStart = *reembed
 	cfg, err := config.Load(*cfgPath)
 	if err != nil {
 		return err
@@ -256,6 +169,8 @@ func run(args []string) error {
 		coreAPI    ipc.CoreAPI
 		eco        *ecosystemWiring
 		factWorker *factEnrichmentWorker
+		evalWorker *memoryEvalWorker // nil ⇒ memory evaluation off (the default)
+		feedWkr    *feedWorker       // nil ⇒ no feed is read (the default)
 	)
 
 	if !locked {
@@ -269,13 +184,14 @@ func run(args []string) error {
 		phr = phraser.NewStub()
 		if cfg.Phraser != nil {
 			pc := phraser.Config{
-				ModelPath:  cfg.Phraser.ModelPath,
+				ModelPath:    cfg.Phraser.ModelPath,
-				BinPath:    cfg.Phraser.BinPath,
+				BinPath:      cfg.Phraser.BinPath,
-				Listen:     cfg.Phraser.Listen,
+				Listen:       cfg.Phraser.Listen,
-				NGpuLayers: cfg.Phraser.NGpuLayers,
+				NGpuLayers:   cfg.Phraser.NGpuLayers,
-				NCtx:       cfg.Phraser.NCtx,
+				NCtx:         cfg.Phraser.NCtx,
-				Timeout:    time.Duration(cfg.Phraser.Timeout),
+				Timeout:      time.Duration(cfg.Phraser.Timeout),
-				Persona:    personaFromCfg(cfg),
+				LLMNudges:    cfg.Phraser.LLMNudges,
+				ContextBlock: contextBlockFn(cfg, time.Now),
 			}
 			if pc.BinPath == "" {
 				pc.BinPath = "llama-server"
@@ -347,21 +263,29 @@ func run(args []string) error {
 		tickInterval := time.Duration(cfg.TickInterval)
 		repeatInterval := time.Duration(cfg.RepeatInterval)
 		autotuneInterval := time.Duration(cfg.AutotuneInterval)
-		tl = newTickLoop(st, gatherer, dispatcher, phr, rules, tickInterval, repeatInterval, autotuneInterval, cfg.Digest, routinesFromConfig(cfg.Routines), config.MorningRoutinesFromConfig(cfg.MorningRoutines))
+		tl = newTickLoop(st, gatherer, dispatcher, phr, rules, tickInterval, repeatInterval, autotuneInterval, cfg.Digest, routinesFromConfig(cfg.Routines), config.MorningRoutinesFromConfig(cfg.MorningRoutines), cfg.PatternProposals)
 		factWorker = newFactEnrichmentWorker(st, eco, time.Duration(cfg.FactEnrichmentInterval))
+		evalWorker = newMemoryEvalWorker(st, phr, cfg)
+		feedWkr = newFeedWorker(ipc.NewStoreAPI(st), embedderOf(voiceW), cfg)
 
 		coreAPI = &daemonAPI{
 			CoreAPI:          ipc.NewStoreAPI(st),
 			getTrace:         tl.trace,
 			getMorningStatus: func(ctx context.Context) []ipc.MorningRoutineStatus { return tl.morningStatus(ctx, time.Now()) },
+			getDayPlan:       func(ctx context.Context) ipc.DayPlan { return tl.dayPlan(ctx, time.Now()) },
 		}
 		if voiceW != nil && voiceW.handler != nil {
 			api := coreAPI.(*daemonAPI)
 			api.chatFn = voiceW.handler.handleText
 		}
 	} else {
-		// locked mode: dummy CoreAPI that returns errLocked for everything
+		// locked mode: no real store yet, so there's no meaningful CoreAPI to
-		coreAPI = &lockedAPI{}
+		// serve. srv.Check below is the actual guard — every CoreAPI call is
+		// refused before it reaches this value. This is just a safe non-nil
+		// placeholder: if the guard is ever bypassed by a bug, calls land
+		// here and fail loudly with ipc.ErrNotImplemented instead of a nil
+		// dereference or, worse, silently succeeding.
+		coreAPI = ipc.UnimplementedCoreAPI{}
 	}
 
 	// ----- IPC boundary (core ↔ modules) -----
@@ -372,7 +296,17 @@ func run(args []string) error {
 
 	passkeySess := webauthn.NewPasskeySession(5 * time.Minute)
 
-	// Set Server.Check — in locked mode, block everything except unlock-path methods.
+	// Set Server.Check — the single authorization guard, run once by
+	// Server.dispatch before any CoreAPI method is called (see
+	// internal/ipc/server.go). In locked mode this is the ONLY thing
+	// standing between an unauthenticated caller and the store: it must
+	// default-deny, with an explicit allowlist for the two methods the
+	// unlock flow itself needs (MethodAssertStepUp, MethodUnlock — neither
+	// of which touches CoreAPI; dispatch handles them directly via
+	// srv.StepUp/srv.UnlockFn). Forgetting to allowlist a new unlock-path
+	// method fails safe (denied); forgetting to guard a new CoreAPI method
+	// is impossible because there is nothing left to forget — every method
+	// not in the allowlist is refused by construction.
 	if locked {
 		srv.Check = func(ctx context.Context, m ipc.Method, _ json.RawMessage) error {
 			switch m {
@@ -388,6 +322,13 @@ func run(args []string) error {
 
 	srv.StepUp = func(ctx context.Context) error { return passkeySess.Assert(ctx, auth.Scope{}) }
 
+	// Mail ingestion (Vikunja #246): the hook stays nil unless an email block is
+	// configured and there is a llama-server to extract with, in which case
+	// ipc.MethodIngestMail reports ErrUnknownMethod.
+	if !locked {
+		wireMailIntake(srv, st, phr, cfg)
+	}
+
 	// WrapKeyFn — wraps the env key with a passkey credential public key and
 	// persists the wrapped blob. Only wired when the daemon has the key in
 	// memory (env key mode). Called by mavweb after passkey enrollment.
@@ -439,13 +380,14 @@ func run(args []string) error {
 			phr = phraser.NewStub()
 			if cfg.Phraser != nil {
 				pc := phraser.Config{
-					ModelPath:  cfg.Phraser.ModelPath,
+					ModelPath:    cfg.Phraser.ModelPath,
-					BinPath:    cfg.Phraser.BinPath,
+					BinPath:      cfg.Phraser.BinPath,
-					Listen:     cfg.Phraser.Listen,
+					Listen:       cfg.Phraser.Listen,
-					NGpuLayers: cfg.Phraser.NGpuLayers,
+					NGpuLayers:   cfg.Phraser.NGpuLayers,
-					NCtx:       cfg.Phraser.NCtx,
+					NCtx:         cfg.Phraser.NCtx,
-					Timeout:    time.Duration(cfg.Phraser.Timeout),
+					Timeout:      time.Duration(cfg.Phraser.Timeout),
-					Persona:    personaFromCfg(cfg),
+					LLMNudges:    cfg.Phraser.LLMNudges,
+					ContextBlock: contextBlockFn(cfg, time.Now),
 				}
 				if pc.BinPath == "" {
 					pc.BinPath = "llama-server"
@@ -508,20 +450,24 @@ func run(args []string) error {
 			tickInterval := time.Duration(cfg.TickInterval)
 			repeatInterval := time.Duration(cfg.RepeatInterval)
 			autotuneInterval := time.Duration(cfg.AutotuneInterval)
-			tl = newTickLoop(st, gatherer, dispatcher, phr, rules, tickInterval, repeatInterval, autotuneInterval, cfg.Digest, routinesFromConfig(cfg.Routines), config.MorningRoutinesFromConfig(cfg.MorningRoutines))
+			tl = newTickLoop(st, gatherer, dispatcher, phr, rules, tickInterval, repeatInterval, autotuneInterval, cfg.Digest, routinesFromConfig(cfg.Routines), config.MorningRoutinesFromConfig(cfg.MorningRoutines), cfg.PatternProposals)
 			factWorker = newFactEnrichmentWorker(st, eco, time.Duration(cfg.FactEnrichmentInterval))
+			evalWorker = newMemoryEvalWorker(st, phr, cfg)
+			feedWkr = newFeedWorker(ipc.NewStoreAPI(st), embedderOf(voiceW), cfg)
 
-			// Swap the CoreAPI from lockedAPI to the real store adapter.
+			// Swap the CoreAPI from the locked placeholder to the real store adapter.
 			newAPI := &daemonAPI{
 				CoreAPI:          ipc.NewStoreAPI(st),
 				getTrace:         tl.trace,
 				getMorningStatus: func(ctx context.Context) []ipc.MorningRoutineStatus { return tl.morningStatus(ctx, time.Now()) },
+				getDayPlan:       func(ctx context.Context) ipc.DayPlan { return tl.dayPlan(ctx, time.Now()) },
 			}
 			if voiceW != nil && voiceW.handler != nil {
 				newAPI.chatFn = voiceW.handler.handleText
 			}
 			srv.SetAPI(newAPI)
 			srv.Check = (&auth.Gate{Enrollment: auth.NewFloorEnrollment(), Session: passkeySess}).Check
+			wireMailIntake(srv, st, phr, cfg)
 
 			// Start voice server.
 			if voiceW != nil {
@@ -546,6 +492,20 @@ func run(args []string) error {
 				factWorker.run(ctx)
 			}()
 
+			// Start background memory evaluation (nil unless configured).
+			if evalWorker != nil {
+				go func() {
+					evalWorker.run(ctx)
+				}()
+			}
+
+			// Start feed reading (nil unless configured).
+			if feedWkr != nil {
+				go func() {
+					feedWkr.run(ctx)
+				}()
+			}
+
 			dl.unlock()
 			log.Printf("mavend: unlocked via passkey assertion")
 			return nil
@@ -584,6 +544,20 @@ func run(args []string) error {
 			defer wg.Done()
 			factWorker.run(ctx)
 		}()
+		if evalWorker != nil {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				evalWorker.run(ctx)
+			}()
+		}
+		if feedWkr != nil {
+			wg.Add(1)
+			go func() {
+				defer wg.Done()
+				feedWkr.run(ctx)
+			}()
+		}
 	}
 
 	<-ctx.Done()
@@ -599,12 +573,29 @@ func run(args []string) error {
 	return nil
 }
 
-// personaFromCfg extracts the voice persona from the config, or returns ""
+// personaFacts reads the optional, deployment-specific facts (his name, his
-// when voice isn't configured. Used to pass a character prompt into the
+// city, the free-text persona string) out of the config. Everything here may
-// LLM phraser without requiring voice to be enabled.
+// be empty — the context block is correct without any of it.
-func personaFromCfg(cfg *config.Config) string {
+func personaFacts(cfg *config.Config) persona.Facts {
-	if cfg.Voice != nil {
+	f := persona.Facts{
-		return cfg.Voice.Persona
+		// Telegram lives outside the voice block, so it counts either way.
+		Telegram: cfg.Telegram != nil && cfg.Telegram.BotToken != "" && cfg.Telegram.ChatID != "",
 	}
-	return ""
+	if cfg.Voice == nil {
+		return f
+	}
+	f.OwnerName = cfg.Voice.OwnerName
+	f.City = cfg.Voice.City
+	f.Static = cfg.Voice.Persona
+	// Same test wireVoice uses to pick the real provider over the stub.
+	f.Weather = cfg.Voice.Weather != nil && cfg.Voice.Weather.Provider == "open-meteo"
+	f.Tools = len(cfg.Voice.Tools) > 0
+	return f
+}
+
+// contextBlockFn returns the per-turn renderer of the shared context block.
+// Per turn, not once at startup, because the block states the current time.
+func contextBlockFn(cfg *config.Config, now func() time.Time) func() string {
+	f := personaFacts(cfg)
+	return func() string { return f.Block(now()) }
 }

cmd/mavend/memoryeval.go

@@ -0,0 +1,88 @@
+// mavend/memoryeval.go — the driver for background memory evaluation
+// (Vikunja #248). The evaluator itself is pure-ish and lives in
+// internal/memeval; this is the one impure part: a ticker, the store, and the
+// resident model's base URL.
+//
+// It is its own goroutine and NOT a step on the main tick, deliberately. The
+// tick runs every 60s and has a delivery deadline behind it; an evaluation is
+// a multi-second LLM round-trip on the same llama-server that answers voice
+// turns, and it happens hourly at most. Bolting it onto the tick would make
+// every hour's tick the slow one for no benefit.
+package main
+
+import (
+	"context"
+	"log"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/memeval"
+	"github.com/kami/maven/internal/phraser"
+	"github.com/kami/maven/internal/store"
+)
+
+// memoryEvalWorker — ticker + evaluator.
+type memoryEvalWorker struct {
+	eval     *memeval.Evaluator
+	interval time.Duration
+}
+
+// newMemoryEvalWorker wires the evaluation loop, or returns nil when it should
+// not run at all. nil is the normal case and every caller must handle it:
+//
+//   - no memory_eval config block ⇒ off (a capability is off unless configured);
+//   - no LLM phraser ⇒ nothing to evaluate with. There is no template fallback
+//     here on purpose: a "memory evaluation" assembled from string templates
+//     would be a fixed sentence pretending to be an observation.
+func newMemoryEvalWorker(st *store.Store, phr phraser.Phraser, cfg *config.Config) *memoryEvalWorker {
+	if cfg.MemoryEval == nil {
+		return nil
+	}
+	lp, ok := phr.(*phraser.LLMPhraser)
+	if !ok {
+		log.Printf("memory eval: configured but no llama-server phraser — evaluation disabled")
+		return nil
+	}
+	interval := time.Duration(cfg.MemoryEval.Interval)
+	if interval <= 0 {
+		interval = config.DefaultMemoryEvalInterval
+	}
+	// A generous per-request timeout: this is a long prompt to a Thinking model
+	// and nobody is waiting on the answer.
+	client := llm.New(lp.BaseURL(), 5*time.Minute)
+	ev := memeval.NewEvaluator(st, st, client, memeval.Config{
+		MaxItems:      cfg.MemoryEval.MaxItems,
+		MinConfidence: cfg.MemoryEval.MinConfidence,
+		ContextBlock:  contextBlockFn(cfg, time.Now),
+	})
+	log.Printf("memory eval: enabled, every %s", interval)
+	return &memoryEvalWorker{eval: ev, interval: interval}
+}
+
+// run evaluates every interval until ctx is canceled.
+//
+// The first evaluation waits a full interval rather than firing at startup, the
+// opposite of the tick loop's cold-start behaviour. A tick that fires late is a
+// nudge that arrives late; an evaluation that fires late is nothing at all, and
+// the alternative is a heavy LLM call competing with startup — including with
+// the first voice turn after a restart.
+func (w *memoryEvalWorker) run(ctx context.Context) {
+	ticker := time.NewTicker(w.interval)
+	defer ticker.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			return
+		case now := <-ticker.C:
+			obs, err := w.eval.Evaluate(ctx, now)
+			if err != nil {
+				log.Printf("memory eval: %v", err)
+				continue
+			}
+			for _, o := range obs {
+				log.Printf("memory eval: noted (%.2f, %s): %s", o.Conf, o.Action, o.Text)
+			}
+		}
+	}
+}

cmd/mavend/patterns.go

@@ -0,0 +1,120 @@
+// mavend/patterns.go — the shared detect+propose step of pattern inference
+// (Vikunja #43). Event *extraction* (fact -> action/object) happens at fact-
+// write time in detectPattern below, tied to whichever channel wrote the
+// fact. Detection — turning a run of events into a proposed routine — is
+// channel-agnostic: it only needs what's already in the events table, so it
+// runs both right after a voice fact-write (for the immediate "напоминать?"
+// confirmation) and, proactively, from the digestion tick (tick.go's
+// detectPatterns) over every action+object pair on record, not just the one
+// that was just talked about.
+package main
+
+import (
+	"context"
+	"errors"
+	"fmt"
+	"log"
+	"time"
+
+	"github.com/kami/maven/internal/pattern"
+	"github.com/kami/maven/internal/store"
+)
+
+// detectAndPropose runs the pattern detector over every recorded event for
+// action+object and, if a stable pattern is found and nothing has been
+// proposed/accepted/dismissed for this pair yet, creates a proposed_routines
+// row. Returns (nil, 0, nil) — not an error — whenever there is nothing new
+// to report: too few events, irregular intervals, or a pair that already has
+// a row in any status. That last case is the one that matters most: it is
+// how a routine the owner already DISMISSED stays dismissed forever, because
+// the row survives dismissal (status flips in place, see
+// store.DismissProposedRoutine) and both the Lookup check here and the
+// table's UNIQUE(action, object) constraint refuse to create a second one.
+func detectAndPropose(ctx context.Context, ds *store.Store, action, object string, ts time.Time) (*pattern.ProposedRoutine, int64, error) {
+	events, err := ds.EventsFor(ctx, action, object)
+	if err != nil {
+		return nil, 0, fmt.Errorf("events for %s/%s: %w", action, object, err)
+	}
+	patEvents := make([]pattern.Event, len(events))
+	for i, e := range events {
+		patEvents[i] = pattern.Event{
+			FactID: e.FactID,
+			Action: e.Action,
+			Object: e.Object,
+			Ts:     e.Ts,
+		}
+	}
+	r, err := pattern.Detect(patEvents)
+	if err != nil {
+		return nil, 0, fmt.Errorf("detect %s/%s: %w", action, object, err)
+	}
+	if r == nil {
+		return nil, 0, nil // not enough data or intervals too irregular
+	}
+
+	// Belt: check first so the common "nothing new" case never even attempts
+	// an insert. Suspenders: CreateProposedRoutine's ON CONFLICT DO NOTHING
+	// (backed by the UNIQUE(action,object) constraint) is the actual
+	// guarantee — this Lookup is an optimization, not the source of truth.
+	existing, err := ds.LookupProposedRoutine(ctx, r.Action, r.Object)
+	if err != nil {
+		return nil, 0, fmt.Errorf("lookup proposed routine %s/%s: %w", action, object, err)
+	}
+	if existing != nil {
+		return nil, 0, nil // already proposed, accepted, or dismissed — say nothing
+	}
+
+	id, err := ds.CreateProposedRoutine(ctx, r.Action, r.Object, r.IntervalDays, ts)
+	if err != nil {
+		if errors.Is(err, store.ErrProposedRoutineExists) {
+			return nil, 0, nil // lost a race with another caller — not an error
+		}
+		return nil, 0, fmt.Errorf("create proposed routine %s/%s: %w", action, object, err)
+	}
+	return r, id, nil
+}
+
+// detectPattern extracts an event from the written fact and runs the pattern
+// detector. If a stable recurring pattern is found and no proposed routine
+// exists for this action+object yet, one is created and the user is prompted
+// to confirm via the park() mechanism. Returns the suggestion phrase when a
+// new proposal was created and parked; "" otherwise.
+func (h *reactiveHandler) detectPattern(ctx context.Context, factID int64, key, value string, ts time.Time) string {
+	ev := pattern.Extract(factID, key, value, ts)
+	if ev == nil {
+		return "" // not an actionable event
+	}
+	if _, err := h.dataStore.CreateEvent(ctx, factID, ev.Action, ev.Object, ts); err != nil {
+		log.Printf("voice: create event: %v", err)
+		return ""
+	}
+	// Detect+propose (Vikunja #43) is shared with the digestion tick's
+	// proactive scan — see detectAndPropose above. Event *extraction* stays
+	// here, tied to this fact write; detection over the accumulated history does
+	// not need to happen right now for the voice path to have already done
+	// its job — it's dedupe-safe to also let the next tick find the same
+	// pattern independently.
+	r, id, err := detectAndPropose(ctx, h.dataStore, ev.Action, ev.Object, ts)
+	if err != nil {
+		log.Printf("voice: detect pattern %s/%s: %v", ev.Action, ev.Object, err)
+		return ""
+	}
+	if r == nil {
+		return "" // not enough data, too irregular, or already proposed/decided
+	}
+	log.Printf("voice: proposed routine: %s/%s every %.1f days", r.Action, r.Object, r.IntervalDays)
+
+	// Park the proposal for voice confirmation.
+	phrase := pattern.PhraseRoutine(r)
+	h.mu.Lock()
+	h.pendingRoutine = &pendingRoutineConfirm{
+		routineID: id,
+		action:    r.Action,
+		object:    r.Object,
+		interval:  r.IntervalDays,
+		phrase:    phrase,
+		expiry:    ts.Add(confirmTTL),
+	}
+	h.mu.Unlock()
+	return phrase
+}

cmd/mavend/patterns_test.go

@@ -0,0 +1,285 @@
+package main
+
+import (
+	"context"
+	"database/sql"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/delivery"
+	"github.com/kami/maven/internal/loop"
+	"github.com/kami/maven/internal/pattern"
+	"github.com/kami/maven/internal/store"
+)
+
+// seedRefillEvents writes N weekly "refill/cat_water" events straight to the
+// events table — this is what the tick reads, independent of any utterance.
+func seedRefillEvents(t *testing.T, st *store.Store, ctx context.Context, base time.Time, n int) {
+	t.Helper()
+	for i := 0; i < n; i++ {
+		factID, err := st.WriteFact(ctx, base.Add(time.Duration(i)*7*24*time.Hour), store.KindSelf,
+			"cat_water", "refill", "test", 1.0, sql.NullInt64{})
+		if err != nil {
+			t.Fatalf("write fact %d: %v", i, err)
+		}
+		if _, err := st.CreateEvent(ctx, factID, "refill", "cat_water", base.Add(time.Duration(i)*7*24*time.Hour)); err != nil {
+			t.Fatalf("create event %d: %v", i, err)
+		}
+	}
+}
+
+// TestTickDetectsPatternFromStoredEvents proves the tick notices a pattern on
+// its own, reading straight from the store — not as a side effect of a live
+// utterance (Vikunja #43). MinEvents weekly events with no voice turn in
+// sight must produce exactly one proposed routine.
+func TestTickDetectsPatternFromStoredEvents(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+
+	tl := newTestTickLoop(t, st, &fakeSink{}, nil)
+	tl.detectPatterns(ctx, now, loop.State{})
+
+	rows, err := st.ListProposedRoutines(ctx)
+	if err != nil {
+		t.Fatalf("list proposed routines: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("proposed routines = %d, want 1: %+v", len(rows), rows)
+	}
+	if rows[0].Action != "refill" || rows[0].Object != "cat_water" {
+		t.Errorf("proposed routine = %s/%s, want refill/cat_water", rows[0].Action, rows[0].Object)
+	}
+}
+
+// TestTickPatternDetectionIsIdempotent proves running the tick's pattern scan
+// twice does not spam a second proposal for the same pair, and that the store
+// itself is what stops the duplicate (not tick-local state) — the whole point
+// of the guard, since the tick has no memory of what it proposed last time.
+func TestTickPatternDetectionIsIdempotent(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+
+	tl := newTestTickLoop(t, st, &fakeSink{}, nil)
+	tl.detectPatterns(ctx, now, loop.State{})
+	tl.detectPatterns(ctx, now.Add(time.Hour), loop.State{})
+
+	rows, err := st.ListProposedRoutines(ctx)
+	if err != nil {
+		t.Fatalf("list proposed routines: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("proposed routines after two ticks = %d, want 1 (no duplicate): %+v", len(rows), rows)
+	}
+}
+
+// TestTickPatternDetectionRespectsDismissal proves the single worst failure
+// mode here — a proposal the owner already said no to coming back on the next
+// tick — cannot happen. Dismissal flips the row's status in place; it must
+// still be there to block re-proposal.
+func TestTickPatternDetectionRespectsDismissal(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+
+	tl := newTestTickLoop(t, st, &fakeSink{}, nil)
+	tl.detectPatterns(ctx, now, loop.State{})
+
+	rows, err := st.ListProposedRoutines(ctx)
+	if err != nil {
+		t.Fatalf("list proposed routines: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("setup: proposed routines = %d, want 1", len(rows))
+	}
+	if err := st.DismissProposedRoutine(ctx, rows[0].ID); err != nil {
+		t.Fatalf("dismiss: %v", err)
+	}
+
+	// More events for the same pair arrive, and the tick runs again — a
+	// dismissed pattern must not resurface.
+	seedRefillEvents(t, st, ctx, now.Add(30*24*time.Hour), pattern.MinEvents)
+	tl.detectPatterns(ctx, now.Add(60*24*time.Hour), loop.State{})
+
+	proposed, err := st.ListProposedRoutinesByStatus(ctx, store.RoutineProposed)
+	if err != nil {
+		t.Fatalf("list proposed: %v", err)
+	}
+	if len(proposed) != 0 {
+		t.Fatalf("a dismissed pattern came back: %+v", proposed)
+	}
+	all, err := st.ListProposedRoutinesByStatus(ctx, "")
+	if err != nil {
+		t.Fatalf("list all: %v", err)
+	}
+	if len(all) != 1 {
+		t.Fatalf("total rows for the pair = %d, want 1 (still dismissed, not duplicated): %+v", len(all), all)
+	}
+	if all[0].Status != store.RoutineDismissed {
+		t.Errorf("status = %s, want dismissed", all[0].Status)
+	}
+}
+
+// proposalRule — the rule name announceProposal uses for the seeded pair.
+const proposalRule = "proposal:refill cat_water"
+
+// TestTickProposalSilentByDefault — detection is always on, announcing is not.
+// With no pattern_proposals block the tick still records the proposal, and says
+// nothing about it: Maven is not autonomous, so a behaviour that speaks without
+// being asked stays off until it is configured.
+func TestTickProposalSilentByDefault(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+	markPresent(t, st, ctx, now)
+
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	tl.tick(ctx, now)
+
+	if n := countSends(sink, proposalRule); n != 0 {
+		t.Fatalf("announced %d proposals with no config, want 0", n)
+	}
+	rows, err := st.ListProposedRoutinesByStatus(ctx, store.RoutineProposed)
+	if err != nil {
+		t.Fatalf("list proposed: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("proposed routines = %d, want 1 (silent, but recorded)", len(rows))
+	}
+}
+
+// TestTickAnnouncesProposalWhenConfigured — with notify on, the proposal goes
+// out once through the ordinary delivery path, worded by the detector itself.
+// Later ticks stay quiet because the pair is already proposed: one pattern is
+// one announcement, ever.
+func TestTickAnnouncesProposalWhenConfigured(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+	markPresent(t, st, ctx, now)
+
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	tl.proposalCfg = &config.PatternProposalConfig{Notify: true}
+	tl.tick(ctx, now)
+
+	var got *delivery.Sendable
+	for i := range sink.sends {
+		if sink.sends[i].RuleName == proposalRule {
+			got = &sink.sends[i]
+		}
+	}
+	if got == nil {
+		t.Fatalf("proposal was not announced; sends=%+v", sink.sends)
+	}
+	if !strings.Contains(got.Body, "напоминать?") {
+		t.Errorf("body = %q, want the detector's own question", got.Body)
+	}
+	if got.Channel != delivery.ChannelVoice {
+		t.Errorf("channel = %v, want voice (sev1, present)", got.Channel)
+	}
+
+	// A month of further ticks: the pair already has a row, so there is
+	// nothing new to detect and nothing more to say.
+	sink.sends = nil
+	later := now.Add(40 * 24 * time.Hour)
+	markPresent(t, st, ctx, later)
+	tl.tick(ctx, later)
+	if n := countSends(sink, proposalRule); n != 0 {
+		t.Fatalf("re-announced an existing proposal %d times, want 0", n)
+	}
+}
+
+// TestTickProposalRespectsGate — a proposal is the least urgent thing Maven can
+// say, so it is sev1 and the restraint gate suppresses it. Away presence means
+// it is not announced at all: it is not held, not retried, it just lives on
+// /routines. The proposal row is still written — noticing is never gated.
+func TestTickProposalRespectsGate(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+	// no presence probes ⇒ away ⇒ care-class gate blocks.
+
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	tl.proposalCfg = &config.PatternProposalConfig{Notify: true}
+	tl.tick(ctx, now)
+
+	if n := countSends(sink, proposalRule); n != 0 {
+		t.Fatalf("away: announced %d proposals, want 0", n)
+	}
+	if !tl.lastProposalAt.IsZero() {
+		t.Error("cooldown clock advanced on a suppressed announcement")
+	}
+	rows, err := st.ListProposedRoutinesByStatus(ctx, store.RoutineProposed)
+	if err != nil {
+		t.Fatalf("list proposed: %v", err)
+	}
+	if len(rows) != 1 {
+		t.Fatalf("proposed routines = %d, want 1 (detection is never gated)", len(rows))
+	}
+}
+
+// TestTickProposalCooldownSpacesAnnouncements — two patterns detected on the
+// same tick must not become two interruptions. The second one waits for the
+// cooldown, and is on /routines meanwhile.
+func TestTickProposalCooldownSpacesAnnouncements(t *testing.T) {
+	st := newTestStore(t)
+	ctx := context.Background()
+	now := refNow()
+	seedRefillEvents(t, st, ctx, now, pattern.MinEvents)
+	for i := 0; i < pattern.MinEvents; i++ {
+		ts := now.Add(time.Duration(i) * 3 * 24 * time.Hour)
+		factID, err := st.WriteFact(ctx, ts, store.KindSelf, "litter_box", "clean", "test", 1.0, sql.NullInt64{})
+		if err != nil {
+			t.Fatalf("write fact: %v", err)
+		}
+		if _, err := st.CreateEvent(ctx, factID, "clean", "litter_box", ts); err != nil {
+			t.Fatalf("create event: %v", err)
+		}
+	}
+	markPresent(t, st, ctx, now)
+
+	sink := &fakeSink{}
+	tl := newTestTickLoop(t, st, sink, nil)
+	tl.proposalCfg = &config.PatternProposalConfig{Notify: true, Cooldown: config.Duration(24 * time.Hour)}
+	tl.tick(ctx, now)
+
+	announced := 0
+	for _, s := range sink.sends {
+		if strings.HasPrefix(s.RuleName, "proposal:") {
+			announced++
+		}
+	}
+	if announced != 1 {
+		t.Fatalf("announced %d proposals on one tick, want exactly 1", announced)
+	}
+	rows, err := st.ListProposedRoutinesByStatus(ctx, store.RoutineProposed)
+	if err != nil {
+		t.Fatalf("list proposed: %v", err)
+	}
+	if len(rows) != 2 {
+		t.Fatalf("proposed routines = %d, want 2 (both recorded, one announced)", len(rows))
+	}
+
+	// Still inside the cooldown: silence, even though a proposal is pending.
+	sink.sends = nil
+	soon := now.Add(time.Hour)
+	markPresent(t, st, ctx, soon)
+	tl.tick(ctx, soon)
+	for _, s := range sink.sends {
+		if strings.HasPrefix(s.RuleName, "proposal:") {
+			t.Fatalf("announced %q inside the cooldown", s.RuleName)
+		}
+	}
+}

cmd/mavend/query_recall_test.go

@@ -0,0 +1,158 @@
+package main
+
+import (
+	"context"
+	"fmt"
+	"math"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/memory"
+	"github.com/kami/maven/internal/phraser"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/voice"
+)
+
+// fixedEmbedder hands back a vector chosen per text, so a test can say exactly
+// how close each stored memory is to the question. The real embedders make
+// scores that are realistic but not controllable, and this test is about the
+// gate, not about the embedder.
+type fixedEmbedder struct{ vecs map[string][]float32 }
+
+func (f *fixedEmbedder) Dim() int     { return 4 }
+func (f *fixedEmbedder) Close() error { return nil }
+
+func (f *fixedEmbedder) Embed(_ context.Context, text string) ([]float32, error) {
+	v, ok := f.vecs[text]
+	if !ok {
+		return nil, fmt.Errorf("fixedEmbedder: no vector for %q", text)
+	}
+	return v, nil
+}
+
+// scoreVec builds a unit vector whose cosine against the query vector
+// (1,0,0,0) is exactly score.
+func scoreVec(score float64) []float32 {
+	rest := math.Sqrt(1 - score*score)
+	return []float32{float32(score), float32(rest), 0, 0}
+}
+
+// recordingPhraser remembers what the query path handed it to phrase, which is
+// how the test can tell which pass produced the answer.
+type recordingPhraser struct {
+	*phraser.Stub
+	notes []string
+}
+
+func (r *recordingPhraser) PhraseQuery(ctx context.Context, utterance string, notes []string) (string, error) {
+	r.notes = notes
+	return r.Stub.PhraseQuery(ctx, utterance, notes)
+}
+
+// recallCase — one stored memory: its text, how close it is to the question,
+// whether it is a note or a fact, and whether the notes table holds it too.
+type recallCase struct {
+	text  string
+	score float64
+	kind  string
+}
+
+// buildRecallHandler stores the given memories and returns a handler whose
+// query path can be run directly. Notes go into BOTH the notes table and the
+// vector index, which is what the daemon does (voice.go's IntentNote).
+func buildRecallHandler(t *testing.T, question string, mems []recallCase) (*reactiveHandler, *recordingPhraser) {
+	t.Helper()
+	ctx := context.Background()
+	st := newTestStore(t)
+	emb := &fixedEmbedder{vecs: map[string][]float32{question: {1, 0, 0, 0}}}
+	mem := memory.NewInMemoryStore()
+	now := time.Now()
+
+	for i, m := range mems {
+		vec := scoreVec(m.score)
+		emb.vecs[m.text] = vec
+		id := fmt.Sprintf("%s:%d", m.kind, i)
+		if m.kind == "note" {
+			if _, err := st.WriteNote(ctx, now, m.text, vec, "tap:voice"); err != nil {
+				t.Fatalf("WriteNote: %v", err)
+			}
+		}
+		if err := mem.Insert(ctx, id, vec, map[string]string{"text": m.text, "type": m.kind}); err != nil {
+			t.Fatalf("memory insert: %v", err)
+		}
+	}
+
+	phr := &recordingPhraser{Stub: phraser.NewStub()}
+	h := &reactiveHandler{
+		api:             ipc.NewStoreAPI(st),
+		embedder:        emb,
+		replier:         voice.NewStubReplier(),
+		phraser:         phr,
+		now:             func() time.Time { return now },
+		memStore:        mem,
+		dataStore:       st,
+		queryMinScore:   0.55,
+		queryMinMargin:  0.008,
+		weatherProvider: nil,
+	}
+	return h, phr
+}
+
+func askQuery(t *testing.T, h *reactiveHandler, question string) string {
+	t.Helper()
+	return h.applyAction(context.Background(), router.Decision{
+		Intent:    router.IntentQuery,
+		Utterance: question,
+	})
+}
+
+// TestQueryRecallNoteCanWin — the note-recall regression (Vikunja #373). Notes
+// and facts share one vector index, and a note that clearly beats everything
+// else must be the answer. Before the fix the memory pass only ran after the
+// notes-only gate had already rejected the same note at the same score, so only
+// a fact could ever come back from it.
+func TestQueryRecallNoteCanWin(t *testing.T) {
+	const q = "где молоко"
+
+	t.Run("a clearly best note answers", func(t *testing.T) {
+		h, phr := buildRecallHandler(t, q, []recallCase{
+			{text: "молоко стоит в холодильнике", score: 0.90, kind: "note"},
+			{text: "выучил пару аккордов", score: 0.50, kind: "note"},
+		})
+		reply := askQuery(t, h, q)
+		if want := "вот что я нашла: молоко стоит в холодильнике"; reply != want {
+			t.Errorf("reply %q, want %q", reply, want)
+		}
+		// One text, the winning memory's — the answer came from the memory
+		// pass, not from handing the phraser every note in the table.
+		if len(phr.notes) != 1 || phr.notes[0] != "молоко стоит в холодильнике" {
+			t.Errorf("phraser got %q, want just the recalled note", phr.notes)
+		}
+	})
+
+	// The other half of "one gate over everything": a fact that matches better
+	// than the best note now answers, instead of losing to a note that only had
+	// to beat other notes.
+	t.Run("the better-matching fact answers", func(t *testing.T) {
+		h, _ := buildRecallHandler(t, q, []recallCase{
+			{text: "молоко стоит в холодильнике", score: 0.80, kind: "note"},
+			{text: "купил молоко в среду", score: 0.95, kind: "fact"},
+		})
+		if reply := askQuery(t, h, q); reply != "купил молоко в среду" {
+			t.Errorf("reply %q, want the fact read back", reply)
+		}
+	})
+
+	// The gate is untouched: two memories this close mean the embedder cannot
+	// tell them apart, and silence still beats a coin flip.
+	t.Run("no clear best stays silent", func(t *testing.T) {
+		h, _ := buildRecallHandler(t, q, []recallCase{
+			{text: "молоко стоит в холодильнике", score: 0.860, kind: "note"},
+			{text: "молоко закончилось", score: 0.858, kind: "note"},
+		})
+		if reply := askQuery(t, h, q); reply != "не знаю." {
+			t.Errorf("reply %q, want silence", reply)
+		}
+	})
+}

cmd/mavend/quiet_toggle.go

@@ -0,0 +1,144 @@
+// Quiet-mode toggle recognition — the pre-route keyword check that lets
+// "тихий режим" flip the daemon-wide quiet_hours config without going through
+// the router. Moved out of voice.go unchanged (Vikunja #321); the tests live in
+// quiet_toggle_test.go.
+package main
+
+import (
+	"context"
+	"log"
+	"strings"
+	"unicode"
+
+	"github.com/kami/maven/internal/ipc"
+)
+
+// resolveQuietToggle — pre-route keyword check. Returns (reply, true) when
+// the utterance is a quiet-on/off command; ("", false) otherwise. Called from
+// runTurn BEFORE the router so a classifier miscue can't drop it — which means
+// both the voice path and the text path (mavweb /api/chat, telegram) reach it,
+// so a false positive here is a network-reachable way to flip a daemon-wide
+// setting. See classifyQuietToggle for the matching rule.
+func (h *reactiveHandler) resolveQuietToggle(ctx context.Context, text string) (string, bool) {
+	on, off := classifyQuietToggle(text)
+	if !on && !off {
+		return "", false
+	}
+	val := "false"
+	reply := "тихий режим выключен."
+	if on {
+		val = "true"
+		reply = "тихий режим включён. буду реже напоминать."
+	}
+	if _, err := h.api.WriteFact(ctx, ipc.WriteFactReq{
+		Ts:         h.now(),
+		Kind:       "config",
+		Key:        "quiet_hours",
+		Value:      val,
+		Source:     "tap:voice",
+		Confidence: 1.0,
+	}); err != nil {
+		log.Printf("voice: write quiet_hours: %v", err)
+		return "не получилось переключить тихий режим.", true
+	}
+	return reply, true
+}
+
+// quietInflections — the inflectional endings a stem may carry and still be
+// the same word. Adjective/adverb/noun/verb endings, all ≤3 letters. This is
+// what separates "тихий"/"тихом"/"тихо" (stem "тих" + a real ending) from
+// "тихонько"/"потихоньку", which are different words: "онько" is not an
+// ending, and "потихоньку" doesn't start with the stem at all.
+var quietInflections = []string{
+	"", "а", "е", "и", "й", "о", "у", "ы", "ю", "я",
+	"ая", "ее", "ей", "ем", "ие", "ий", "им", "их", "ия", "ию", "ое", "ой", "ом", "ую", "ые", "ый", "ым", "ых", "ья",
+	"ами", "ого", "ому", "ыми", "ать", "ить", "ять",
+}
+
+// quietStem reports whether tok is the given stem carrying at most one
+// inflectional ending. Word boundaries come from tokenisation (see
+// quietTokens), not from a regexp — Go's \b is ASCII-oriented and treats every
+// Cyrillic letter as a non-word character, so `\bтих\b` would happily match
+// inside "тихонько". Comparing whole tokens sidesteps that entirely.
+func quietStem(tok, stem string) bool {
+	if !strings.HasPrefix(tok, stem) {
+		return false
+	}
+	suffix := tok[len(stem):]
+	for _, e := range quietInflections {
+		if suffix == e {
+			return true
+		}
+	}
+	return false
+}
+
+// quietTokens splits an utterance into lowercase word tokens, dropping
+// punctuation and spacing. Unicode-aware, so Cyrillic words tokenise the same
+// way ASCII ones do.
+func quietTokens(text string) []string {
+	return strings.FieldsFunc(strings.ToLower(strings.TrimSpace(text)), func(r rune) bool {
+		return !unicode.IsLetter(r) && !unicode.IsDigit(r)
+	})
+}
+
+// quietPhrase matches a pattern (a sequence of stems) against the token list.
+// Multi-word patterns match any contiguous run of tokens — "включи тихий
+// режим" carries "тихий режим". Single-word patterns match ONLY when they are
+// the whole utterance: bare "тихо" is a command, but "в комнате тихо" is a
+// remark about the room and must not flip a daemon-wide setting.
+func quietPhrase(tokens, pattern []string) bool {
+	if len(pattern) == 0 || len(tokens) < len(pattern) {
+		return false
+	}
+	if len(pattern) == 1 {
+		return len(tokens) == 1 && quietStem(tokens[0], pattern[0])
+	}
+	for i := 0; i+len(pattern) <= len(tokens); i++ {
+		hit := true
+		for j, stem := range pattern {
+			if !quietStem(tokens[i+j], stem) {
+				hit = false
+				break
+			}
+		}
+		if hit {
+			return true
+		}
+	}
+	return false
+}
+
+// quietOffPhrases / quietOnPhrases — the toggle vocabulary, as stem sequences.
+var (
+	quietOffPhrases = [][]string{
+		{"quiet", "off"}, {"quiet", "end"},
+		{"громк", "режим"}, {"шумн", "режим"},
+		{"отмен", "тих"}, {"выключ", "тих"}, {"не", "тих"},
+	}
+	quietOnPhrases = [][]string{
+		{"quiet", "on"}, {"quiet", "mode"},
+		{"тих", "режим"}, {"не", "шум"}, {"не", "беспоко"},
+		{"тих"},
+	}
+)
+
+// classifyQuietToggle reads an utterance as a quiet-mode command. OFF is
+// resolved before ON for the same reason classifyConfirm checks negatives
+// first: the OFF phrases are built out of the ON words ("выключи тихий"
+// contains "тихий"), so scanning ON first would shadow them and "выключи
+// тихий режим" would turn quiet mode on. Negation wins.
+func classifyQuietToggle(text string) (on, off bool) {
+	tokens := quietTokens(text)
+	for _, p := range quietOffPhrases {
+		if quietPhrase(tokens, p) {
+			return false, true
+		}
+	}
+	for _, p := range quietOnPhrases {
+		if quietPhrase(tokens, p) {
+			return true, false
+		}
+	}
+	return false, false
+}

cmd/mavend/quiet_toggle_test.go

@@ -0,0 +1,114 @@
+package main
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+)
+
+// quietFakeAPI records the WriteFact the toggle performs.
+type quietFakeAPI struct {
+	ipc.UnimplementedCoreAPI
+	got  ipc.WriteFactReq
+	call int
+}
+
+func (a *quietFakeAPI) WriteFact(_ context.Context, req ipc.WriteFactReq) (int64, error) {
+	a.got, a.call = req, a.call+1
+	return 1, nil
+}
+
+// quietVerdict — what a phrase should do to the setting.
+type quietVerdict int
+
+const (
+	quietNone quietVerdict = iota
+	quietOn
+	quietOff
+)
+
+func TestResolveQuietToggle(t *testing.T) {
+	cases := []struct {
+		text string
+		want quietVerdict
+	}{
+		// ON vocabulary.
+		{"quiet on", quietOn},
+		{"quiet mode", quietOn},
+		{"тихий режим", quietOn},
+		{"тихий", quietOn},
+		{"не шуми", quietOn},
+		{"не беспокоить", quietOn},
+		{"тихо", quietOn},
+		// ON, inflected / embedded in a sentence.
+		{"включи тихий режим", quietOn},
+		{"побудь в тихом режиме", quietOn},
+		{"Тихий Режим!", quietOn},
+		{"тихая", quietOn},
+
+		// OFF vocabulary — all seven, incl. the three that used to say ON.
+		{"quiet off", quietOff},
+		{"quiet end", quietOff},
+		{"громкий режим", quietOff},
+		{"шумный режим", quietOff},
+		{"отмени тихий", quietOff},
+		{"выключи тихий", quietOff},
+		{"не тихо", quietOff},
+		// OFF wins over the ON words it contains.
+		{"выключи тихий режим", quietOff},
+		{"отмени тихий режим пожалуйста", quietOff},
+		{"верни громкий режим", quietOff},
+
+		// False positives: "тихо"/"тихий" as ordinary Russian.
+		{"очень тихий сегодня день", quietNone},
+		{"в комнате тихо", quietNone},
+		{"тихонько напомни", quietNone},
+		{"потихоньку", quietNone},
+		{"тихонько", quietNone},
+		{"он говорил тихим голосом весь вечер", quietNone},
+
+		// Unrelated.
+		{"напомни завтра позвонить маме", quietNone},
+		{"какая погода", quietNone},
+		{"", quietNone},
+	}
+
+	for _, tc := range cases {
+		t.Run(tc.text, func(t *testing.T) {
+			api := &quietFakeAPI{}
+			h := &reactiveHandler{api: api, now: func() time.Time { return time.Unix(0, 0).UTC() }}
+			reply, handled := h.resolveQuietToggle(context.Background(), tc.text)
+
+			if tc.want == quietNone {
+				if handled || reply != "" {
+					t.Fatalf("%q: got (%q, %v), want no match", tc.text, reply, handled)
+				}
+				if api.call != 0 {
+					t.Fatalf("%q: wrote a fact on a non-match", tc.text)
+				}
+				return
+			}
+			if !handled {
+				t.Fatalf("%q: not handled, want %v", tc.text, tc.want)
+			}
+			wantReply, wantVal := "тихий режим выключен.", "false"
+			if tc.want == quietOn {
+				wantReply, wantVal = "тихий режим включён. буду реже напоминать.", "true"
+			}
+			if reply != wantReply {
+				t.Errorf("%q: reply = %q, want %q", tc.text, reply, wantReply)
+			}
+			if api.call != 1 {
+				t.Fatalf("%q: WriteFact called %d times, want 1", tc.text, api.call)
+			}
+			if api.got.Kind != "config" || api.got.Key != "quiet_hours" || api.got.Source != "tap:voice" || api.got.Confidence != 1.0 {
+				t.Errorf("%q: request shape = %+v", tc.text, api.got)
+			}
+			if api.got.Value != wantVal {
+				t.Errorf("%q: value = %q, want %q", tc.text, api.got.Value, wantVal)
+			}
+		})
+	}
+}

cmd/mavend/recall.go

@@ -2,24 +2,24 @@ package main
 
 import "github.com/kami/maven/internal/memory"
 
-// bestRecall is the read side of the long-term memory store: the top hit's
+// bestRecall is the read side of the long-term memory store: the top hit when
-// stored text when it clears the confidence gate. This recalls across BOTH
+// it clears the confidence gate. The index holds BOTH notes and facts, and
-// notes and facts (facts aren't in the notes table, so this is the only path
+// either can win — the caller looks at the returned hit's meta["type"] to see
-// that can answer "when did I last …?" from a captured fact). A note hit here
+// which. Facts aren't in the notes table, so this is the only path that can
-// is redundant with the notes-RAG path — by design; the two indexes can diverge
+// answer "when did I last …?" from a captured fact.
-// once the backend is swapped for a persistent/external store. ok=false when
+//
-// there's no hit above the threshold or the hit carries no text.
+// The whole hit is returned, not just its text, because "which memory answered"
-func bestRecall(results []memory.Result, min float64) (string, bool) {
+// decides how the answer is said: a note gets phrased in Maven's voice, a fact
-	if len(results) == 0 {
+// is read back as stored.
-		return "", false
+//
+// ok=false when the hit fails the confidence gate (see memory.Confident: an
+// absolute floor plus a margin over the runner-up) or carries no text.
+func bestRecall(results []memory.Result, minScore, minMargin float64) (memory.Result, bool) {
+	if !memory.Confident(results, minScore, minMargin) {
+		return memory.Result{}, false
 	}
-	top := results[0]
+	if results[0].Meta["text"] == "" {
-	if top.Score < min {
+		return memory.Result{}, false
-		return "", false
 	}
-	text := top.Meta["text"]
+	return results[0], true
-	if text == "" {
-		return "", false
-	}
-	return text, true
 }

cmd/mavend/recall_test.go

@@ -8,23 +8,24 @@ import (
 
 func TestBestRecall(t *testing.T) {
 	const min = 0.55
+	const margin = 0.008
 
 	t.Run("empty results", func(t *testing.T) {
-		if _, ok := bestRecall(nil, min); ok {
+		if _, ok := bestRecall(nil, min, margin); ok {
 			t.Error("empty results returned ok")
 		}
 	})
 
 	t.Run("top below threshold", func(t *testing.T) {
 		res := []memory.Result{{Score: 0.4, Meta: map[string]string{"text": "выпил воды"}}}
-		if _, ok := bestRecall(res, min); ok {
+		if _, ok := bestRecall(res, min, margin); ok {
 			t.Error("below-threshold hit returned ok")
 		}
 	})
 
 	t.Run("hit without text meta", func(t *testing.T) {
 		res := []memory.Result{{Score: 0.9, Meta: map[string]string{"type": "fact"}}}
-		if _, ok := bestRecall(res, min); ok {
+		if _, ok := bestRecall(res, min, margin); ok {
 			t.Error("textless hit returned ok")
 		}
 	})
@@ -34,12 +35,43 @@ func TestBestRecall(t *testing.T) {
 			{Score: 0.82, Meta: map[string]string{"text": "выпил воды в три часа", "type": "fact"}},
 			{Score: 0.60, Meta: map[string]string{"text": "другое"}},
 		}
-		got, ok := bestRecall(res, min)
+		got, ok := bestRecall(res, min, margin)
 		if !ok {
 			t.Fatal("clearing hit not returned")
 		}
-		if got != "выпил воды в три часа" {
+		if got.Meta["text"] != "выпил воды в три часа" {
-			t.Errorf("wrong text: %q", got)
+			t.Errorf("wrong text: %q", got.Meta["text"])
+		}
+		if got.Meta["type"] != "fact" {
+			t.Errorf("kind lost: %q", got.Meta["type"])
+		}
+	})
+
+	// The index holds notes and facts together, so a note has to be able to win
+	// it — for a long time it could not (Vikunja #373).
+	t.Run("a note can win", func(t *testing.T) {
+		res := []memory.Result{
+			{Score: 0.86, Meta: map[string]string{"text": "молоко в холодильнике", "type": "note"}},
+			{Score: 0.61, Meta: map[string]string{"text": "выпил воды", "type": "fact"}},
+		}
+		got, ok := bestRecall(res, min, margin)
+		if !ok {
+			t.Fatal("clearly-best note not returned")
+		}
+		if got.Meta["type"] != "note" || got.Meta["text"] != "молоко в холодильнике" {
+			t.Errorf("got %v, want the note", got.Meta)
+		}
+	})
+
+	// The runner-up is almost as close, so the embedder cannot tell the two
+	// notes apart. Silence beats reading back a coin flip.
+	t.Run("runner-up too close", func(t *testing.T) {
+		res := []memory.Result{
+			{Score: 0.860, Meta: map[string]string{"text": "выпил воды в три часа"}},
+			{Score: 0.858, Meta: map[string]string{"text": "другое"}},
+		}
+		if _, ok := bestRecall(res, min, margin); ok {
+			t.Error("thin-margin hit returned ok")
 		}
 	})
 }

cmd/mavend/replier_llm.go

@@ -7,6 +7,7 @@ import (
 	"time"
 
 	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/persona"
 	"github.com/kami/maven/internal/router"
 	"github.com/kami/maven/internal/voice"
 )
@@ -23,13 +24,19 @@ type completer interface {
 type llmReplier struct {
 	c    completer
 	stub *voice.StubReplier
+
+	// block renders the shared context block per turn (who he is, the time).
+	// nil ⇒ the prompt stands alone.
+	block func() string
 }
 
-func newLLMReplier(c completer) *llmReplier {
+func newLLMReplier(c completer, block func() string) *llmReplier {
-	return &llmReplier{c: c, stub: voice.NewStubReplier()}
+	return &llmReplier{c: c, stub: voice.NewStubReplier(), block: block}
 }
 
-const replySystem = `Ты — Maven, домашняя ассистентка (о себе — в женском роде). Подтверди действие РОВНО ОДНИМ коротким предложением (≤120 символов), тепло и по-русски. Не задавай вопросов, не повторяй слова, не добавляй ничего после точки. Respond ONLY with valid JSON: {"response": "...", "mood": "neutral"}.`
+const replySystem = `Ты — Maven, домашняя ассистентка (о себе — в женском роде). Владелец — мужчина, говоришь с ним на "ты", в единственном числе; никогда не "вы"/"ваш" и не "он"/"его". Подтверди действие РОВНО ОДНИМ коротким предложением (≤120 символов), по-русски, спокойно и без официальных формулировок. Не задавай вопросов, не повторяй слова, не добавляй ничего после точки. Отвечай ТОЛЬКО одним объектом JSON с полями "response" (текст) и "mood" (ровно одно из: neutral, happy, thinking, tired, confused).
+Пример: {"response": "Записала, что ты выпил стакан воды.", "mood": "neutral"}
+Никогда не пиши "..." в поле response.`
 
 func (r *llmReplier) Reply(d router.Decision) string {
 	if d.Clarify {
@@ -38,7 +45,7 @@ func (r *llmReplier) Reply(d router.Decision) string {
 	ctx, cancel := context.WithTimeout(context.Background(), 60*time.Second)
 	defer cancel()
 	out, err := r.c.Complete(ctx, llm.Req{
-		System:        replySystem,
+		System:        persona.Prepend(r.block, replySystem),
 		User:          replyContext(d),
 		MaxTokens:     512,
 		RepeatPenalty: 1.3,

cmd/mavend/replier_llm_test.go

@@ -17,7 +17,7 @@ type mockCompleter struct {
 func (m mockCompleter) Complete(_ context.Context, _ llm.Req) (string, error) { return m.out, m.err }
 
 func TestLLMReplierReturnsLLMReply(t *testing.T) {
-	r := newLLMReplier(mockCompleter{out: `{"response":"записала, кофе закончился","mood":"neutral"}`})
+	r := newLLMReplier(mockCompleter{out: `{"response":"записала, кофе закончился","mood":"neutral"}`}, nil)
 	got := r.Reply(router.Decision{Intent: router.IntentNote, Slots: router.Slots{Text: "кофе закончился"}})
 	if got != "записала, кофе закончился" {
 		t.Errorf("got %q, want %q", got, "записала, кофе закончился")
@@ -25,7 +25,7 @@ func TestLLMReplierReturnsLLMReply(t *testing.T) {
 }
 
 func TestLLMReplierFallsBackToPlainText(t *testing.T) {
-	r := newLLMReplier(mockCompleter{out: "записала, кофе закончился"})
+	r := newLLMReplier(mockCompleter{out: "записала, кофе закончился"}, nil)
 	got := r.Reply(router.Decision{Intent: router.IntentNote, Slots: router.Slots{Text: "кофе закончился"}})
 	if got != "записала, кофе закончился" {
 		t.Errorf("got %q, want %q", got, "записала, кофе закончился")
@@ -33,7 +33,7 @@ func TestLLMReplierFallsBackToPlainText(t *testing.T) {
 }
 
 func TestLLMReplierFallsBackToStubOnError(t *testing.T) {
-	r := newLLMReplier(mockCompleter{err: errTestLLMDown})
+	r := newLLMReplier(mockCompleter{err: errTestLLMDown}, nil)
 	noteDec := router.Decision{Intent: router.IntentNote}
 	got := r.Reply(noteDec)
 	want := voice.NewStubReplier().Reply(noteDec)
@@ -43,7 +43,7 @@ func TestLLMReplierFallsBackToStubOnError(t *testing.T) {
 }
 
 func TestLLMReplierFallsBackToStubOnEmpty(t *testing.T) {
-	r := newLLMReplier(mockCompleter{out: ""})
+	r := newLLMReplier(mockCompleter{out: ""}, nil)
 	noteDec := router.Decision{Intent: router.IntentNote}
 	got := r.Reply(noteDec)
 	want := voice.NewStubReplier().Reply(noteDec)
@@ -53,7 +53,7 @@ func TestLLMReplierFallsBackToStubOnEmpty(t *testing.T) {
 }
 
 func TestLLMReplierClarifyUsesStub(t *testing.T) {
-	r := newLLMReplier(mockCompleter{out: "я всё поняла"})
+	r := newLLMReplier(mockCompleter{out: "я всё поняла"}, nil)
 	clarifyDec := router.Decision{Clarify: true}
 	got := r.Reply(clarifyDec)
 	want := voice.NewStubReplier().Reply(clarifyDec)

cmd/mavend/ruwords.go

@@ -0,0 +1,180 @@
+// Package main — ruwords.go holds Russian language + calendar/time formatting
+// helpers used by the voice reply paths (replySystem, the reminder/routine
+// phrasing, etc). Pure functions, no receivers: weekday/month name tables,
+// plural agreement, clock/date rendering, and the "do I actually know this
+// place/day" guards that pick an honest reply over a confidently wrong one.
+// Extend this file rather than voice.go for anything in that shape.
+package main
+
+import (
+	"fmt"
+	"strconv"
+	"strings"
+	"time"
+)
+
+var ruWeekdays = []string{
+	"воскресенье", "понедельник", "вторник", "среда",
+	"четверг", "пятница", "суббота",
+}
+
+var ruMonths = []string{
+	"января", "февраля", "марта", "апреля", "мая", "июня",
+	"июля", "августа", "сентября", "октября", "ноября", "декабря",
+}
+
+// onlyLocalTimeReply — the honest answer when the user asks the time somewhere
+// other than here. She only keeps one clock, and saying so is better than
+// naming the wrong city's time.
+//
+// There used to be a city→time-zone table here. It was removed on purpose: the
+// user only ever asks for local time, so the table was a second list of cities
+// to keep in step with the weather one for no gain.
+const onlyLocalTimeReply = "я знаю только местное время, про другие города пока не скажу."
+
+// notPlaceAfterV — words that follow "в" without naming a place, so
+// mentionsUnknownPlace does not mistake them for a city.
+var notPlaceAfterV = map[string]bool{
+	"данный": true, "данную": true, "этот": true, "эту": true,
+	"котором": true, "какое": true, "какой": true, "который": true,
+	"общем": true, "точности": true, "курсе": true, "сутках": true,
+	"часах": true, "минутах": true, "секундах": true, "неделе": true,
+}
+
+// mentionsUnknownPlace reports whether the question has a "в <слово>" phrase
+// that looks like a place we do not know ("который час в киеве"). Used only to
+// pick the honest "local time only" reply instead of answering local time as
+// if it were the city's.
+func mentionsUnknownPlace(u string) bool {
+	toks := strings.Fields(u)
+	for i := 0; i+1 < len(toks); i++ {
+		if toks[i] != "в" && toks[i] != "во" {
+			continue
+		}
+		next := strings.Trim(toks[i+1], ".,?!")
+		if next == "" || notPlaceAfterV[next] {
+			continue
+		}
+		// A number after "в" is a clock ("в 5 часов"), not a place.
+		if _, err := strconv.Atoi(strings.SplitN(next, ":", 2)[0]); err == nil {
+			continue
+		}
+		return true
+	}
+	return false
+}
+
+// onlyNearDaysReply — she can work out today, tomorrow, the day after and
+// yesterday, and nothing further. Said out loud instead of answering today's
+// date for a day she did not understand.
+const onlyNearDaysReply = "я считаю только сегодня, завтра, послезавтра и вчера — про другие дни пока не скажу."
+
+// dayWords — day references the calendar parser cannot resolve. A weekday name
+// or a "через …" phrase means he asked about a specific other day.
+var dayWords = []string{
+	"понедельник", "вторник", "сред", "четверг", "пятниц", "суббот", "воскресен",
+	"через", "monday", "tuesday", "wednesday", "thursday", "friday", "saturday", "sunday",
+}
+
+// mentionsUnknownDay reports whether the question names a day the calendar
+// parser could not resolve. Mirror of mentionsUnknownPlace: it exists only to
+// pick an honest reply over a confidently wrong one.
+//
+// Only called after ParseCalendarDate has already failed, so "завтра" and the
+// other words it does know never reach here.
+func mentionsUnknownDay(u string) bool {
+	for _, w := range dayWords {
+		if strings.Contains(u, w) {
+			return true
+		}
+	}
+	return false
+}
+
+// ruClock renders the clock part of the time reply: "15 часов 4 минуты".
+func ruClock(t time.Time) string {
+	h, m := t.Hour(), t.Minute()
+	hourWord := ruPlural(h, "час", "часа", "часов")
+	if m == 0 {
+		return fmt.Sprintf("%d %s ровно", h, hourWord)
+	}
+	return fmt.Sprintf("%d %s %d %s", h, hourWord, m, ruPlural(m, "минута", "минуты", "минут"))
+}
+
+// dayPrefix names the day relative to now ("завтра", "вчера", …) so the date
+// reply opens the way a person would say it.
+func dayPrefix(now, day time.Time) string {
+	base := time.Date(now.Year(), now.Month(), now.Day(), 0, 0, 0, 0, now.Location())
+	switch int(day.Sub(base).Hours() / 24) {
+	case -1:
+		return "вчера"
+	case 0:
+		return "сегодня"
+	case 1:
+		return "завтра"
+	case 2:
+		return "послезавтра"
+	}
+	return "это"
+}
+
+func ruPlural(n int, one, two, many string) string {
+	n = n % 100
+	if n > 10 && n < 20 {
+		return many
+	}
+	n = n % 10
+	switch n {
+	case 1:
+		return one
+	case 2, 3, 4:
+		return two
+	default:
+		return many
+	}
+}
+
+// hasDurationWords checks whether u is asking about elapsed/remaining time
+// rather than the current clock — guards replySystem from replying "сейчас
+// X часов" to "сколько времени прошло". Mirrors the stage0.go build filter.
+func hasDurationWords(u string) bool {
+	s := strings.ToLower(strings.TrimSpace(u))
+	// First-word duration markers (same keywords as timeQueryBuild in stage0).
+	first := strings.Fields(s)
+	if len(first) > 0 {
+		switch first[0] {
+		case "прошло", "осталось", "пройдет", "минуло", "проходит":
+			return true
+		}
+	}
+	// Broader duration keywords appearing anywhere in the utterance.
+	if strings.Contains(s, "прошло") || strings.Contains(s, "осталось") {
+		return true
+	}
+	if strings.Contains(s, " до ") {
+		return true
+	}
+	return false
+}
+
+// formatTime returns a human-readable Russian time string for a fact timestamp.
+// Used by the query handler when answering "когда я это сделал?"-style questions.
+func formatTime(t time.Time) string {
+	now := time.Now()
+	if t.After(now.Add(-2*time.Minute)) && t.Before(now.Add(2*time.Minute)) {
+		return "только что"
+	}
+	diff := now.Sub(t)
+	switch {
+	case diff < 10*time.Minute:
+		return "несколько минут назад"
+	case diff < 60*time.Minute:
+		return fmt.Sprintf("%d минут назад", int(diff.Minutes()))
+	case diff < 2*time.Hour:
+		return "час назад"
+	case diff < 24*time.Hour:
+		return fmt.Sprintf("%d часа назад", int(diff.Hours()))
+	default:
+		return t.Format("2 января 15:04")
+	}
+}

cmd/mavend/strutil.go

@@ -0,0 +1,85 @@
+// Package main — strutil.go holds small, receiver-free string utilities used
+// across the voice reply paths: trimming a wake token, pulling out the first
+// word or first line, and a minimal JSON string encoder for the one payload
+// shape that needs it. Extend this file rather than voice.go for anything in
+// that shape.
+package main
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// stripWake removes a leading wake token (any script the STT phonetically
+// transcribes "Maven" as) so the verb is the first word.
+func stripWake(u string) string {
+	stripped, had := router.StripWakeToken(u)
+	if !had {
+		return strings.TrimSpace(u)
+	}
+	return stripped
+}
+
+// firstWord returns the first whitespace-delimited token (lowercased) — the
+// proposed tool's name.
+func firstWord(s string) string {
+	f := strings.Fields(s)
+	if len(f) == 0 {
+		return ""
+	}
+	return strings.ToLower(f[0])
+}
+
+// firstLine — the first non-empty line of a tool's output, for a short spoken
+// reply (the full output goes to the log, not the TTS). Trimmed to keep the
+// utterance sane if a command dumps a wall of text.
+func firstLine(s string) string {
+	for _, line := range strings.Split(s, "\n") {
+		line = strings.TrimSpace(line)
+		if line != "" {
+			if len(line) > 200 {
+				line = line[:200]
+			}
+			return line
+		}
+	}
+	return ""
+}
+
+// jsonString — a one-line JSON string encoder without dragging encoding/json
+// into the top of this file. Used to wrap a reminder payload's text field;
+// the router's reminder Slots are already absolute (DateTimeParser resolved
+// relative→absolute), the payload shape is conventional {"text":...}.
+func jsonString(s string) string {
+	// minimal JSON string escape — quotes + backslash + control chars.
+	// adequate for the reminder payload's text field; not a general JSON
+	// encoder. The chroma / RAG modules (when they land) use a real json
+	// encoder for richer payloads. Keep it inline here so the import
+	// direction stays narrow.
+	var b []byte
+	b = append(b, '"')
+	for _, r := range s {
+		switch r {
+		case '"':
+			b = append(b, '\\', '"')
+		case '\\':
+			b = append(b, '\\', '\\')
+		case '\n':
+			b = append(b, '\\', 'n')
+		case '\r':
+			b = append(b, '\\', 'r')
+		case '\t':
+			b = append(b, '\\', 't')
+		default:
+			if r < 0x20 {
+				b = append(b, []byte(fmt.Sprintf("\\u%04x", r))...)
+			} else {
+				b = append(b, []byte(string(r))...)
+			}
+		}
+	}
+	b = append(b, '"')
+	return string(b)
+}

cmd/mavend/system_reply_test.go

@@ -0,0 +1,78 @@
+package main
+
+import (
+	"context"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/router"
+)
+
+// systemHandler — a handler with nothing but a fixed clock, which is all
+// replySystem needs.
+func systemHandler(now time.Time) *reactiveHandler {
+	return &reactiveHandler{now: func() time.Time { return now }}
+}
+
+// TestReplySystemDateOffset — "какое число завтра" must answer tomorrow's
+// date, not today's (Vikunja #388).
+func TestReplySystemDateOffset(t *testing.T) {
+	// Thursday, 30 July 2026.
+	now := time.Date(2026, 7, 30, 14, 5, 0, 0, time.UTC)
+	h := systemHandler(now)
+	cases := []struct{ utterance, want string }{
+		{"какое сегодня число", "сегодня четверг, 30 июля 2026 года"},
+		{"какое число", "сегодня четверг, 30 июля 2026 года"},
+		{"какое число завтра", "завтра пятница, 31 июля 2026 года"},
+		{"какое число послезавтра", "послезавтра суббота, 1 августа 2026 года"},
+		{"какое было число вчера", "вчера среда, 29 июля 2026 года"},
+	}
+	for _, c := range cases {
+		got := h.replySystem(context.Background(), router.Decision{Utterance: c.utterance})
+		if got != c.want {
+			t.Errorf("replySystem(%q) = %q, want %q", c.utterance, got, c.want)
+		}
+	}
+}
+
+// A day she cannot work out must not come back as today's date — that is the
+// same silent wrong answer #388 was about, one step further out.
+func TestReplySystemUnknownDayIsHonest(t *testing.T) {
+	now := time.Date(2026, 7, 30, 14, 5, 0, 0, time.UTC)
+	h := systemHandler(now)
+	for _, u := range []string{
+		"какое число в пятницу",
+		"какое число через неделю",
+		"какое число в понедельник",
+	} {
+		got := h.replySystem(context.Background(), router.Decision{Utterance: u})
+		if got != onlyNearDaysReply {
+			t.Errorf("replySystem(%q) = %q, want the honest reply", u, got)
+		}
+	}
+	// The days she does know must not be caught by the same guard.
+	if got := h.replySystem(context.Background(), router.Decision{Utterance: "какое число завтра"}); got == onlyNearDaysReply {
+		t.Error("завтра was treated as an unknown day")
+	}
+}
+
+// TestReplySystemClockCity — the clock arm must not answer local time for a
+// question about another city (Vikunja #388). She keeps one clock, so every
+// named place gets the honest "local time only" answer.
+func TestReplySystemClockCity(t *testing.T) {
+	now := time.Date(2026, 7, 30, 12, 0, 0, 0, time.UTC)
+	h := systemHandler(now)
+	cases := []struct{ utterance, want string }{
+		{"который час", "сейчас 12 часов ровно"},
+		{"который час в киеве", onlyLocalTimeReply},
+		{"сколько времени в москве", onlyLocalTimeReply},
+		{"который час в лондоне", onlyLocalTimeReply},
+		{"который час в бишкеке", onlyLocalTimeReply},
+	}
+	for _, c := range cases {
+		got := h.replySystem(context.Background(), router.Decision{Utterance: c.utterance})
+		if got != c.want {
+			t.Errorf("replySystem(%q) = %q, want %q", c.utterance, got, c.want)
+		}
+	}
+}

cmd/mavend/tick.go

@@ -24,6 +24,7 @@ import (
 	"github.com/kami/maven/internal/ipc"
 	"github.com/kami/maven/internal/loop"
 	"github.com/kami/maven/internal/morning"
+	"github.com/kami/maven/internal/pattern"
 	"github.com/kami/maven/internal/phraser"
 	"github.com/kami/maven/internal/routine"
 	"github.com/kami/maven/internal/store"
@@ -67,6 +68,14 @@ type tickLoop struct {
 	morningRoutines []morning.Routine
 	morningLast     map[string]time.Time
 
+	// proposalCfg — announcement policy for routines the tick inferred itself.
+	// nil ⇒ detect silently, never announce (the default). lastProposalAt is
+	// the cooldown clock, in-memory on purpose: a restart is allowed to permit
+	// one more announcement, and a restart-per-day loop is a bigger problem
+	// than a duplicate proposal notice.
+	proposalCfg    *config.PatternProposalConfig
+	lastProposalAt time.Time
+
 	// digestQ — in-memory queue of eligible nudges waiting for batch flush.
 	// populated when digestCfg != nil && digestCfg.Enabled.
 	digestQ []QueuedNudge
@@ -92,6 +101,7 @@ func newTickLoop(
 	digestCfg *config.DigestConfig,
 	routines []routine.Routine,
 	morningRoutines []morning.Routine,
+	proposalCfg *config.PatternProposalConfig,
 ) *tickLoop {
 	return &tickLoop{
 		store:            st,
@@ -108,6 +118,7 @@ func newTickLoop(
 		routineLast:      make(map[string]time.Time),
 		morningRoutines:  morningRoutines,
 		morningLast:      make(map[string]time.Time),
+		proposalCfg:      proposalCfg,
 		lastPhrase:       make(map[string]delivery.PhrasedNudge),
 	}
 }
@@ -180,6 +191,17 @@ func (t *tickLoop) tick(ctx context.Context, now time.Time) {
 	// (with dedup) avoids re-queueing the same rule after a flush.
 	t.maybeFlush(ctx, now, state)
 
+	// gate-suppressed digest (Vikunja #281): rules the restraint gate held
+	// back this tick (quiet hours / away / calendar-busy), not because they
+	// weren't due, but because it wasn't the moment. Some of those are worth
+	// resurfacing later instead of just being lost — loop.DigestEligible
+	// draws that line. This is a SEPARATE mechanism from the in-memory
+	// digestQ above: that one batches candidates the gate already ALLOWED to
+	// fire; this one durably holds candidates the gate BLOCKED.
+	t.enqueueSuppressedDigest(ctx, trace, state, now)
+	t.expireStaleDigest(ctx, now)
+	t.maybeDrainDigest(ctx, state, now)
+
 	// routines: operator-declared scheduled behaviors. fire the ones whose cron
 	// crossed since last fire, delivered through the normal routing (voice when
 	// present, away channels otherwise). bodies are literal operator text — not
@@ -195,6 +217,14 @@ func (t *tickLoop) tick(ctx context.Context, now time.Time) {
 	// nudge time. See internal/morning for the "why not four timers" rationale.
 	t.fireMorningRoutines(ctx, now, state)
 
+	// pattern detection: scan every action+object pair with recorded events
+	// and propose a routine for any stable one not already decided (Vikunja
+	// #43). This used to only run as a side effect of the voice fact-write
+	// path, so a pattern already sitting in history went unnoticed until he
+	// happened to mention it again by voice. See patterns.go and
+	// detectPatterns below for how idempotence and dismissal are respected.
+	t.detectPatterns(ctx, now, state)
+
 	// reminders: gate-bypassing class. fired once, marked after a successful
 	// delivery.  a failed send leaves the reminder pending — the next tick
 	// re-gathers and re-attempts.
@@ -342,6 +372,235 @@ func (t *tickLoop) flushDigest(ctx context.Context, now time.Time, state loop.St
 	t.digestQ = nil
 }
 
+// detectPatterns runs the pattern detector proactively over every
+// action+object pair that has ever produced an event, independent of
+// whichever fact write (or channel) last touched it (Vikunja #43). This is
+// what makes pattern inference actually proactive: it fires on the daemon's
+// own schedule reading accumulated history, not only as a side effect of a
+// live voice turn.
+//
+// Idempotence and noise are handled by the store, not here — this function
+// is safe to call every tick:
+//   - Same pattern, tick after tick: detectAndPropose's LookupProposedRoutine
+//     check plus proposed_routines' UNIQUE(action, object) constraint (with
+//     CreateProposedRoutine's ON CONFLICT DO NOTHING) mean a pair that
+//     already has a row — in ANY status — produces no second row and no log
+//     spam beyond the one line at genuine creation.
+//   - A DISMISSED proposal must never come back. DismissProposedRoutine flips
+//     status in place; the row is never deleted. So the same Lookup check
+//     that stops a duplicate "proposed" also stops a "dismissed" one from
+//     resurrecting — there is nothing tick-specific to get right here beyond
+//     calling the same shared path the voice route already used.
+//
+// By default this only creates a row for the /routines page to show: it does
+// not notify, ring, or speak. Detection is not the same act as disturbing him
+// about it, and Maven is "not a nag, not autonomous" (CLAUDE.md). Announcing
+// is opt-in through the pattern_proposals config block — see announceProposal
+// for the restraints that apply even then. A proposal only starts producing
+// recurring nudges once he accepts it (fireAcceptedRoutines).
+func (t *tickLoop) detectPatterns(ctx context.Context, now time.Time, state loop.State) {
+	pairs, err := t.store.DistinctEventPairs(ctx)
+	if err != nil {
+		log.Printf("tick: distinct event pairs: %v", err)
+		return
+	}
+	announced := false
+	for _, p := range pairs {
+		r, _, err := detectAndPropose(ctx, t.store, p.Action, p.Object, now)
+		if err != nil {
+			log.Printf("tick: detect pattern %s/%s: %v", p.Action, p.Object, err)
+			continue
+		}
+		if r == nil {
+			continue // no stable pattern, or already proposed/accepted/dismissed
+		}
+		log.Printf("tick: proposed routine: %s/%s every %.1f days", r.Action, r.Object, r.IntervalDays)
+		// One announcement per tick at most, whatever the scan turned up. The
+		// rest are on /routines; they are not lost, they are just not shouted.
+		if announced {
+			continue
+		}
+		announced = t.announceProposal(ctx, r, now, state)
+	}
+}
+
+// announceProposal offers a freshly inferred routine through the ordinary
+// care-delivery path, if announcing is switched on at all. Returns true when
+// something was actually sent.
+//
+// Everything here is restraint. The feature is off unless configured; when on
+// it is sev1 (the lowest severity, so quiet hours, away presence and snooze
+// all suppress it via loop.Gate exactly like a care nudge); it is spaced by
+// proposalCfg.Cooldown across every pair, not per pair; and a suppressed or
+// dropped announcement is NOT retried — the cooldown clock advances only on a
+// real send, but the proposal row already exists, so the next tick will not
+// re-detect it and nothing queues up behind it. A missed announcement means
+// he reads it on /routines instead, which is the whole point of the page.
+//
+// The body is the detector's own literal Russian phrasing (pattern.PhraseRoutine
+// — "ты заправляешь поилку раз в 7 дней — напоминать?"), not LLM-generated, so
+// an inferred routine cannot arrive worded as something Maven never observed.
+func (t *tickLoop) announceProposal(ctx context.Context, r *pattern.ProposedRoutine, now time.Time, state loop.State) bool {
+	if !t.proposalCfg.AnnounceProposals() {
+		return false
+	}
+	cooldown := time.Duration(t.proposalCfg.Cooldown)
+	if cooldown <= 0 {
+		cooldown = config.DefaultProposalCooldown
+	}
+	if !t.lastProposalAt.IsZero() && now.Sub(t.lastProposalAt) < cooldown {
+		return false
+	}
+
+	rule := loop.Rule{Name: "proposal:" + r.Action + " " + r.Object, Severity: loop.Sev1}
+	if !loop.Gate(state, rule) {
+		return false
+	}
+	body := pattern.PhraseRoutine(r)
+	pn := delivery.PhrasedNudge{
+		Candidate: loop.Candidate{Rule: rule, Severity: rule.Severity, State: state},
+		Body:      body,
+		Summary:   body,
+	}
+	sent, err := t.dispatcher.DispatchNudge(ctx, pn, now)
+	if err != nil {
+		log.Printf("tick: announce proposal %s/%s: %v", r.Action, r.Object, err)
+		return false
+	}
+	if len(sent) == 0 {
+		return false // routing dropped it — /routines still has it.
+	}
+	t.lastProposalAt = now
+	return true
+}
+
+// digestExpiry — how long a gate-suppressed care nudge stays worth
+// resurfacing. 24h: these are daily-cadence rules (water/meal/break run on
+// hour-scale cooldowns and re-derive from facts that reset every day), so a
+// digest entry that outlives one full day is describing a day that's already
+// over — "you skipped a break yesterday" said tomorrow evening is noise, not
+// news. Bounding at one day also means a digest can never silently span a
+// weekend of quiet hours into an unbounded backlog.
+const digestExpiry = 24 * time.Hour
+
+// maxDigestSpokenItems — the bundle read-out is capped so "batched, not
+// dropped" cannot regress into "she dumps twelve things on me the moment I
+// walk in" — a digest that nags in bulk is worse than the drops it replaced.
+// Anything beyond the cap is still marked drained (it did get its moment;
+// the cap limits WORDS, not whether it counted) and folded into a trailing
+// count instead of being spoken in full.
+const maxDigestSpokenItems = 3
+
+// enqueueSuppressedDigest scans this tick's trace for care candidates the
+// gate blocked for a genuine restraint reason and durably records the
+// digest-eligible ones (loop.DigestEligible). Phrasing happens once, here,
+// at enqueue time — not re-derived at drain time — the same way queueNudge
+// phrases once and caches, so a rule suppressed for hours isn't re-prompting
+// the LLM every tick it stays blocked (EnqueueDigestEntry's rule+body dedupe
+// makes repeat calls here harmless, but skipping the phrase call entirely
+// when a pending entry already exists avoids the LLM round-trip too).
+func (t *tickLoop) enqueueSuppressedDigest(ctx context.Context, trace *loop.TickTrace, state loop.State, now time.Time) {
+	if trace == nil {
+		return
+	}
+	for _, tr := range trace.RuleTraces {
+		if !tr.PredicateResult || tr.GateResult {
+			continue // didn't want to fire, or wasn't suppressed
+		}
+		if !loop.DigestEligible(tr.Severity, tr.GateBlockedBy) {
+			continue
+		}
+		rule := loop.Rule{Name: tr.RuleName, Severity: tr.Severity}
+		cand := loop.Candidate{Rule: rule, Severity: tr.Severity, State: state}
+		pn, err := t.phraser.PhraseNudge(ctx, cand)
+		if err != nil {
+			log.Printf("tick: phrase digest candidate %s: %v", tr.RuleName, err)
+			continue
+		}
+		expires := now.Add(digestExpiry)
+		if _, deduped, err := t.store.EnqueueDigestEntry(ctx, tr.RuleName, int(tr.Severity), pn.Body, now, expires); err != nil {
+			log.Printf("tick: enqueue digest entry %s: %v", tr.RuleName, err)
+		} else if deduped {
+			// same suppressed nudge already pending — nothing new to say.
+			continue
+		}
+	}
+}
+
+// expireStaleDigest sweeps entries past their expiry once per tick — cheap
+// bookkeeping, mirrors ReconcileStaleDeliveryAttempts's shape.
+func (t *tickLoop) expireStaleDigest(ctx context.Context, now time.Time) {
+	n, err := t.store.ExpireStaleDigestEntries(ctx, now)
+	if err != nil {
+		log.Printf("tick: expire stale digest entries: %v", err)
+		return
+	}
+	if n > 0 {
+		log.Printf("tick: expired %d stale digest entr(y/ies) unspoken", n)
+	}
+}
+
+// maybeDrainDigest speaks the pending digest bundle once the gate's
+// suppression reasons have actually cleared — quiet hours over, back from
+// away, out of the meeting. Draining while still suppressed would just be a
+// second way to nag through quiet hours; the bundle waits for the same "is
+// it allowed right now" condition a live nudge already waits for.
+func (t *tickLoop) maybeDrainDigest(ctx context.Context, state loop.State, now time.Time) {
+	if state.QuietHours || state.CalendarBusy || state.Presence == store.Away {
+		return
+	}
+	entries, err := t.store.PendingDigestEntries(ctx, now)
+	if err != nil {
+		log.Printf("tick: pending digest entries: %v", err)
+		return
+	}
+	if len(entries) == 0 {
+		return
+	}
+
+	spoken := entries
+	extra := 0
+	if len(spoken) > maxDigestSpokenItems {
+		spoken = entries[:maxDigestSpokenItems]
+		extra = len(entries) - maxDigestSpokenItems
+	}
+	var b strings.Builder
+	maxSev := 0
+	for i, e := range spoken {
+		if i > 0 {
+			b.WriteString(" · ")
+		}
+		b.WriteString(e.Body)
+		if e.Severity > maxSev {
+			maxSev = e.Severity
+		}
+	}
+	if extra > 0 {
+		fmt.Fprintf(&b, " · и ещё %d", extra)
+	}
+	body := b.String()
+	summary := fmt.Sprintf("%d отложенных уведомлений", len(entries))
+
+	cand := loop.Candidate{
+		Rule:     loop.Rule{Name: "digest", Severity: loop.Severity(maxSev)},
+		Severity: loop.Severity(maxSev),
+		State:    state,
+	}
+	pn := delivery.PhrasedNudge{Candidate: cand, Body: body, Summary: summary}
+	t.cachePhrase(pn)
+	if _, err := t.dispatcher.DispatchNudge(ctx, pn, now); err != nil {
+		log.Printf("tick: dispatch digest bundle: %v", err)
+		return // leave entries pending; retried next tick
+	}
+	ids := make([]int64, len(entries))
+	for i, e := range entries {
+		ids[i] = e.ID
+	}
+	if err := t.store.DrainDigestEntries(ctx, ids, now); err != nil {
+		log.Printf("tick: drain digest entries: %v", err)
+	}
+}
+
 // routinesFromConfig maps the config's routine blocks to the engine type.
 // Validation (cron parses, name/body present, severity defaulted) already ran
 // in config.Load, so this is a pure field copy.
@@ -530,6 +789,77 @@ func (t *tickLoop) morningStatus(ctx context.Context, now time.Time) []ipc.Morni
 	return out
 }
 
+// dayPlan is the read-only "what does today hold" query (Vikunja #128). It is
+// the impure half of morning.BuildPlan: it reads the calendar events, the
+// pending reminders and the checklist facts, and the pure builder orders them.
+//
+// It never dispatches. Asking for the plan is a query like any other; the only
+// unprompted delivery in maven stays with the morning nudge and the
+// dispatcher's policy.
+func (t *tickLoop) dayPlan(ctx context.Context, now time.Time) ipc.DayPlan {
+	y, m, d := now.Date()
+	dayStart := time.Date(y, m, d, 0, 0, 0, 0, now.Location())
+	dayEnd := dayStart.AddDate(0, 0, 1)
+
+	var events []morning.PlanEntry
+	facts, err := t.store.CalendarEvents(ctx, dayStart, dayEnd)
+	if err != nil {
+		log.Printf("tick: day plan: calendar events: %v", err)
+	}
+	for _, f := range facts {
+		events = append(events, morning.PlanEntry{
+			At:   f.Ts,
+			Text: f.Value,
+			Kind: morning.PlanEvent,
+			// Provenance below a calendar read (an ambient relay, #126) is
+			// hedged rather than recited as fact.
+			Uncertain: f.Confidence < 1.0,
+		})
+	}
+
+	var reminders []morning.PlanEntry
+	rems, err := t.store.ListReminders(ctx, dayPlanMaxReminders)
+	if err != nil {
+		log.Printf("tick: day plan: list reminders: %v", err)
+	}
+	for _, r := range rems {
+		if r.Status != "pending" {
+			continue
+		}
+		fire := r.NextFireTs
+		if fire.IsZero() {
+			fire = r.FireTs
+		}
+		reminders = append(reminders, morning.PlanEntry{
+			At:   fire,
+			Text: strings.TrimSpace(r.Payload),
+			Kind: morning.PlanReminder,
+		})
+	}
+
+	var checklistFacts map[string]store.Fact
+	if len(t.morningRoutines) > 0 {
+		checklistFacts = t.gatherMorningFacts(ctx)
+	}
+	plan := morning.BuildPlan(t.morningRoutines, checklistFacts, events, reminders, now)
+
+	out := ipc.DayPlan{Date: plan.Date, Spoken: plan.FormatRU()}
+	out.Items = make([]ipc.DayPlanItem, len(plan.Items))
+	for i, it := range plan.Items {
+		out.Items[i] = ipc.DayPlanItem{
+			At:        it.At,
+			Text:      it.Text,
+			Kind:      string(it.Kind),
+			Uncertain: it.Uncertain,
+		}
+	}
+	return out
+}
+
+// dayPlanMaxReminders bounds the reminder scan. The plan covers one day; a
+// pending queue longer than this is a bug elsewhere, not a plan to recite.
+const dayPlanMaxReminders = 500
+
 // tune — the feedback auto-tuner's impure step. runs on a slow cadence
 // (autotuneInterval, see run) so it doesn't write a fact every tick. for each
 // rule:
@@ -609,6 +939,7 @@ type daemonAPI struct {
 	ipc.CoreAPI
 	getTrace         func() *loop.TickTrace
 	getMorningStatus func(ctx context.Context) []ipc.MorningRoutineStatus
+	getDayPlan       func(ctx context.Context) ipc.DayPlan
 	chatFn           func(ctx context.Context, text string) string
 }
 
@@ -634,6 +965,13 @@ func (d *daemonAPI) MorningStatus(ctx context.Context) ([]ipc.MorningRoutineStat
 	return d.getMorningStatus(ctx), nil
 }
 
+func (d *daemonAPI) DayPlan(ctx context.Context) (ipc.DayPlan, error) {
+	if d.getDayPlan == nil {
+		return ipc.DayPlan{}, errors.New("mavend: day plan not available")
+	}
+	return d.getDayPlan(ctx), nil
+}
+
 func toIPCTickTrace(t loop.TickTrace) ipc.TickTrace {
 	rules := make([]ipc.RuleTrace, len(t.RuleTraces))
 	for i, r := range t.RuleTraces {

cmd/mavend/tick_test.go

@@ -46,7 +46,7 @@ func newTestTickLoop(t *testing.T, st *store.Store, sink delivery.Sink, digestCf
 		Nudges:    st,
 		Reminders: st,
 	})
-	return newTickLoop(st, g, d, phraser.NewStub(), rules, time.Second, 5*time.Minute, 0, digestCfg, nil, nil)
+	return newTickLoop(st, g, d, phraser.NewStub(), rules, time.Second, 5*time.Minute, 0, digestCfg, nil, nil, nil)
 }
 
 func TestTickFiresRoutineWhenScheduleCrosses(t *testing.T) {
@@ -63,7 +63,7 @@ func TestTickFiresRoutineWhenScheduleCrosses(t *testing.T) {
 	sink := &fakeSink{}
 	d := delivery.NewDispatcher(delivery.Config{Voice: sink, Ntfy: sink, Telegram: sink, Nudges: st, Reminders: st})
 	rs := []routine.Routine{{Name: "morning", Cron: "0 12 * * *", Body: "полдень, время воды", Severity: 1}}
-	tl := newTickLoop(st, g, d, phraser.NewStub(), rules, time.Second, 5*time.Minute, 0, nil, rs, nil)
+	tl := newTickLoop(st, g, d, phraser.NewStub(), rules, time.Second, 5*time.Minute, 0, nil, rs, nil, nil)
 
 	// first tick: seeds, does not fire the routine.
 	tl.tick(ctx, now)

cmd/mavend/voicewire.go

@@ -0,0 +1,435 @@
+package main
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+	"log"
+	"os"
+	"path/filepath"
+	"strings"
+	"time"
+
+	"github.com/kami/maven/internal/config"
+	"github.com/kami/maven/internal/delivery"
+	"github.com/kami/maven/internal/delivery/voicesink"
+	"github.com/kami/maven/internal/dialogue"
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/memory"
+	"github.com/kami/maven/internal/phraser"
+	"github.com/kami/maven/internal/router"
+	"github.com/kami/maven/internal/store"
+	"github.com/kami/maven/internal/stt"
+	"github.com/kami/maven/internal/tool"
+	"github.com/kami/maven/internal/tts"
+	"github.com/kami/maven/internal/voice"
+	"github.com/kami/maven/internal/weather"
+	"github.com/kami/maven/internal/worker"
+)
+
+// voiceWiring — everything the daemon needs to run the audio path. Held by
+// cmd/mavend/main.go alongside the other wirings; closed on shutdown.
+type voiceWiring struct {
+	server    *voice.Server
+	sessions  *voice.Sessions
+	voiceSink delivery.Sink
+	embedder  router.Embedder
+	handler   *reactiveHandler // the reactive handler for IPC Chat
+	// worker clients (set when configured as Remote): closed on shutdown so
+	// mavsttd / mavttsd don't keep a stale conn into a restarting daemon.
+	sttClient *worker.Client
+	ttsClient *worker.Client
+}
+
+// close releases the listener + worker conns. Safe to call on nil (when
+// voice is not wired — wireVoice returns nil,nil).
+func (w *voiceWiring) close() {
+	if w == nil {
+		return
+	}
+	if w.embedder != nil {
+		_ = w.embedder.Close()
+	}
+	if w.server != nil {
+		_ = w.server.Close()
+	}
+	if w.sttClient != nil {
+		_ = w.sttClient.Close()
+	}
+	if w.ttsClient != nil {
+		_ = w.ttsClient.Close()
+	}
+}
+
+// wireVoice builds the audio path from cfg + a CoreAPI + a router. Returns
+// nil wiring + nil error when voice isn't enabled (the caller's voice sink
+// stays nil; the dispatcher's ChannelVoice routing drops silently).
+//
+// When voice is enabled, MUST wire a voicesink into the dispatcher's Voice
+// slot using w.sessions (the caller does that — see main.go).
+func wireVoice(cfg *config.Config, coreAPI ipc.CoreAPI, phr phraser.Phraser, memStore memory.Store, dataStore *store.Store, eco *ecosystemWiring) (*voiceWiring, error) {
+	if cfg.Voice == nil || !cfg.Voice.Enabled {
+		return nil, nil
+	}
+	w := &voiceWiring{}
+
+	// ----- stt (Stub in-process OR Remote via worker socket) -----
+	var transcriber stt.Transcriber
+	if cfg.Voice.Stt != nil && cfg.Voice.Stt.Socket != "" {
+		c := worker.Dial(cfg.Voice.Stt.Socket)
+		w.sttClient = c
+		lang := cfg.Voice.Stt.Lang
+		if lang == "" {
+			lang = cfg.Voice.Lang
+		}
+		transcriber = stt.NewRemote(c, lang)
+	} else {
+		transcriber = stt.NewStub()
+	}
+
+	// ----- tts (Stub in-process OR Remote) -----
+	var synthesizer tts.Synthesizer
+	if cfg.Voice.Tts != nil && cfg.Voice.Tts.Socket != "" {
+		c := worker.Dial(cfg.Voice.Tts.Socket)
+		w.ttsClient = c
+		lang := cfg.Voice.Tts.Lang
+		if lang == "" {
+			lang = cfg.Voice.Lang
+		}
+		synthesizer = tts.NewRemote(c, lang, cfg.Voice.Tts.Voice)
+	} else {
+		synthesizer = tts.NewStub()
+	}
+
+	// ----- router: embedder (ONNX when configured, floor HashEmbedder otherwise) -----
+	var emb router.Embedder
+	if cfg.Voice.Embedder != nil {
+		onnx, err := router.NewONNXEmbedder(
+			cfg.Voice.Embedder.ModelPath,
+			cfg.Voice.Embedder.TokenizerPath,
+			cfg.Voice.Embedder.LibPath,
+		)
+		if err != nil {
+			w.close()
+			return nil, fmt.Errorf("embedder: %w", err)
+		}
+		log.Printf("voice: onnx embedder loaded (%d dim)", onnx.Dim())
+		emb = onnx
+	} else {
+		log.Printf("voice: embedder not configured, using HashEmbedder floor")
+		emb = router.NewHashEmbedder(1024)
+	}
+	w.embedder = emb
+	checkStoredEmbedder(dataStore, emb)
+
+	// ----- tool executor (the enabled act allowlist, store-backed) -----
+	// Config tools are the declarative bootstrap: seed them into the store as
+	// enabled (editing mavend.json IS the human enable act). Ad-hoc tools are
+	// enabled later through the authed mavweb surface. The executor + matcher
+	// both read the store live, so a newly-enabled tool is runnable without a
+	// daemon restart.
+	seedTools(coreAPI, cfg.Voice.Tools)
+	exec := tool.NewExecutor(coreAPI, time.Duration(cfg.Voice.ToolTimeout))
+	matcher := tool.NewMatcher(coreAPI)
+
+	// ----- weather provider (Open-Meteo when configured, Stub otherwise) -----
+	var weatherProvider weather.Provider
+	var weatherLocation string
+	if cfg.Voice.Weather != nil && cfg.Voice.Weather.Provider == "open-meteo" {
+		weatherProvider = weather.NewOpenMeteoProvider()
+		weatherLocation = cfg.Voice.Weather.DefaultLocation
+		log.Printf("voice: weather provider: open-meteo (default location: %s)", cfg.Voice.Weather.DefaultLocation)
+	} else {
+		weatherProvider = weather.NewStubProvider()
+		log.Printf("voice: weather provider: stub (not configured)")
+	}
+
+	// The replier uses the same llama-server as the phraser.
+	var llmClient *llm.Client
+	if lp, ok := phr.(*phraser.LLMPhraser); ok {
+		llmClient = llm.New(lp.BaseURL(), 60*time.Second)
+	}
+	// ----- router (the cascade; floor examples seed the classifier) -----
+	// The act matcher's allowlist is exactly the enabled tool names — the
+	// router only matches acts the executor can run (one source of truth).
+	threshold := cfg.Voice.RouterThreshold
+	if threshold <= 0 {
+		threshold = config.DefaultRouterThreshold
+	}
+	// The resident model routes by default: 63.2% of held-out intents right
+	// against the classifier's 50.0%, at about 1s a turn instead of 30ms (see
+	// config.VoiceConfig.LLMRouter). The classifier always stays wired as the
+	// fallback, so a model error never breaks a turn.
+	rtr := buildRouter(emb, matcher, threshold, pickLLMRouter(cfg.Voice.UseLLMRouter(), llmClient))
+
+	// ----- sessions registry (shared with voicesink) -----
+	sessions := voice.NewSessions()
+	w.sessions = sessions
+
+	// ----- voice sink (proactive nudges: dispatcher → voicesink → tts → push to client) -----
+	w.voiceSink = voicesink.New(synthesizer, sessions)
+
+	// ----- memory (long-term vector storage) -----
+	// Persistent (store-backed, survives restarts) when the daemon passes one;
+	// falls back to the in-memory floor otherwise (tests / no-store paths).
+	if memStore == nil {
+		memStore = memory.NewInMemoryStore()
+	}
+
+	// ----- dialogue (multi-turn slot carry-over; 2-min follow-up window) -----
+	// Store-backed when the daemon passes a store, so a restart mid-conversation
+	// keeps the thread (Vikunja #363). Sessions past their TTL are dropped on
+	// load, never revived. Clarify's parked question stays in memory only.
+	var dialogueSessions *dialogue.SessionStore
+	if dataStore != nil {
+		dialogueSessions = dialogue.NewPersistentSessionStore(2*time.Minute, dataStore)
+		if err := dialogueSessions.Load(context.Background(), time.Now()); err != nil {
+			log.Printf("dialogue: load saved sessions: %v", err)
+		}
+	} else {
+		dialogueSessions = dialogue.NewSessionStore(2 * time.Minute)
+	}
+	clarifyStore := dialogue.NewClarifyStore(clarifyTTL)
+	timeParser := router.NewPythonDateParser()
+
+	// ----- replier (LLM-backed when the engine is on, Stub floor otherwise) -----
+	replier := voice.Replier(voice.NewStubReplier())
+	if llmClient != nil {
+		replier = newLLMReplier(llmClient, contextBlockFn(cfg, time.Now))
+	}
+
+	// ----- the handler (the reactive path; closes over stt / tts / router / coreAPI / memory) -----
+	h := &reactiveHandler{
+		stt:              transcriber,
+		tts:              synthesizer,
+		router:           rtr,
+		embedder:         emb,
+		api:              coreAPI,
+		tools:            exec,
+		matcher:          matcher,
+		replier:          replier,
+		phraser:          phr,
+		now:              time.Now,
+		feedsOn:          cfg.Feeds != nil,
+		weatherProvider:  weatherProvider,
+		weatherLocation:  weatherLocation,
+		memStore:         memStore,
+		dataStore:        dataStore,
+		dialogueSessions: dialogueSessions,
+		clarifyStore:     clarifyStore,
+		// 0 here (unset config) ⇒ the dialogue default.
+		clarifyMaxAttempts: cfg.Voice.ClarifyMaxAttempts,
+		extractor:          router.Extractor{Time: timeParser, Acts: matcher, Facts: router.DefaultFactParser{}},
+		queryMinScore:      cfg.Voice.QueryMinScore,
+		queryMinMargin:     cfg.Voice.QueryMinMargin,
+		timeParser:         timeParser,
+		ecosystem:          eco,
+	}
+
+	// ----- the server (TCP listener) -----
+	srv := voice.NewServer(cfg.Voice.Bind, h, sessions)
+	if err := srv.Listen(); err != nil {
+		w.close()
+		return nil, fmt.Errorf("voice listen: %w", err)
+	}
+	w.server = srv
+	w.handler = h
+
+	return w, nil
+}
+
+// pickLLMRouter returns the LLM router when the operator asked for it and there
+// is a llama-server to talk to, and nil otherwise. nil is safe: the cascade then
+// routes with the classifier, so an unusable setting costs accuracy, not turns.
+func pickLLMRouter(enabled bool, c *llm.Client) *router.LLMRouter {
+	if !enabled {
+		return nil
+	}
+	if c == nil {
+		log.Printf("voice: voice.llm_router is on but there is no llama-server to route with (the phraser is not an LLM phraser) — using the classifier instead")
+		return nil
+	}
+	log.Printf("voice: LLM router enabled")
+	return router.NewLLMRouter(c)
+}
+
+// buildRouter constructs the reactive-path router with the given embedder
+// and confidence threshold.
+//   - stage-0 grammars from DefaultActMatcher whose fn allowlist is exactly
+//     the enabled tool names (actFns) — the router only matches acts the
+//     executor can run. Empty ⇒ every act refuses at the matcher.
+//   - The embedder is provided by wireVoice: HashEmbedder (floor) when no
+//     embedder config is present, or the ONNX multilingual model when
+//     configured — same interface, one constructor change.
+//   - 6 bootstrap examples covering the 5 intents + one compound-capture
+//     placeholder. Spec calls for ~10 per intent at production; this is the
+//     bootstrapping floor swapped by tuning the seed set later.
+//   - Threshold is from voice.router_threshold config (default 0.55).
+func buildRouter(emb router.Embedder, acts router.ActMatcher, threshold float64, llmR *router.LLMRouter) *router.Router {
+	cls := router.NewClassifier(emb)
+	seedClassifier(cls)
+	grammars := router.DefaultGrammars(acts)
+	grammars = append(grammars, router.SystemTimeDateGrammars()...)
+	grammars = append(grammars, router.ReminderGrammar())
+	return router.New(router.Config{
+		Grammars:   grammars,
+		Classifier: cls,
+		Extractor: router.Extractor{
+			Time:  router.NewPythonDateParser(),
+			Acts:  acts,
+			Facts: router.DefaultFactParser{},
+		},
+		Threshold: threshold,
+		LLM:       llmR,
+	})
+}
+
+// seedDir is the directory containing intent seed files. Each file is named
+// <intent>.txt and contains one training example per line (blank lines and
+// lines starting with # are ignored). Relative to the working directory.
+const seedDir = "models/seeds"
+
+// seedClassifier floors the embedded examples so the cold-boot path
+// doesn't return ErrNoIntents. Loads examples from seedDir — one file per
+// intent (act.txt, reminder.txt, fact.txt, note.txt, query.txt). When the
+// classifier can't decide it falls through to Clarify — the last-resort
+// path asks the user to rephrase rather than guessing wrong.
+func seedClassifier(c *router.Classifier) {
+	intents := []router.Intent{
+		router.IntentAct,
+		router.IntentReminder,
+		router.IntentFact,
+		router.IntentNote,
+		router.IntentQuery,
+		router.IntentChat,
+		router.IntentSystem,
+	}
+	total := 0
+	for _, intent := range intents {
+		n, err := loadSeedFile(c, intent)
+		if err != nil {
+			log.Printf("voice: seed %s: %v", intent, err)
+			continue
+		}
+		total += n
+	}
+	log.Printf("voice: loaded %d seed examples from %s", total, seedDir)
+}
+
+func loadSeedFile(c *router.Classifier, intent router.Intent) (int, error) {
+	path := filepath.Join(seedDir, string(intent)+".txt")
+	f, err := os.Open(path)
+	if err != nil {
+		return 0, fmt.Errorf("open %s: %w", path, err)
+	}
+	defer f.Close()
+
+	var count int
+	sc := bufio.NewScanner(f)
+	for sc.Scan() {
+		line := strings.TrimSpace(sc.Text())
+		if line == "" || strings.HasPrefix(line, "#") {
+			continue
+		}
+		if err := c.AddExample(context.Background(), intent, line); err != nil {
+			log.Printf("voice: seed %s: skipping %q: %v", intent, line, err)
+			continue
+		}
+		count++
+	}
+	if err := sc.Err(); err != nil {
+		return count, fmt.Errorf("scan %s: %w", path, err)
+	}
+	return count, nil
+}
+
+// seedTools upserts the config-declared tools into the store as enabled. Editing
+// mavend.json is a human act, so a config tool is enabled by definition; this
+// makes the declarative config the reproducible bootstrap while the store stays
+// the single runtime source of truth (mavweb enables ad-hoc ones on top).
+func seedTools(api ipc.CoreAPI, tools []config.ToolConfig) {
+	ctx := context.Background()
+	now := time.Now()
+	n := 0
+	for _, tc := range tools {
+		if tc.Name == "" || len(tc.Cmd) == 0 {
+			log.Printf("voice: skipping malformed tool config %+v", tc)
+			continue
+		}
+		if err := api.EnableTool(ctx, tc.Name, tc.Cmd, tc.Destructive, tc.Scope, now); err != nil {
+			log.Printf("voice: seed tool %q: %v", tc.Name, err)
+			continue
+		}
+		n++
+	}
+	log.Printf("voice: seeded %d act tools from config", n)
+}
+
+// reembedOnStart is the -reembed flag (set in run()). Opt-in on purpose: see
+// runReembed.
+var reembedOnStart bool
+
+// checkStoredEmbedder compares the embedder we just loaded with the one that
+// wrote the vectors already in the DB (Vikunja #378).
+//
+// The two models we have both make 384-dim vectors, so a size check catches
+// nothing: after a swap, recall silently compares vectors from different
+// spaces and the scores are noise. So we say it out loud. Recall itself is not
+// changed here — the fix is `mavend -reembed`.
+func checkStoredEmbedder(dataStore *store.Store, emb router.Embedder) {
+	if dataStore == nil {
+		return
+	}
+	current := router.EmbedderID(emb)
+	if reembedOnStart {
+		runReembed(dataStore, emb, current)
+		return
+	}
+	stored, mismatch, err := dataStore.CheckEmbedder(context.Background(), current)
+	if err != nil {
+		log.Printf("voice: embedder marker check failed: %v", err)
+		return
+	}
+	if mismatch {
+		log.Printf("voice: WARNING embedder MISMATCH — stored vectors were written by %q but the configured embedder is %q; recall scores are noise until the notes and facts are re-embedded — run `mavend -reembed` once (Vikunja #378)", stored, current)
+		return
+	}
+	log.Printf("voice: embedder marker ok (%s)", current)
+}
+
+// runReembed is the one-shot backfill behind -reembed.
+//
+// Why a flag and not automatic on mismatch: the embedder is ONNX on the
+// laptop's CPU, so a few thousand notes is minutes of work. Doing that silently
+// inside a normal start would look like the daemon hanging on boot. So the user
+// runs it once, deliberately, after an embedder swap; the mismatch warning
+// above tells them to. It re-embeds, logs what it did, and then the daemon
+// carries on serving as usual — no separate binary, no second start needed.
+func runReembed(dataStore *store.Store, emb router.Embedder, current string) {
+	log.Printf("voice: re-embedding stored notes and facts with %s — this can take a few minutes, do not interrupt", current)
+	res, err := dataStore.ReembedAll(context.Background(), current,
+		// EmbedPassage, not EmbedQuery: these are stored texts being searched
+		// FOR, which is the side they were written with.
+		func(ctx context.Context, text string) ([]float32, error) {
+			return router.EmbedPassage(ctx, emb, text)
+		})
+	if err != nil {
+		log.Printf("voice: re-embed FAILED, nothing was changed and no marker was written — safe to run again: %v", err)
+		return
+	}
+	if res.Skipped {
+		log.Printf("voice: re-embed skipped — the stored vectors were already written by %s", current)
+		return
+	}
+	log.Printf("voice: re-embed done — %d notes in the notes table, %d notes and %d facts in the memory index, took %s; stored vectors now belong to %s",
+		res.Notes, res.MemNotes, res.Facts, res.Took.Round(time.Second), current)
+
+	// A row with no text cannot be re-embedded, so its vector is still the old
+	// model's noise while the marker now says everything is current. Both write
+	// paths always store the text, so this should be zero — say it loudly
+	// rather than bury it in the line above if it ever isn't.
+	if res.NoText > 0 {
+		log.Printf("voice: WARNING %d stored rows had no text, so their vectors could not be re-embedded and are still noise; they will never match anything useful (Vikunja #378)", res.NoText)
+	}
+}

cmd/mavend/weatherq.go

@@ -0,0 +1,50 @@
+// Package main — weatherq.go holds the weather-query keyword helpers: does
+// this utterance ask about weather at all, and which city (if any) did it
+// name. Both are plain substring/lookup matching, not NLU — extend this file
+// rather than voice.go for anything in that shape.
+package main
+
+import "strings"
+
+// isWeatherQuery returns true if the utterance is about weather.
+func isWeatherQuery(u string) bool {
+	lower := strings.ToLower(u)
+	return strings.Contains(lower, "погод") ||
+		strings.Contains(lower, "градус") ||
+		strings.Contains(lower, "температур") ||
+		strings.Contains(lower, "дожд") ||
+		strings.Contains(lower, "холод") ||
+		strings.Contains(lower, "тепл") ||
+		strings.Contains(lower, "weather") ||
+		strings.Contains(lower, "temperature")
+}
+
+// extractWeatherLocation parses a location from the utterance, or falls back
+// to the configured default. Very basic: just checks for known city names.
+func extractWeatherLocation(u, defaultLoc string) string {
+	lower := strings.ToLower(u)
+	cities := map[string]string{
+		"москв":    "Moscow",
+		"moscow":   "Moscow",
+		"питер":    "Saint Petersburg",
+		"spb":      "Saint Petersburg",
+		"петербур": "Saint Petersburg",
+		"лондон":   "London",
+		"london":   "London",
+		"париж":    "Paris",
+		"paris":    "Paris",
+		"берлин":   "Berlin",
+		"berlin":   "Berlin",
+		"нью-йорк": "New York",
+		"new york": "New York",
+	}
+	for substr, name := range cities {
+		if strings.Contains(lower, substr) {
+			return name
+		}
+	}
+	if defaultLoc != "" {
+		return defaultLoc
+	}
+	return "Moscow"
+}

cmd/mavmaild/main.go

@@ -0,0 +1,332 @@
+// mavmaild — the mail reader module (Vikunja #246,
+// docs/plans/01-email-reader.md).
+//
+// Every so often it opens one IMAP mailbox read-only, fetches the messages it
+// has not read yet, and hands each one to core over ipc.MethodIngestMail. Core
+// runs the extraction on the resident model and writes what comes back as task
+// CANDIDATES he reviews on /tasks. Nothing here writes to the store, nothing
+// here can create a reminder, and nothing here speaks.
+//
+// Why a separate daemon rather than a loop inside mavend, when extraction has
+// to happen in mavend anyway: the credential. mavpoll set the precedent with the
+// zenmoney token (#125) — the module that talks to a third party holds the
+// secret, reads it from a FILE so it never appears in `ps`, in
+// docker-compose.yml or in shell history, and core never sees it. Core learns
+// that mail exists only as message text on one IPC method; it cannot connect to
+// the mailbox even if it wanted to, and a compromised core yields no mail
+// password.
+//
+// Off unless configured: without -password-file there is nothing to run, and
+// the daemon says so and exits. If core has no `email` block the very first
+// ingest comes back ErrUnknownMethod and this daemon stops polling instead of
+// hammering a socket that will keep refusing.
+//
+// Mail is personal, so the log is counts and UIDs: how many messages were
+// fetched, how many were bulk, how many candidates came back. No subject, no
+// sender, no body, ever — reviewing a candidate is what /tasks is for.
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"errors"
+	"flag"
+	"fmt"
+	"log"
+	"os"
+	"os/signal"
+	"path/filepath"
+	"sort"
+	"strings"
+	"syscall"
+	"time"
+
+	"github.com/kami/maven/internal/email"
+	"github.com/kami/maven/internal/ipc"
+)
+
+func main() {
+	if err := run(os.Args[1:]); err != nil {
+		fmt.Fprintln(os.Stderr, "mavmaild:", err)
+		os.Exit(1)
+	}
+}
+
+func run(args []string) error {
+	fs := flag.NewFlagSet("mavmaild", flag.ContinueOnError)
+	socket := fs.String("socket", "", "core IPC socket path (required)")
+	server := fs.String("imap", "", "IMAP server, host or host:993 (required)")
+	user := fs.String("user", "", "IMAP username (required)")
+	passFile := fs.String("password-file", "", "file holding the IMAP password (required — never passed as a flag value)")
+	mailbox := fs.String("mailbox", "INBOX", "mailbox to read, read-only")
+	interval := fs.Duration("interval", 15*time.Minute, "how often to read the mailbox")
+	lookback := fs.Duration("lookback", 72*time.Hour, "how far back to search on each poll")
+	max := fs.Int("max", 25, "most messages to fetch in one poll")
+	timeout := fs.Duration("timeout", 30*time.Second, "IMAP network timeout")
+	statePath := fs.String("state", "", "file remembering which UIDs were read (default: none — every poll re-reads the window)")
+	if err := fs.Parse(args); err != nil {
+		return err
+	}
+	if *socket == "" {
+		return fmt.Errorf("-socket is required")
+	}
+	if *server == "" || *user == "" || *passFile == "" {
+		return fmt.Errorf("mail reading is off unless configured: set -imap, -user and -password-file")
+	}
+
+	// The password is read from a file, never taken as a flag value: an argv
+	// secret is visible in `ps` to every user on the box and lands in the compose
+	// file and the shell history. Read once at start — a rotated password means a
+	// restart, which is cheaper than re-reading his credential every quarter hour.
+	raw, err := os.ReadFile(*passFile)
+	if err != nil {
+		return fmt.Errorf("read password file: %w", err)
+	}
+	password := strings.TrimSpace(string(raw))
+	if password == "" {
+		return fmt.Errorf("password file %s is empty", *passFile)
+	}
+
+	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
+	defer stop()
+
+	core, err := ipc.DialWait(*socket, 60*time.Second)
+	if err != nil {
+		return err
+	}
+	defer core.Close()
+
+	r := &reader{
+		core:     core,
+		addr:     *server,
+		user:     *user,
+		mailbox:  *mailbox,
+		lookback: *lookback,
+		max:      *max,
+		timeout:  *timeout,
+		state:    newSeenState(*statePath),
+	}
+	if err := r.state.load(); err != nil {
+		// A missing or corrupt state file must not stop mail from being read: the
+		// worst case is re-reading the window, and capture dedupes on text.
+		log.Printf("mavmaild: state: %v (starting from an empty seen-set)", err)
+	}
+
+	// The password is never logged, not even its length.
+	log.Printf("mavmaild: reading %s on %s every %s (lookback %s, max %d/poll)",
+		*mailbox, *server, *interval, *lookback, *max)
+
+	r.pollOnce(ctx, password) // don't idle a full interval on start
+	t := time.NewTicker(*interval)
+	defer t.Stop()
+	for {
+		select {
+		case <-ctx.Done():
+			log.Printf("mavmaild: bye")
+			return nil
+		case <-t.C:
+			if r.disabled {
+				// Core told us mail ingestion is not configured. Nothing will change
+				// without a core restart, and a restart restarts us too.
+				log.Printf("mavmaild: core does not accept mail — idling")
+				return nil
+			}
+			r.pollOnce(ctx, password)
+		}
+	}
+}
+
+// mailIngester — the slice of core this daemon uses. One method: hand over a
+// message. It cannot write a fact, create a reminder or read the store, and the
+// interface says so.
+type mailIngester interface {
+	IngestMail(ctx context.Context, req ipc.IngestMailReq) (ipc.IngestMailResp, error)
+}
+
+type reader struct {
+	core     mailIngester
+	addr     string
+	user     string
+	mailbox  string
+	lookback time.Duration
+	max      int
+	timeout  time.Duration
+	state    *seenState
+
+	// dial — connection seam for the tests; nil ⇒ implicit TLS.
+	dial func(addr string, timeout time.Duration) (*email.Conn, error)
+
+	// disabled — core answered ErrUnknownMethod, i.e. it has no email block.
+	disabled bool
+}
+
+// pollOnce — one read of the mailbox, then one ingest per message.
+//
+// A fetch error aborts this poll and nothing else; the next tick tries again.
+// An ingest error for one message does not skip the rest — one mail the model
+// choked on should not hide the four behind it.
+func (r *reader) pollOnce(ctx context.Context, password string) {
+	msgs, err := r.fetch(password)
+	if err != nil {
+		// The error may name a UID; it never names a subject or a sender.
+		log.Printf("mavmaild: fetch: %v", err)
+		if len(msgs) == 0 {
+			return
+		}
+	}
+	var junk, candidates, created int
+	for _, m := range msgs {
+		if ctx.Err() != nil {
+			return
+		}
+		if m.Junk {
+			junk++
+			// Marked seen without a model call: the header filter already decided,
+			// and re-classifying it every quarter hour would be pure waste.
+			r.state.mark(m.UID)
+			continue
+		}
+		resp, err := r.core.IngestMail(ctx, ipc.IngestMailReq{
+			Mailbox: r.mailbox,
+			UID:     m.UID,
+			From:    m.From,
+			Subject: m.Subject,
+			Date:    m.Date,
+			Body:    m.Body,
+		})
+		if errors.Is(err, ipc.ErrUnknownMethod) {
+			log.Printf("mavmaild: core has no email block configured — mail ingestion is off; stopping")
+			r.disabled = true
+			return
+		}
+		if err != nil {
+			// Not marked seen: an ingest that failed should be retried next poll.
+			log.Printf("mavmaild: ingest uid %d: %v", m.UID, err)
+			continue
+		}
+		r.state.mark(m.UID)
+		candidates += len(resp.TaskIDs)
+		created += resp.Created
+	}
+	if err := r.state.save(); err != nil {
+		log.Printf("mavmaild: state: %v", err)
+	}
+	log.Printf("mavmaild: %s: %d read, %d bulk, %d candidate(s), %d new", r.mailbox, len(msgs), junk, candidates, created)
+}
+
+// fetch reads the mailbox. Messages already in the seen-set are not fetched at
+// all, so a steady mailbox costs one SEARCH per poll and nothing else.
+func (r *reader) fetch(password string) ([]email.Message, error) {
+	f := email.FetchSince{
+		Addr:    r.addr,
+		User:    r.user,
+		Mailbox: r.mailbox,
+		Timeout: r.timeout,
+		Since:   time.Now().Add(-r.lookback),
+		Max:     r.max,
+		Skip:    r.state.seen,
+	}
+	return f.RunWith(password, r.dial)
+}
+
+// ---- seen state ------------------------------------------------------------
+
+// seenState — the UIDs already handed to core, persisted so a restart does not
+// re-read (and re-extract, at multi-second LLM cost) the whole lookback window.
+//
+// Correctness does not depend on it: ipc.CaptureTask dedupes on normalised text
+// among live tasks, so a re-read produces no duplicate rows. This exists to save
+// the model's time, which is why a broken state file is a log line rather than a
+// failure.
+//
+// UIDs are per-mailbox and monotonic, so the set is kept as a high-water mark
+// plus the stragglers above it. If the server ever changes UIDVALIDITY, UIDs
+// reset and the window is simply re-read once — dedupe absorbs it.
+type seenState struct {
+	path  string
+	high  uint32
+	set   map[uint32]bool
+	dirty bool
+}
+
+func newSeenState(path string) *seenState {
+	return &seenState{path: path, set: map[uint32]bool{}}
+}
+
+type seenFile struct {
+	High uint32   `json:"high"`
+	UIDs []uint32 `json:"uids,omitempty"`
+}
+
+func (s *seenState) seen(uid uint32) bool {
+	return uid <= s.high || s.set[uid]
+}
+
+func (s *seenState) mark(uid uint32) {
+	if s.seen(uid) {
+		return
+	}
+	s.set[uid] = true
+	s.dirty = true
+	// Advance the high-water mark through any contiguous run, so the explicit set
+	// stays small on a mailbox read in order.
+	for {
+		next := s.high + 1
+		if !s.set[next] {
+			break
+		}
+		delete(s.set, next)
+		s.high = next
+	}
+}
+
+func (s *seenState) load() error {
+	if s.path == "" {
+		return nil
+	}
+	b, err := os.ReadFile(s.path)
+	if errors.Is(err, os.ErrNotExist) {
+		return nil // first run
+	}
+	if err != nil {
+		return err
+	}
+	var f seenFile
+	if err := json.Unmarshal(b, &f); err != nil {
+		return fmt.Errorf("parse %s: %w", s.path, err)
+	}
+	s.high = f.High
+	for _, u := range f.UIDs {
+		s.set[u] = true
+	}
+	return nil
+}
+
+// save writes the state atomically (temp file + rename), 0600: it is a list of
+// message ids from his mailbox, which is metadata about his mail.
+func (s *seenState) save() error {
+	if s.path == "" || !s.dirty {
+		return nil
+	}
+	uids := make([]uint32, 0, len(s.set))
+	for u := range s.set {
+		uids = append(uids, u)
+	}
+	sort.Slice(uids, func(i, j int) bool { return uids[i] < uids[j] })
+	b, err := json.Marshal(seenFile{High: s.high, UIDs: uids})
+	if err != nil {
+		return err
+	}
+	tmp := s.path + ".tmp"
+	if err := os.MkdirAll(filepath.Dir(s.path), 0o700); err != nil {
+		return err
+	}
+	if err := os.WriteFile(tmp, b, 0o600); err != nil {
+		return err
+	}
+	if err := os.Rename(tmp, s.path); err != nil {
+		return err
+	}
+	s.dirty = false
+	return nil
+}

cmd/mavmaild/main_test.go

@@ -0,0 +1,254 @@
+package main
+
+import (
+	"bufio"
+	"context"
+	"fmt"
+	"net"
+	"os"
+	"path/filepath"
+	"strconv"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/email"
+	"github.com/kami/maven/internal/ipc"
+)
+
+// ---- a scripted IMAP server, same shape internal/email's tests use ---------
+
+type fakeIMAP struct {
+	msgs map[uint32]string
+	uids []uint32
+	cmds []string
+}
+
+func (f *fakeIMAP) serve(c net.Conn) {
+	defer c.Close()
+	fmt.Fprint(c, "* OK fake ready\r\n")
+	r := bufio.NewReader(c)
+	for {
+		line, err := r.ReadString('\n')
+		if err != nil {
+			return
+		}
+		parts := strings.SplitN(strings.TrimRight(line, "\r\n"), " ", 2)
+		if len(parts) != 2 {
+			return
+		}
+		tag, cmd := parts[0], parts[1]
+		f.cmds = append(f.cmds, cmd)
+		upper := strings.ToUpper(cmd)
+		switch {
+		case strings.HasPrefix(upper, "LOGIN"), strings.HasPrefix(upper, "EXAMINE"):
+			fmt.Fprintf(c, "%s OK\r\n", tag)
+		case strings.HasPrefix(upper, "UID SEARCH"):
+			var ids []string
+			for _, u := range f.uids {
+				ids = append(ids, strconv.FormatUint(uint64(u), 10))
+			}
+			fmt.Fprintf(c, "* SEARCH %s\r\n%s OK\r\n", strings.Join(ids, " "), tag)
+		case strings.HasPrefix(upper, "UID FETCH"):
+			uid, _ := strconv.ParseUint(strings.Fields(cmd)[2], 10, 32)
+			if raw, ok := f.msgs[uint32(uid)]; ok {
+				fmt.Fprintf(c, "* 1 FETCH (UID %d BODY[] {%d}\r\n%s)\r\n", uid, len(raw), raw)
+			}
+			fmt.Fprintf(c, "%s OK\r\n", tag)
+		case strings.HasPrefix(upper, "LOGOUT"):
+			fmt.Fprintf(c, "* BYE\r\n%s OK\r\n", tag)
+			return
+		default:
+			fmt.Fprintf(c, "%s BAD\r\n", tag)
+		}
+	}
+}
+
+func (f *fakeIMAP) dial(_ string, timeout time.Duration) (*email.Conn, error) {
+	cli, srv := net.Pipe()
+	go f.serve(srv)
+	return email.NewConn(cli, timeout)
+}
+
+// ---- a fake core -----------------------------------------------------------
+
+type fakeCore struct {
+	got  []ipc.IngestMailReq
+	resp ipc.IngestMailResp
+	err  error
+}
+
+func (c *fakeCore) IngestMail(_ context.Context, req ipc.IngestMailReq) (ipc.IngestMailResp, error) {
+	c.got = append(c.got, req)
+	if c.err != nil {
+		return ipc.IngestMailResp{}, c.err
+	}
+	return c.resp, nil
+}
+
+func mail(subject, body string, extraHeaders ...string) string {
+	h := "Subject: " + subject + "\r\nFrom: a@b.c\r\nContent-Type: text/plain; charset=utf-8\r\n"
+	for _, e := range extraHeaders {
+		h += e + "\r\n"
+	}
+	return h + "\r\n" + body + "\r\n"
+}
+
+func newTestReader(t *testing.T, f *fakeIMAP, core *fakeCore, statePath string) *reader {
+	t.Helper()
+	return &reader{
+		core: core, addr: "mail.example:993", user: "kami", mailbox: "INBOX",
+		lookback: 72 * time.Hour, max: 25, timeout: 5 * time.Second,
+		state: newSeenState(statePath),
+		dial:  f.dial,
+	}
+}
+
+func TestPollHandsMessagesToCore(t *testing.T) {
+	f := &fakeIMAP{
+		uids: []uint32{1, 2},
+		msgs: map[uint32]string{
+			1: mail("Счёт", "Оплатить до 5 августа."),
+			2: mail("Скидки", "Sale!", "List-Unsubscribe: <mailto:u@x>"),
+		},
+	}
+	core := &fakeCore{resp: ipc.IngestMailResp{TaskIDs: []int64{1}, Created: 1}}
+	r := newTestReader(t, f, core, "")
+	r.pollOnce(context.Background(), "secret")
+
+	// The newsletter is filtered before core is asked: only the real mail crosses.
+	if len(core.got) != 1 {
+		t.Fatalf("core saw %d messages, want 1 (the bulk one must not cross): %+v", len(core.got), core.got)
+	}
+	got := core.got[0]
+	if got.UID != 1 || got.Mailbox != "INBOX" || got.Subject != "Счёт" {
+		t.Errorf("ingest req = %+v", got)
+	}
+	if !strings.Contains(got.Body, "Оплатить") {
+		t.Errorf("body = %q", got.Body)
+	}
+}
+
+// A second poll must not re-send what core already saw — extraction is a
+// multi-second LLM call per message.
+func TestPollSkipsSeenUIDs(t *testing.T) {
+	f := &fakeIMAP{uids: []uint32{5}, msgs: map[uint32]string{5: mail("Счёт", "текст")}}
+	core := &fakeCore{}
+	r := newTestReader(t, f, core, "")
+	r.pollOnce(context.Background(), "secret")
+	r.pollOnce(context.Background(), "secret")
+	if len(core.got) != 1 {
+		t.Errorf("core saw %d messages over two polls, want 1", len(core.got))
+	}
+}
+
+// An ingest that failed is NOT marked seen: the next poll retries it.
+func TestPollRetriesFailedIngest(t *testing.T) {
+	f := &fakeIMAP{uids: []uint32{5}, msgs: map[uint32]string{5: mail("Счёт", "текст")}}
+	core := &fakeCore{err: fmt.Errorf("llama-server is warming up")}
+	r := newTestReader(t, f, core, "")
+	r.pollOnce(context.Background(), "secret")
+	core.err = nil
+	r.pollOnce(context.Background(), "secret")
+	if len(core.got) != 2 {
+		t.Errorf("core saw %d attempts, want 2 (a failed ingest is retried)", len(core.got))
+	}
+}
+
+// Core without an email block ⇒ stop, don't hammer the socket.
+func TestPollStopsWhenCoreRefusesMail(t *testing.T) {
+	f := &fakeIMAP{uids: []uint32{1, 2}, msgs: map[uint32]string{1: mail("a", "b"), 2: mail("c", "d")}}
+	core := &fakeCore{err: fmt.Errorf("call: %w", ipc.ErrUnknownMethod)}
+	r := newTestReader(t, f, core, "")
+	r.pollOnce(context.Background(), "secret")
+	if !r.disabled {
+		t.Error("ErrUnknownMethod must disable the reader")
+	}
+	if len(core.got) != 1 {
+		t.Errorf("core saw %d messages, want 1 — stop at the first refusal", len(core.got))
+	}
+}
+
+func TestSeenStatePersists(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "state", "seen.json")
+	f := &fakeIMAP{uids: []uint32{9}, msgs: map[uint32]string{9: mail("Счёт", "текст")}}
+	core := &fakeCore{}
+	r := newTestReader(t, f, core, path)
+	r.pollOnce(context.Background(), "secret")
+
+	fi, err := os.Stat(path)
+	if err != nil {
+		t.Fatalf("state file: %v", err)
+	}
+	// A list of message ids from his mailbox is metadata about his mail.
+	if perm := fi.Mode().Perm(); perm != 0o600 {
+		t.Errorf("state file mode = %v, want 0600", perm)
+	}
+
+	// A fresh reader with the same state file must not re-read the message.
+	core2 := &fakeCore{}
+	r2 := newTestReader(t, f, core2, path)
+	if err := r2.state.load(); err != nil {
+		t.Fatalf("load: %v", err)
+	}
+	r2.pollOnce(context.Background(), "secret")
+	if len(core2.got) != 0 {
+		t.Errorf("after a restart core saw %d messages, want 0", len(core2.got))
+	}
+}
+
+func TestSeenStateHighWaterMark(t *testing.T) {
+	s := newSeenState("")
+	s.mark(1)
+	s.mark(3)
+	s.mark(2)
+	if s.high != 3 {
+		t.Errorf("high = %d, want 3 (contiguous run collapses)", s.high)
+	}
+	if len(s.set) != 0 {
+		t.Errorf("explicit set = %v, want empty", s.set)
+	}
+	if !s.seen(2) || s.seen(4) {
+		t.Errorf("seen(2)=%v seen(4)=%v", s.seen(2), s.seen(4))
+	}
+}
+
+func TestSeenStateCorruptFileIsNotFatal(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "seen.json")
+	if err := os.WriteFile(path, []byte("{not json"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	s := newSeenState(path)
+	if err := s.load(); err == nil {
+		t.Error("a corrupt state file should report an error the caller logs")
+	}
+	if s.seen(1) {
+		t.Error("a corrupt state file must leave an empty seen-set, not a poisoned one")
+	}
+}
+
+// Off unless configured, and the credential is never a flag value.
+func TestRunRequiresConfig(t *testing.T) {
+	if err := run([]string{}); err == nil {
+		t.Error("no -socket must be an error")
+	}
+	if err := run([]string{"-socket", "/tmp/nope.sock"}); err == nil {
+		t.Error("no mailbox configuration must be an error, not a default mailbox")
+	}
+	// There is no -password flag at all: only -password-file.
+	if err := run([]string{"-socket", "/x", "-imap", "h", "-user", "u", "-password", "p"}); err == nil ||
+		!strings.Contains(err.Error(), "flag provided but not defined") {
+		t.Errorf("a -password flag must not exist; err = %v", err)
+	}
+}
+
+func TestRunRejectsEmptyPasswordFile(t *testing.T) {
+	path := filepath.Join(t.TempDir(), "pass")
+	if err := os.WriteFile(path, []byte("  \n"), 0o600); err != nil {
+		t.Fatal(err)
+	}
+	err := run([]string{"-socket", "/x/y.sock", "-imap", "h", "-user", "u", "-password-file", path})
+	if err == nil || !strings.Contains(err.Error(), "empty") {
+		t.Errorf("an empty password file must be refused before dialling; err = %v", err)
+	}
+}

cmd/mavpoll/main.go

@@ -9,6 +9,12 @@
 // Two sources, each its own provenance (the loop's rules trust source):
 //   - netdata  → poll:netdata      resource alarms (disk/mem/cert/temp)
 //   - kuma     → poll:uptimekuma   service up/down (the source of truth for it)
+//   - zenmoney → poll:zenmoney     spending/income totals (Vikunja #125)
+//
+// The zenmoney source is why the token lives HERE and not in core: the poller
+// already owns every other third-party credential, it holds no store key, and
+// core never needs to know an account exists to answer a question about a fact
+// the poller wrote. It is off unless -zenmoney-token-file is given.
 //
 // Netdata needs no auth over the wg-fronted net. Kuma's /metrics needs an API
 // key (basic-auth); without -kuma the whole kuma path is skipped (netdata-only
@@ -37,6 +43,7 @@ import (
 	"time"
 
 	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/zenmoney"
 )
 
 func main() {
@@ -52,6 +59,9 @@ func run(args []string) error {
 	netdataURL := fs.String("netdata", "http://127.0.0.1:19999", "netdata base URL ('' to disable)")
 	kumaURL := fs.String("kuma", "", "uptime-kuma metrics URL, e.g. http://127.0.0.1:3001/metrics ('' to disable)")
 	kumaKey := fs.String("kuma-key", "", "uptime-kuma API key (basic-auth username)")
+	zenTokenFile := fs.String("zenmoney-token-file", "", "file holding the zenmoney API token ('' disables money tracking)")
+	zenURL := fs.String("zenmoney-url", zenmoney.DefaultBaseURL, "zenmoney API base URL (tests/self-hosted proxies)")
+	zenInterval := fs.Duration("zenmoney-interval", time.Hour, "how often to read zenmoney (money does not move every minute)")
 	wgIface := fs.String("wg", "", "wireguard interface for the presence signal, e.g. wg0 or 'all' ('' to disable)")
 	wgCmd := fs.String("wg-cmd", "wg", "wg binary (use e.g. 'sudo wg' if the poller lacks CAP_NET_ADMIN)")
 	interval := fs.Duration("interval", 60*time.Second, "poll cadence")
@@ -62,8 +72,24 @@ func run(args []string) error {
 	if *socket == "" {
 		return fmt.Errorf("-socket is required")
 	}
-	if *netdataURL == "" && *kumaURL == "" && *wgIface == "" {
+	if *netdataURL == "" && *kumaURL == "" && *wgIface == "" && *zenTokenFile == "" {
-		return fmt.Errorf("nothing to poll: set -netdata, -kuma and/or -wg")
+		return fmt.Errorf("nothing to poll: set -netdata, -kuma, -wg and/or -zenmoney-token-file")
+	}
+
+	// The token is read from a file, never taken as a flag value: an argv token
+	// is visible in `ps` to every user on the box and lands in the compose file
+	// and the shell history. Read once at start — a rotated token means a
+	// restart, which is cheaper than re-reading his credential every hour.
+	var zen *zenmoney.Client
+	if *zenTokenFile != "" {
+		raw, err := os.ReadFile(*zenTokenFile)
+		if err != nil {
+			return fmt.Errorf("read zenmoney token: %w", err)
+		}
+		zen, err = zenmoney.New(strings.TrimSpace(string(raw)), *zenURL, *timeout*3)
+		if err != nil {
+			return err
+		}
 	}
 
 	ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
@@ -83,9 +109,13 @@ func run(args []string) error {
 		kumaKey:    *kumaKey,
 		wgIface:    *wgIface,
 		wgCmd:      *wgCmd,
+		zen:        zen,
+		zenEvery:   *zenInterval,
 	}
 
-	log.Printf("mavpoll: polling every %s (netdata=%q kuma=%q wg=%q)", *interval, *netdataURL, *kumaURL, *wgIface)
+	// The token is never logged, not even its length.
+	log.Printf("mavpoll: polling every %s (netdata=%q kuma=%q wg=%q zenmoney=%v every %s)",
+		*interval, *netdataURL, *kumaURL, *wgIface, zen != nil, *zenInterval)
 	p.pollOnce(ctx) // fire immediately; don't idle a full interval on start
 	t := time.NewTicker(*interval)
 	defer t.Stop()
@@ -108,6 +138,12 @@ type poller struct {
 	kumaKey    string
 	wgIface    string
 	wgCmd      string
+
+	// zen is nil unless a token file was configured — money tracking is a
+	// capability, off by default like weather and telegram.
+	zen      *zenmoney.Client
+	zenEvery time.Duration
+	zenLast  time.Time
 }
 
 // pollOnce — one sweep of both sources. A failure in one source logs and does
@@ -129,6 +165,66 @@ func (p *poller) pollOnce(ctx context.Context) {
 			log.Printf("mavpoll: wg: %v", err)
 		}
 	}
+	// Money on its own, much slower cadence: a bank feed that updates hourly
+	// polled every minute is 60 pointless reads of his financial history.
+	if p.zen != nil && now.Sub(p.zenLast) >= p.zenEvery {
+		p.zenLast = now
+		if err := p.pollZenmoney(ctx, now); err != nil {
+			log.Printf("mavpoll: zenmoney: %v", err)
+		}
+	}
+}
+
+// ---- zenmoney: spending/income totals → money facts ------------------------
+
+// pollZenmoney reads today's and this month's totals and writes them as
+// facts(kind=env, source=poll:zenmoney) (Vikunja #125).
+//
+// Two properties this function exists to hold:
+//
+//   - An empty or failed read writes NOTHING. zenmoney.Summary.Value() refuses
+//     to encode a summary built from zero transactions, so a poller that cannot
+//     reach the API leaves the last good fact in place rather than overwriting
+//     it with a zero Maven would then recite as fact.
+//   - Nothing about the money leaves the box except the diff request itself, to
+//     the service that already holds his bank sessions. The totals are written
+//     to the store and read back only when he asks; they are never search input
+//     and no tick rule fires on them.
+//
+// Both windows are read from one diff call each. Two calls an hour against an
+// API whose whole job is this is not worth caching.
+// moneyWindow — one fact key and the period it covers.
+type moneyWindow struct {
+	key      string
+	from, to time.Time
+}
+
+func (p *poller) pollZenmoney(ctx context.Context, now time.Time) error {
+	dFrom, dTo := zenmoney.DayWindow(now)
+	mFrom, mTo := zenmoney.MonthWindow(now)
+	windows := []moneyWindow{
+		{zenmoney.KeySpentToday, dFrom, dTo},
+		{zenmoney.KeySpentMonth, mFrom, mTo},
+	}
+	var firstErr error
+	for _, w := range windows {
+		sum, err := p.zen.Since(ctx, w.from, w.to)
+		if err != nil {
+			if firstErr == nil {
+				firstErr = err
+			}
+			continue
+		}
+		val, ok := sum.Value()
+		if !ok {
+			// Nothing read. Silence, not a zero.
+			continue
+		}
+		if err := p.writeIfChangedRaw(ctx, w.key, zenmoney.Source, val, now); err != nil && firstErr == nil {
+			firstErr = err
+		}
+	}
+	return firstErr
 }
 
 // ---- wireguard: latest handshake → presence signal -------------------------
@@ -301,6 +397,31 @@ func (p *poller) writeIfChanged(ctx context.Context, key, source, val string, no
 	return nil
 }
 
+// writeIfChangedRaw is writeIfChanged for values that are already JSON (the
+// money facts store an object, not a string). Kept separate rather than
+// generalising writeIfChanged, because the string-valued env facts encoding
+// their own value is the convention the rules rely on.
+//
+// The log line names the key and the source, never the figures: mavpoll's log
+// is not the place his spending ends up.
+func (p *poller) writeIfChangedRaw(ctx context.Context, key, source, jsonVal string, now time.Time) error {
+	prev, err := p.core.LatestFactBySource(ctx, key, source)
+	switch {
+	case err == nil && prev.Value == jsonVal:
+		return nil
+	case err != nil && err != ipc.ErrNoFact && !isNoFact(err):
+		return fmt.Errorf("read %s: %w", key, err)
+	}
+	if _, err := p.core.WriteFact(ctx, ipc.WriteFactReq{
+		Ts: now, Kind: "env", Key: key, Value: jsonVal,
+		Source: source, Confidence: 1.0,
+	}); err != nil {
+		return fmt.Errorf("write %s: %w", key, err)
+	}
+	log.Printf("mavpoll: %s updated (%s)", key, source)
+	return nil
+}
+
 // isNoFact — ErrNoFact rehydrated over the wire is wrapped (fmt.Errorf %w), so
 // errors.Is is the right check; keep a helper so the switch above reads clean.
 func isNoFact(err error) bool {

cmd/mavpoll/main_test.go

@@ -1,8 +1,17 @@
 package main
 
 import (
+	"context"
 	"encoding/json"
+	"net/http"
+	"net/http/httptest"
+	"os"
+	"strings"
 	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/zenmoney"
 )
 
 func TestMaxSeverity(t *testing.T) {
@@ -62,3 +71,120 @@ func TestParseMaxHandshake(t *testing.T) {
 		}
 	}
 }
+
+// ---- zenmoney (Vikunja #125) ----------------------------------------------
+
+// factCore records the facts the poller wrote and answers "no fact yet".
+type factCore struct {
+	ipc.UnimplementedCoreAPI
+
+	written []ipc.WriteFactReq
+	prev    map[string]string
+}
+
+func (c *factCore) LatestFactBySource(_ context.Context, key, source string) (ipc.Fact, error) {
+	if v, ok := c.prev[key+"|"+source]; ok {
+		return ipc.Fact{Key: key, Source: source, Value: v}, nil
+	}
+	return ipc.Fact{}, ipc.ErrNoFact
+}
+
+func (c *factCore) WriteFact(_ context.Context, req ipc.WriteFactReq) (int64, error) {
+	c.written = append(c.written, req)
+	return int64(len(c.written)), nil
+}
+
+func zenFixtureServer(t *testing.T, body []byte, status int) *httptest.Server {
+	t.Helper()
+	return httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		if status != http.StatusOK {
+			w.WriteHeader(status)
+			return
+		}
+		w.Write(body)
+	}))
+}
+
+func TestPollZenmoneyWritesMoneyFacts(t *testing.T) {
+	body, err := os.ReadFile("../../internal/zenmoney/testdata/diff.json")
+	if err != nil {
+		t.Fatal(err)
+	}
+	srv := zenFixtureServer(t, body, http.StatusOK)
+	defer srv.Close()
+	zen, err := zenmoney.New("tok", srv.URL, time.Second)
+	if err != nil {
+		t.Fatal(err)
+	}
+	core := &factCore{}
+	p := &poller{core: core, zen: zen}
+	now := time.Date(2026, 8, 1, 21, 0, 0, 0, time.UTC)
+	if err := p.pollZenmoney(context.Background(), now); err != nil {
+		t.Fatal(err)
+	}
+	if len(core.written) != 2 {
+		t.Fatalf("wrote %d facts, want today + month", len(core.written))
+	}
+	for _, f := range core.written {
+		if f.Kind != "env" || f.Source != zenmoney.Source {
+			t.Errorf("fact = %+v, want kind=env source=%s", f, zenmoney.Source)
+		}
+		if _, err := zenmoney.ParseFactValue(f.Value); err != nil {
+			t.Errorf("fact value %q does not decode: %v", f.Value, err)
+		}
+	}
+}
+
+// A read that returns nothing for the window writes NOTHING. Silence, not a
+// zero: an invented 0 would be recited back to him as fact.
+func TestPollZenmoneyWritesNothingWhenEmpty(t *testing.T) {
+	srv := zenFixtureServer(t, []byte(`{"serverTimestamp":1,"instrument":[],"transaction":[]}`), http.StatusOK)
+	defer srv.Close()
+	zen, _ := zenmoney.New("tok", srv.URL, time.Second)
+	core := &factCore{}
+	p := &poller{core: core, zen: zen}
+	if err := p.pollZenmoney(context.Background(), time.Now()); err != nil {
+		t.Fatal(err)
+	}
+	if len(core.written) != 0 {
+		t.Errorf("wrote %+v, want no fact at all", core.written)
+	}
+}
+
+// An API failure must not overwrite the last good total either.
+func TestPollZenmoneyFailureWritesNothing(t *testing.T) {
+	srv := zenFixtureServer(t, nil, http.StatusUnauthorized)
+	defer srv.Close()
+	zen, _ := zenmoney.New("bad", srv.URL, time.Second)
+	core := &factCore{}
+	p := &poller{core: core, zen: zen}
+	if err := p.pollZenmoney(context.Background(), time.Now()); err == nil {
+		t.Error("want the 401 reported")
+	}
+	if len(core.written) != 0 {
+		t.Errorf("wrote %+v on a failed read", core.written)
+	}
+}
+
+// Unchanged totals do not churn the facts table.
+func TestWriteIfChangedRawSkipsUnchanged(t *testing.T) {
+	core := &factCore{prev: map[string]string{
+		zenmoney.KeySpentToday + "|" + zenmoney.Source: `{"count":1}`,
+	}}
+	p := &poller{core: core}
+	if err := p.writeIfChangedRaw(context.Background(), zenmoney.KeySpentToday, zenmoney.Source, `{"count":1}`, time.Now()); err != nil {
+		t.Fatal(err)
+	}
+	if len(core.written) != 0 {
+		t.Errorf("wrote %+v for an unchanged value", core.written)
+	}
+}
+
+// Money tracking is off unless configured: no token file, no zenmoney client,
+// and the poller still refuses to start with nothing at all to poll.
+func TestRunRequiresSomethingToPoll(t *testing.T) {
+	err := run([]string{"-socket", "/tmp/nope.sock", "-netdata", "", "-kuma", "", "-wg", ""})
+	if err == nil || !strings.Contains(err.Error(), "nothing to poll") {
+		t.Errorf("err = %v, want a 'nothing to poll' refusal", err)
+	}
+}

cmd/mavweb/ambient.go

@@ -0,0 +1,135 @@
+package main
+
+import (
+	"crypto/subtle"
+	"encoding/json"
+	"errors"
+	"io"
+	"log"
+	"net/http"
+	"strings"
+
+	"github.com/kami/maven/internal/calendar"
+	"github.com/kami/maven/internal/ipc"
+)
+
+// POST /api/ambient — the work calendar read (Vikunja #126).
+//
+// Maven does not hold a work credential. A corp mail or calendar session on the
+// homelab ties the box's blast radius to the employer's data, so the work
+// calendar is read as a SIGNAL instead: an Android notification-listener on the
+// owner's phone posts meeting notifications here over wg/LAN, and the ones that
+// clearly describe a meeting become calendar events at source=ambient:notif,
+// confidence below 1.0. Mail as a notification signal, not a mailbox.
+//
+// Off unless configured: no -ambient-token, no route. The token is a shared
+// secret because the poster is a phone service, not a browser — WebAuthn has no
+// answer for a background Android service. The endpoint is write-only and
+// accepts exactly one shape of write; it cannot read anything back out.
+//
+// A notification with no recognisable clock reading stores NOTHING. Maven is
+// not a guesser-of-truth, and a mailbox of noise rendered as invented meetings
+// is worse than a gap.
+
+// ambientMaxBody bounds the request. A notification is two short lines.
+const ambientMaxBody = 8 << 10
+
+type ambientResp struct {
+	Stored bool   `json:"stored"`
+	Key    string `json:"key,omitempty"`
+	Reason string `json:"reason,omitempty"`
+}
+
+// handleAmbient ingests one relayed notification. token is the configured
+// shared secret; an empty token means the capability is off and the handler is
+// never registered, so it is treated as a hard failure here too.
+func handleAmbient(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI, token string) {
+	if r.Method != http.MethodPost {
+		http.Error(w, "POST only", http.StatusMethodNotAllowed)
+		return
+	}
+	if token == "" {
+		http.Error(w, "ambient ingest disabled (no -ambient-token)", http.StatusServiceUnavailable)
+		return
+	}
+	if !ambientAuthorized(r, token) {
+		http.Error(w, "unauthorized", http.StatusUnauthorized)
+		return
+	}
+	if core == nil {
+		http.Error(w, "ambient ingest disabled (no -core)", http.StatusServiceUnavailable)
+		return
+	}
+
+	var n calendar.Notification
+	body, err := io.ReadAll(io.LimitReader(r.Body, ambientMaxBody))
+	if err != nil {
+		http.Error(w, "read failed", http.StatusBadRequest)
+		return
+	}
+	if err := json.Unmarshal(body, &n); err != nil {
+		http.Error(w, "bad json", http.StatusBadRequest)
+		return
+	}
+	if n.Posted.IsZero() {
+		writeAmbient(w, http.StatusBadRequest, ambientResp{Reason: "posted_at is required"})
+		return
+	}
+
+	ev, ok := calendar.EventFromNotification(n)
+	if !ok {
+		// Not an event. 202: the relay did its job, there is just nothing here
+		// worth remembering, and it must not retry.
+		writeAmbient(w, http.StatusAccepted, ambientResp{Reason: "no meeting time in notification"})
+		return
+	}
+
+	key := calendar.FactKey(ev)
+	val := calendar.FactValue(ev)
+
+	// Append-only discipline, same as cmd/mavcaldav: a phone reposts the same
+	// notification many times, and each repost is the same event.
+	if prev, err := core.LatestFactBySource(r.Context(), key, calendar.SourceAmbient); err == nil && prev.Value == val {
+		writeAmbient(w, http.StatusOK, ambientResp{Stored: false, Key: key, Reason: "unchanged"})
+		return
+	} else if err != nil && !errors.Is(err, ipc.ErrNoFact) {
+		log.Printf("ambient: read %s: %v", key, err)
+		http.Error(w, "read failed", http.StatusBadGateway)
+		return
+	}
+
+	// kind=env: an observation about the world, never a self-fact — a passive
+	// signal does not write truth about the owner. Confidence below 1.0 is the
+	// honest part: this is a notification about a meeting, not a reading of a
+	// calendar, and the query path hedges when it recites one.
+	if _, err := core.WriteFact(r.Context(), ipc.WriteFactReq{
+		Ts:         ev.Start,
+		Kind:       "env",
+		Key:        key,
+		Value:      val,
+		Source:     calendar.SourceAmbient,
+		Confidence: calendar.AmbientConfidence,
+	}); err != nil {
+		log.Printf("ambient: write %s: %v", key, err)
+		http.Error(w, "write failed", http.StatusBadGateway)
+		return
+	}
+	log.Printf("ambient: %s=%s (%s, pkg=%s)", key, val, calendar.SourceAmbient, n.Package)
+	writeAmbient(w, http.StatusCreated, ambientResp{Stored: true, Key: key})
+}
+
+// ambientAuthorized accepts the token as a bearer header or as an X-Maven-Token
+// header, compared in constant time.
+func ambientAuthorized(r *http.Request, token string) bool {
+	got := strings.TrimSpace(strings.TrimPrefix(r.Header.Get("Authorization"), "Bearer"))
+	if got == "" {
+		got = strings.TrimSpace(r.Header.Get("X-Maven-Token"))
+	}
+	return subtle.ConstantTimeCompare([]byte(got), []byte(token)) == 1
+}
+
+func writeAmbient(w http.ResponseWriter, code int, resp ambientResp) {
+	w.Header().Set("Content-Type", "application/json")
+	w.WriteHeader(code)
+	json.NewEncoder(w).Encode(resp)
+}

cmd/mavweb/ambient_test.go

@@ -0,0 +1,223 @@
+package main
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"net/http"
+	"net/http/httptest"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/calendar"
+	"github.com/kami/maven/internal/ipc"
+)
+
+const ambientTestToken = "s3cret"
+
+// ambientCore adds provenance-scoped reads to fakeCore, which the dedupe path
+// needs.
+type ambientCore struct {
+	fakeCore
+	latest  map[string]ipc.Fact // "key|source" → fact
+	readErr error
+}
+
+func (c *ambientCore) LatestFactBySource(_ context.Context, key, source string) (ipc.Fact, error) {
+	if c.readErr != nil {
+		return ipc.Fact{}, c.readErr
+	}
+	f, ok := c.latest[key+"|"+source]
+	if !ok {
+		return ipc.Fact{}, ipc.ErrNoFact
+	}
+	return f, nil
+}
+
+func postAmbient(t *testing.T, core ipc.CoreAPI, token string, n calendar.Notification) (*httptest.ResponseRecorder, ambientResp) {
+	t.Helper()
+	body, err := json.Marshal(n)
+	if err != nil {
+		t.Fatal(err)
+	}
+	req := httptest.NewRequest(http.MethodPost, "/api/ambient", strings.NewReader(string(body)))
+	req.Header.Set("Authorization", "Bearer "+ambientTestToken)
+	rr := httptest.NewRecorder()
+	handleAmbient(rr, req, core, token)
+	var resp ambientResp
+	json.Unmarshal(rr.Body.Bytes(), &resp)
+	return rr, resp
+}
+
+func meetingNotification() calendar.Notification {
+	return calendar.Notification{
+		Package: "com.google.android.gm",
+		Title:   "Планёрка",
+		Text:    "10:00-10:30",
+		Posted:  time.Date(2026, 8, 3, 9, 40, 0, 0, time.UTC),
+	}
+}
+
+func TestHandleAmbientStoresMeeting(t *testing.T) {
+	core := &ambientCore{}
+	rr, resp := postAmbient(t, core, ambientTestToken, meetingNotification())
+
+	if rr.Code != http.StatusCreated {
+		t.Fatalf("status = %d, want 201: %s", rr.Code, rr.Body)
+	}
+	if !resp.Stored {
+		t.Errorf("resp = %+v, want stored", resp)
+	}
+	if len(core.writeLog) != 1 {
+		t.Fatalf("expected 1 fact write, got %d", len(core.writeLog))
+	}
+	got := core.writeLog[0]
+	if got.Source != calendar.SourceAmbient {
+		t.Errorf("source = %q, want %q", got.Source, calendar.SourceAmbient)
+	}
+	if got.Confidence >= 1.0 {
+		t.Errorf("confidence = %v — a notification is not a calendar read", got.Confidence)
+	}
+	if got.Confidence != calendar.AmbientConfidence {
+		t.Errorf("confidence = %v, want %v", got.Confidence, calendar.AmbientConfidence)
+	}
+	if got.Kind != "env" {
+		t.Errorf("kind = %q — a passive signal never writes a self-fact", got.Kind)
+	}
+	if want := "calendar_event_20260803_"; !strings.HasPrefix(got.Key, want) {
+		t.Errorf("key = %q, want prefix %q", got.Key, want)
+	}
+	if got.Value != "Планёрка @ 10:00-10:30" {
+		t.Errorf("value = %q", got.Value)
+	}
+}
+
+// A phone reposts the same notification many times. Each repost is the same
+// event, and the append-only log must not fill with duplicates.
+func TestHandleAmbientDedupesReposts(t *testing.T) {
+	core := &ambientCore{}
+	postAmbient(t, core, ambientTestToken, meetingNotification())
+	if len(core.writeLog) != 1 {
+		t.Fatalf("first post did not write")
+	}
+	w := core.writeLog[0]
+	core.latest = map[string]ipc.Fact{w.Key + "|" + w.Source: {Value: w.Value}}
+
+	rr, resp := postAmbient(t, core, ambientTestToken, meetingNotification())
+	if rr.Code != http.StatusOK {
+		t.Errorf("status = %d, want 200 for an unchanged repost", rr.Code)
+	}
+	if resp.Stored {
+		t.Error("a repost must not be stored again")
+	}
+	if len(core.writeLog) != 1 {
+		t.Errorf("wrote %d facts, want 1", len(core.writeLog))
+	}
+}
+
+// The conservative half: noise stores nothing at all.
+func TestHandleAmbientIgnoresNonMeetings(t *testing.T) {
+	core := &ambientCore{}
+	rr, resp := postAmbient(t, core, ambientTestToken, calendar.Notification{
+		Package: "com.google.android.gm",
+		Title:   "3 новых письма",
+		Posted:  time.Now(),
+	})
+	if rr.Code != http.StatusAccepted {
+		t.Errorf("status = %d, want 202 (accepted, nothing to store — the relay must not retry)", rr.Code)
+	}
+	if resp.Stored {
+		t.Error("a notification with no meeting time must store nothing")
+	}
+	if len(core.writeLog) != 0 {
+		t.Fatalf("wrote %d facts for a non-meeting", len(core.writeLog))
+	}
+}
+
+func TestHandleAmbientAuth(t *testing.T) {
+	body := `{"title":"Планёрка 10:00","posted_at":"2026-08-03T09:40:00Z"}`
+
+	newReq := func(hdr, val string) *http.Request {
+		r := httptest.NewRequest(http.MethodPost, "/api/ambient", strings.NewReader(body))
+		if hdr != "" {
+			r.Header.Set(hdr, val)
+		}
+		return r
+	}
+
+	t.Run("no token rejected", func(t *testing.T) {
+		core := &ambientCore{}
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, newReq("", ""), core, ambientTestToken)
+		if rr.Code != http.StatusUnauthorized {
+			t.Errorf("status = %d, want 401", rr.Code)
+		}
+		if len(core.writeLog) != 0 {
+			t.Error("an unauthorized post must not write")
+		}
+	})
+
+	t.Run("wrong token rejected", func(t *testing.T) {
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, newReq("Authorization", "Bearer nope"), &ambientCore{}, ambientTestToken)
+		if rr.Code != http.StatusUnauthorized {
+			t.Errorf("status = %d, want 401", rr.Code)
+		}
+	})
+
+	t.Run("X-Maven-Token accepted", func(t *testing.T) {
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, newReq("X-Maven-Token", ambientTestToken), &ambientCore{}, ambientTestToken)
+		if rr.Code != http.StatusCreated {
+			t.Errorf("status = %d, want 201: %s", rr.Code, rr.Body)
+		}
+	})
+
+	t.Run("capability off", func(t *testing.T) {
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, newReq("Authorization", "Bearer "+ambientTestToken), &ambientCore{}, "")
+		if rr.Code != http.StatusServiceUnavailable {
+			t.Errorf("status = %d, want 503 when no token is configured", rr.Code)
+		}
+	})
+
+	t.Run("GET rejected", func(t *testing.T) {
+		rr := httptest.NewRecorder()
+		r := httptest.NewRequest(http.MethodGet, "/api/ambient", nil)
+		handleAmbient(rr, r, &ambientCore{}, ambientTestToken)
+		if rr.Code != http.StatusMethodNotAllowed {
+			t.Errorf("status = %d, want 405 — the ingest is write-only", rr.Code)
+		}
+	})
+}
+
+func TestHandleAmbientBadInput(t *testing.T) {
+	t.Run("bad json", func(t *testing.T) {
+		req := httptest.NewRequest(http.MethodPost, "/api/ambient", strings.NewReader("{nope"))
+		req.Header.Set("X-Maven-Token", ambientTestToken)
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, req, &ambientCore{}, ambientTestToken)
+		if rr.Code != http.StatusBadRequest {
+			t.Errorf("status = %d, want 400", rr.Code)
+		}
+	})
+
+	t.Run("missing posted_at", func(t *testing.T) {
+		req := httptest.NewRequest(http.MethodPost, "/api/ambient", strings.NewReader(`{"title":"Планёрка 10:00"}`))
+		req.Header.Set("X-Maven-Token", ambientTestToken)
+		rr := httptest.NewRecorder()
+		handleAmbient(rr, req, &ambientCore{}, ambientTestToken)
+		if rr.Code != http.StatusBadRequest {
+			t.Errorf("status = %d, want 400", rr.Code)
+		}
+	})
+
+	t.Run("read error surfaces", func(t *testing.T) {
+		core := &ambientCore{readErr: fmt.Errorf("socket closed")}
+		rr, _ := postAmbient(t, core, ambientTestToken, meetingNotification())
+		if rr.Code != http.StatusBadGateway {
+			t.Errorf("status = %d, want 502", rr.Code)
+		}
+	})
+}

cmd/mavweb/handlers_test.go

@@ -17,11 +17,12 @@ import (
 )
 
 // fakeCore records the mutating calls handleTools makes and returns canned
-// tool lists / errors. Embedding ipc.CoreAPI (nil) satisfies the large
+// tool lists / errors. Embedding ipc.UnimplementedCoreAPI satisfies the large
 // interface — only the methods the handlers touch are overridden; any other
-// call would nil-panic, which is fine since the handlers never make them.
+// call returns ipc.ErrNotImplemented instead of nil-panicking, so a test that
+// accidentally exercises an undeclared method fails loudly.
 type fakeCore struct {
-	ipc.CoreAPI
+	ipc.UnimplementedCoreAPI
 
 	proposed, enabled []ipc.Tool
 	listErr           error
@@ -62,6 +63,18 @@ type fakeCore struct {
 	// for handleTrace tests
 	tickTrace ipc.TickTrace
 	traceErr  error
+
+	// for handleChatAPI tests
+	chatText string
+	chatErr  error
+}
+
+func (f *fakeCore) Chat(_ context.Context, text string) (string, error) {
+	f.chatText = text
+	if f.chatErr != nil {
+		return "", f.chatErr
+	}
+	return "поняла", nil
 }
 
 func (f *fakeCore) EnableTool(_ context.Context, name string, cmd []string, destructive bool, scope string, _ time.Time) error {
@@ -1044,3 +1057,64 @@ func TestHandleRoutines_NilCore_503(t *testing.T) {
 		t.Fatalf("status = %d, want 503", rr.Code)
 	}
 }
+
+// --- handleChatAPI step-up gate (Vikunja #317) ---
+//
+// POST /api/chat reaches the router, the LLM and the act path, so it carries
+// the same gate as POST /tools and POST /api/revert.
+
+func postChat(text string) *http.Request {
+	req := httptest.NewRequest(http.MethodPost, "/api/chat", strings.NewReader("text="+url.QueryEscape(text)))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	return req
+}
+
+func TestHandleChatAPI_RequireStepUp_FailsClosed(t *testing.T) {
+	core := &fakeCore{}
+	rr := httptest.NewRecorder()
+	handleChatAPI(rr, postChat("выключи свет"), core, nil, true)
+	if rr.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403; body=%s", rr.Code, rr.Body.String())
+	}
+	if core.chatText != "" {
+		t.Errorf("core.Chat called with %q, but -require-stepup should deny", core.chatText)
+	}
+}
+
+func TestHandleChatAPI_UnassertedSession_Denied(t *testing.T) {
+	core := &fakeCore{}
+	rr := httptest.NewRecorder()
+	handleChatAPI(rr, postChat("выключи свет"), core, webauthn.NewPasskeySession(5*time.Minute), false)
+	if rr.Code != http.StatusForbidden {
+		t.Fatalf("status = %d, want 403", rr.Code)
+	}
+	if core.chatText != "" {
+		t.Errorf("core.Chat called with %q despite an unasserted session", core.chatText)
+	}
+}
+
+func TestHandleChatAPI_AssertedSession_PassesGate(t *testing.T) {
+	core := &fakeCore{}
+	rr := httptest.NewRecorder()
+	handleChatAPI(rr, postChat("привет"), core, stepUpSession(), true)
+	if rr.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303; body=%s", rr.Code, rr.Body.String())
+	}
+	if core.chatText != "привет" {
+		t.Errorf("core.Chat text = %q, want %q", core.chatText, "привет")
+	}
+}
+
+// Default deploy: WebAuthn unconfigured and -require-stepup off ⇒ chat keeps
+// working, resting on the transport-level auth in front of mavweb.
+func TestHandleChatAPI_FailOpenByDefault(t *testing.T) {
+	core := &fakeCore{}
+	rr := httptest.NewRecorder()
+	handleChatAPI(rr, postChat("привет"), core, nil, false)
+	if rr.Code != http.StatusSeeOther {
+		t.Fatalf("status = %d, want 303", rr.Code)
+	}
+	if core.chatText != "привет" {
+		t.Errorf("core.Chat text = %q, want %q", core.chatText, "привет")
+	}
+}

cmd/mavweb/main.go

@@ -25,6 +25,7 @@ import (
 	"github.com/kami/maven/internal/audio"
 	"github.com/kami/maven/internal/ipc"
 	"github.com/kami/maven/internal/pattern"
+	"github.com/kami/maven/internal/tasks"
 	"github.com/kami/maven/internal/voice"
 	"github.com/kami/maven/internal/webauthn"
 )
@@ -58,6 +59,9 @@ var notificationsHTML string
 //go:embed reminders.html
 var remindersHTML string
 
+//go:embed tasks.html
+var tasksHTML string
+
 //go:embed voice.html
 var voiceHTML string
 
@@ -97,6 +101,7 @@ var sidebarSections = []struct {
 		Pages: []struct{ Label, URL, Key string }{
 			{Label: "Rule Trace", URL: "/trace", Key: "trace"},
 			{Label: "Notifications", URL: "/notifications", Key: "notifications"},
+			{Label: "Tasks", URL: "/tasks", Key: "tasks"},
 			{Label: "Reminders", URL: "/reminders", Key: "reminders"},
 			{Label: "Routines", URL: "/routines", Key: "routines"},
 			{Label: "Morning", URL: "/morning", Key: "morning"},
@@ -170,6 +175,8 @@ func pageIcon(key string) string {
 		return `<svg class=icon width="14" height="14"><use href="/ethos-icons.svg#i-wave"/></svg>`
 	case "notifications":
 		return `<svg class=icon width="14" height="14"><use href="/ethos-icons.svg#i-bell"/></svg>`
+	case "tasks":
+		return `<svg class=icon width="14" height="14"><use href="/ethos-icons.svg#i-grid"/></svg>`
 	case "reminders":
 		return `<svg class=icon width="14" height="14"><use href="/ethos-icons.svg#i-calendar"/></svg>`
 	case "routines":
@@ -202,6 +209,8 @@ func pageTitle(key string) string {
 		return "Rule Trace"
 	case "notifications":
 		return "Notifications"
+	case "tasks":
+		return "Tasks"
 	case "reminders":
 		return "Reminders"
 	case "routines":
@@ -314,7 +323,7 @@ func noCache(h http.Handler) http.Handler {
 }
 
 func main() {
-	addr := flag.String("addr", ":9200", "HTTP listen address")
+	addr := flag.String("addr", "127.0.0.1:9200", "HTTP listen address (loopback by default; pass e.g. \":9200\" or a LAN IP deliberately for wider exposure — POST /chat and /routines are state-changing)")
 	voiceAddr := flag.String("voice", "127.0.0.1:9100", "voice server TCP addr (host:port)")
 	// ntfyWS: the ntfy WebSocket subscribe URL the PWA connects to for in-app
 	// nudge delivery, e.g. wss://ntfy.kvmx.ru/maven/ws?auth=<base64-token>. The
@@ -329,11 +338,15 @@ func main() {
 	coreSock := flag.String("core", "", "mavend IPC socket path for presence-signal ingest (empty = disabled)")
 	pkOrigin := flag.String("webauthn-origin", "", "WebAuthn origin URL (e.g. https://maven.kvmx.ru)")
 	pkRPID := flag.String("webauthn-rpid", "", "WebAuthn RP ID (e.g. maven.kvmx.ru)")
-	requireStepUp := flag.Bool("require-stepup", false, "fail closed on step-up-gated actions (/tools POST, /api/revert) when WebAuthn step-up cannot be asserted; default false preserves the historical fail-open behaviour")
+	requireStepUp := flag.Bool("require-stepup", false, "fail closed on step-up-gated actions (POST /tools, /routines, /api/revert, /api/chat) when WebAuthn step-up cannot be asserted; default false preserves the historical fail-open behaviour")
 	pkFile := flag.String("passkey-file", "./passkeys.json", "path to WebAuthn credential store (JSON)")
 	nexusURL := flag.String("nexus", "", "Nexus base URL for the /ecosystem panel (empty = not configured)")
 	praxisURL := flag.String("praxis", "", "Praxis base URL for the /ecosystem panel (empty = not configured)")
 	hexisURL := flag.String("hexis", "", "Hexis base URL for the /ecosystem panel (empty = not configured)")
+	// Shared secret for POST /api/ambient, the notification-relay ingest that
+	// reads the work calendar as a signal instead of holding a work credential
+	// (see ambient.go). Empty ⇒ the route is not registered at all.
+	ambientToken := flag.String("ambient-token", "", "shared secret for POST /api/ambient notification ingest (empty = ingest disabled, route not registered)")
 	flag.Parse()
 
 	var core ipc.CoreAPI
@@ -381,6 +394,14 @@ func main() {
 	mux.HandleFunc("/api/signal", func(w http.ResponseWriter, r *http.Request) {
 		handleSignal(w, r, core)
 	})
+	// Off unless configured: no token, no route — an unconfigured ingest is not
+	// a 503 waiting to be probed, it does not exist.
+	if *ambientToken != "" {
+		mux.HandleFunc("/api/ambient", func(w http.ResponseWriter, r *http.Request) {
+			handleAmbient(w, r, core, *ambientToken)
+		})
+		log.Printf("mavweb: ambient notification ingest enabled at POST /api/ambient")
+	}
 	mux.HandleFunc("/dash", func(w http.ResponseWriter, r *http.Request) {
 		handleDash(w, r, core)
 	})
@@ -396,6 +417,11 @@ func main() {
 	mux.HandleFunc("/reminders", func(w http.ResponseWriter, r *http.Request) {
 		handleReminders(w, r, core)
 	})
+	// /tasks — capture + review. POST is not step-up gated; see handleTasks for
+	// why a task write is not in the same class as /tools or /routines.
+	mux.HandleFunc("/tasks", func(w http.ResponseWriter, r *http.Request) {
+		handleTasks(w, r, core)
+	})
 	mux.HandleFunc("/morning", func(w http.ResponseWriter, r *http.Request) {
 		handleMorning(w, r, core)
 	})
@@ -434,9 +460,9 @@ func main() {
 	}
 	if stepUpSession == nil {
 		if *requireStepUp {
-			log.Printf("SECURITY: step-up verification is DISABLED (-webauthn-origin/-webauthn-rpid unset) and -require-stepup is set: POST /tools (tool enable/disable/dismiss — defines and executes arbitrary argv) and POST /api/revert will be DENIED (403). Set -webauthn-origin and -webauthn-rpid to enable passkey step-up.")
+			log.Printf("SECURITY: step-up verification is DISABLED (-webauthn-origin/-webauthn-rpid unset) and -require-stepup is set: POST /tools (tool enable/disable/dismiss — defines and executes arbitrary argv), POST /routines (accepting schedules recurring firing), POST /api/revert and POST /api/chat (reaches the router, the LLM and the act path) will be DENIED (403). Set -webauthn-origin and -webauthn-rpid to enable passkey step-up.")
 		} else {
-			log.Printf("SECURITY WARNING: step-up verification is DISABLED because -webauthn-origin/-webauthn-rpid are unset. UNGUARDED SURFACES: POST /tools (defines arbitrary argv via name+cmd, which internal/tool then EXECUTES) and POST /api/revert (voids the latest fact for a key). These are protected only by whatever transport-level auth sits in front of mavweb (wg+nginx+auth) — do NOT expose -addr on a public interface. Set -webauthn-origin and -webauthn-rpid to require passkey step-up, or pass -require-stepup to fail closed instead.")
+			log.Printf("SECURITY WARNING: step-up verification is DISABLED because -webauthn-origin/-webauthn-rpid are unset. UNGUARDED SURFACES: POST /tools (defines arbitrary argv via name+cmd, which internal/tool then EXECUTES), POST /routines (accepting schedules recurring firing), POST /api/revert (voids the latest fact for a key) and POST /api/chat (reaches the router, the LLM and, through applyAction, the act path). These are protected only by whatever transport-level auth sits in front of mavweb (wg+nginx+auth) — do NOT expose -addr on a public interface. Set -webauthn-origin and -webauthn-rpid to require passkey step-up, or pass -require-stepup to fail closed instead.")
 		}
 	}
 
@@ -454,14 +480,26 @@ func main() {
 		handleRoutines(w, r, core, stepUpSession, *requireStepUp)
 	})
 
-	// /api/revert voids the latest fact for a key — a store mutation, so it
+	// State-changing routes on this server, and their gate (Vikunja #317):
-	// sits behind the same passkey step-up as tool enable (nil session ⇒
+	//
-	// WebAuthn unconfigured ⇒ transport-level auth only, same as /tools).
+	//   POST /tools      step-up  — defines argv that internal/tool executes
+	//   POST /routines   step-up  — accepting schedules recurring firing
+	//   POST /api/revert step-up  — voids the latest fact for a key
+	//   POST /api/chat   step-up  — reaches the router, LLM and the act path
+	//   POST /api/signal none     — appends a presence fact, no argv, no act
+	//   POST /api/ptt, /ws  none  — proxy audio to mavend's voice port, which
+	//                               is itself only reachable inside the deploy
+	//
+	// "step-up" means stepUpOK: asserted passkey when WebAuthn is configured,
+	// otherwise fail-open unless -require-stepup, which denies.
+	//
+	// GET /chat only renders the page and echoes back the q/r query params the
+	// POST redirect set — nothing to gate.
 	mux.HandleFunc("/chat", func(w http.ResponseWriter, r *http.Request) {
 		handleChatPage(w, r, core)
 	})
 	mux.HandleFunc("/api/chat", func(w http.ResponseWriter, r *http.Request) {
-		handleChatAPI(w, r, core)
+		handleChatAPI(w, r, core, stepUpSession, *requireStepUp)
 	})
 	mux.HandleFunc("/api/revert", func(w http.ResponseWriter, r *http.Request) {
 		handleRevert(w, r, core, stepUpSession, *requireStepUp)
@@ -720,6 +758,8 @@ var passkeyTmpl = template.Must(template.New("passkey").Funcs(shellFuncs()).Pars
 
 var voiceTmpl = template.Must(template.New("voice").Funcs(shellFuncs()).Parse(shellTopHTML + voiceHTML + shellBottomHTML))
 
+var tasksTmpl = template.Must(template.New("tasks").Funcs(shellFuncs()).Parse(shellTopHTML + tasksHTML + shellBottomHTML))
+
 var routinesTmpl = template.Must(template.New("routines").Funcs(shellFuncs()).Parse(shellTopHTML + routinesHTML + shellBottomHTML))
 
 var traceTmpl = template.Must(template.New("trace").Funcs(func() template.FuncMap {
@@ -790,6 +830,187 @@ func handleReminders(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
 	}
 }
 
+// taskRow is one line on /tasks, with every timestamp already formatted so the
+// template holds no date logic.
+type taskRow struct {
+	ID       int64
+	Text     string
+	Source   string
+	Evidence string
+	Status   string
+	Due      string
+	Created  string
+	Resolved string
+	// Why — the ranker's reason for this row's position (Vikunja #129), in
+	// Russian, empty when nothing distinguished the task. Blank is the honest
+	// rendering: he never said this one mattered more.
+	Why string
+}
+
+// handleTasks serves the task review surface (GET) and the four writes it
+// offers (POST): add, confirm, done, drop.
+//
+// Not step-up gated, unlike /tools and /routines, and the difference is the
+// point: enabling a tool defines argv Maven will execute, and accepting a
+// routine hands the tick loop a new standing reason to interrupt him. A task is
+// neither — nothing in the tick loop reads the tasks table, so the worst a
+// weaker caller can do here is write a line onto a list he reads himself. It
+// still sits behind whatever transport auth fronts mavweb, like every other
+// page.
+//
+// "confirm" is the only interesting move: it promotes a candidate Maven derived
+// from something she read into work he owns. That review step is why derived
+// tasks are captured as candidates in the first place.
+func handleTasks(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
+	if core == nil {
+		http.Error(w, "tasks disabled (no -core)", http.StatusServiceUnavailable)
+		return
+	}
+	ctx := r.Context()
+	var msg, errMsg string
+	if r.Method == http.MethodPost {
+		var err error
+		msg, err = applyTaskPost(ctx, core, r)
+		if err != nil {
+			log.Printf("tasks: %v", err)
+			errMsg = err.Error()
+		}
+	}
+
+	all, err := core.ListTasks(ctx, "")
+	if err != nil {
+		log.Printf("tasks: list: %v", err)
+		http.Error(w, "tasks error: "+err.Error(), http.StatusBadGateway)
+		return
+	}
+	// Live rows are ordered by the same ranker the spoken list uses, so the page
+	// and the voice reply can never disagree about what comes first. Resolved
+	// rows keep store order (newest first) — ranking finished work is pointless.
+	var live []tasks.Item
+	var resolved []taskRow
+	for _, t := range all {
+		switch t.Status {
+		case "candidate", "open":
+			live = append(live, tasks.Item{
+				ID: t.ID, Text: t.Text, Status: t.Status,
+				Created: t.CreatedTs, Due: t.Due, Weight: t.Weight,
+			})
+		default:
+			resolved = append(resolved, taskRow{
+				ID: t.ID, Text: t.Text, Source: t.Source, Evidence: t.Evidence,
+				Status: t.Status, Created: fmtTaskTime(&t.CreatedTs),
+				Due: fmtTaskDate(t.Due), Resolved: fmtTaskTime(t.Resolved),
+			})
+		}
+	}
+	byID := make(map[int64]ipc.Task, len(all))
+	for _, t := range all {
+		byID[t.ID] = t
+	}
+	var cands, open []taskRow
+	for _, r := range tasks.Rank(live, time.Now()) {
+		t := byID[r.ID]
+		row := taskRow{
+			ID: t.ID, Text: t.Text, Source: t.Source, Evidence: t.Evidence,
+			Status: t.Status, Created: fmtTaskTime(&t.CreatedTs),
+			Due: fmtTaskDate(t.Due), Resolved: fmtTaskTime(t.Resolved),
+			Why: r.Reason,
+		}
+		if t.Status == "candidate" {
+			// A candidate's due date is Maven's reading of a mail, so its
+			// ranking reason is not shown as if he had set a priority.
+			row.Why = ""
+			cands = append(cands, row)
+		} else {
+			open = append(open, row)
+		}
+	}
+	w.Header().Set("Content-Type", "text/html; charset=utf-8")
+	if err := tasksTmpl.Execute(w, struct {
+		Msg, Err   string
+		Candidates []taskRow
+		Open       []taskRow
+		Resolved   []taskRow
+	}{msg, errMsg, cands, open, resolved}); err != nil {
+		log.Printf("tasks render: %v", err)
+	}
+}
+
+// applyTaskPost performs one write and returns the message to show. A bad
+// request returns an error, which the page renders inline rather than as a
+// bare 400 — this is a form surface, not an API.
+func applyTaskPost(ctx context.Context, core ipc.CoreAPI, r *http.Request) (string, error) {
+	action := r.FormValue("action")
+	if action == "add" {
+		text := strings.TrimSpace(r.FormValue("text"))
+		if text == "" {
+			return "", errors.New("empty task text")
+		}
+		req := ipc.CaptureTaskReq{Text: text, Source: "tap:web", Status: "open", Ts: time.Now()}
+		// Importance is his, stated on the form. Out-of-range values are
+		// clamped rather than rejected — a bad select is not worth a 400.
+		if v := r.FormValue("weight"); v != "" {
+			var wgt int
+			if n, _ := fmt.Sscanf(v, "%d", &wgt); n != 1 || wgt < 0 {
+				return "", fmt.Errorf("bad weight %q", v)
+			}
+			if wgt > tasks.MaxWeight {
+				wgt = tasks.MaxWeight
+			}
+			req.Weight = wgt
+		}
+		if d := r.FormValue("due"); d != "" {
+			due, err := time.ParseInLocation("2006-01-02", d, time.Local)
+			if err != nil {
+				return "", fmt.Errorf("bad due date %q", d)
+			}
+			req.Due = &due
+		}
+		resp, err := core.CaptureTask(ctx, req)
+		if err != nil {
+			return "", err
+		}
+		if !resp.Created {
+			return "already on the list", nil
+		}
+		return "added task", nil
+	}
+
+	var id int64
+	if n, _ := fmt.Sscanf(r.FormValue("id"), "%d", &id); n != 1 {
+		return "", errors.New("invalid id")
+	}
+	var status, msg string
+	switch action {
+	case "confirm":
+		status, msg = "open", "confirmed task"
+	case "done":
+		status, msg = "done", "task done"
+	case "drop":
+		status, msg = "dropped", "dropped task"
+	default:
+		return "", fmt.Errorf("unknown action %q", action)
+	}
+	if err := core.SetTaskStatus(ctx, id, status, time.Now()); err != nil {
+		return "", err
+	}
+	return msg, nil
+}
+
+func fmtTaskTime(t *time.Time) string {
+	if t == nil || t.IsZero() {
+		return "—"
+	}
+	return t.Local().Format("02 Jan 15:04")
+}
+
+func fmtTaskDate(t *time.Time) string {
+	if t == nil || t.IsZero() {
+		return "—"
+	}
+	return t.Local().Format("02 Jan")
+}
+
 // routineRow is one line on the page: what maven noticed, in her words, and
 // how long ago she noticed it.
 type routineRow struct {
@@ -935,12 +1156,32 @@ func handleMorning(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
 		http.Error(w, "core read failed", http.StatusBadGateway)
 		return
 	}
+	view := morningView{Routines: status}
+	// The day plan (#128) shows on this page because it is the same question at
+	// a different scale. A plan read that fails must not take the checklist
+	// down with it — the page degrades to what it had before.
+	plan, err := core.DayPlan(ctx)
+	if err != nil {
+		log.Printf("morning: day plan: %v", err)
+		view.PlanErr = err.Error()
+	} else {
+		view.Plan = &plan
+	}
 	w.Header().Set("Content-Type", "text/html; charset=utf-8")
-	if err := morningTmpl.Execute(w, status); err != nil {
+	if err := morningTmpl.Execute(w, view); err != nil {
 		log.Printf("morning render: %v", err)
 	}
 }
 
+// morningView — what /morning renders: today's plan on top, the checklist
+// state under it. PlanErr is set instead of Plan when the core could not build
+// a plan, so the page says so rather than showing an empty day.
+type morningView struct {
+	Plan     *ipc.DayPlan
+	PlanErr  string
+	Routines []ipc.MorningRoutineStatus
+}
+
 func handleVoice(w http.ResponseWriter, r *http.Request) {
 	w.Header().Set("Content-Type", "text/html; charset=utf-8")
 	if err := voiceTmpl.Execute(w, nil); err != nil {
@@ -1258,7 +1499,14 @@ func handleChatPage(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
 }
 
 // handleChatAPI processes a chat message POST and redirects back to /chat.
-func handleChatAPI(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
+//
+// State-changing, and the widest surface on this server: the text reaches the
+// router, the LLM, and through mavend's applyAction the whole action path
+// including `act` — so it is gated on the same step-up as POST /tools and
+// POST /api/revert (Vikunja #317). With WebAuthn unconfigured the gate is
+// fail-open exactly like the others (see stepUpOK); with -require-stepup it
+// denies, which is the point of that flag.
+func handleChatAPI(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI, session *webauthn.PasskeySession, requireStepUp bool) {
 	if r.Method != http.MethodPost {
 		http.Error(w, "POST only", http.StatusMethodNotAllowed)
 		return
@@ -1267,6 +1515,10 @@ func handleChatAPI(w http.ResponseWriter, r *http.Request, core ipc.CoreAPI) {
 		http.Error(w, "chat disabled (no -core)", http.StatusServiceUnavailable)
 		return
 	}
+	if !stepUpOK(session, requireStepUp) {
+		http.Error(w, "step-up required: assert a passkey first", http.StatusForbidden)
+		return
+	}
 	text := strings.TrimSpace(r.FormValue("text"))
 	if text == "" {
 		http.Redirect(w, r, "/chat", http.StatusSeeOther)

cmd/mavweb/morning.html

@@ -1,9 +1,27 @@
 {{template "shellTop" "morning"}}
+<h1>Today</h1>
+{{with .Plan}}
+<div class=hint>{{.Date.Format "02.01.2006"}}</div>
+{{if not .Items}}
+<div class=hint>nothing planned</div>
+{{else}}
+<div class=scroll><table class=mono>
+<tr><th>at<th>kind<th>what</tr>
+{{range .Items}}<tr>
+<td>{{.At.Format "15:04"}}</td>
+<td class=gray>{{.Kind}}</td>
+<td>{{if .Uncertain}}<span class=hint title="relayed notification, not a calendar read">похоже,</span> {{end}}{{.Text}}</td>
+</tr>{{end}}
+</table></div>
+{{end}}
+{{end}}
+{{if .PlanErr}}<div class=hint>plan unavailable: {{.PlanErr}}</div>{{end}}
+
 <h1>Morning Routines</h1>
-{{if not .}}
+{{if not .Routines}}
 <div class=hint>no morning routines configured</div>
 {{else}}
-{{range .}}
+{{range .Routines}}
 <div class="mb-4">
   <div><strong>{{.Name}}</strong>
     <span class={{if .Active}}green{{else}}gray{{end}}>{{if .Active}}active now{{else}}outside window{{end}}</span>

cmd/mavweb/tasks.html

@@ -0,0 +1,83 @@
+{{template "shellTop" "tasks"}}
+<h1>Tasks</h1>
+{{if .Msg}}<div class="msg msg-ok">{{.Msg}}</div>{{end}}
+{{if .Err}}<div class="msg msg-err">{{.Err}}</div>{{end}}
+
+<section class=card>
+<h2 class=card-title>add</h2>
+<form method=post action=/tasks class=inline-form>
+<input type=hidden name=action value=add>
+<input type=text name=text placeholder="что нужно сделать" size=44 required>
+<input type=date name=due title="due date (optional)">
+<select name=weight title="importance (optional)">
+<option value=0>normal</option>
+<option value=2>важно</option>
+<option value=3>срочно</option>
+</select>
+<button class=btn>add</button>
+</form>
+</section>
+
+{{if .Candidates}}
+<section class=card>
+<h2 class=card-title>found, not confirmed <span class=badge>{{len .Candidates}}</span></h2>
+<div class=hint>maven derived these from something she read. nothing counts as your work until you confirm it.</div>
+<div class=scroll><table>
+<tr><th>task</th><th>where from</th><th>due</th><th>captured</th><th></th><th></th></tr>
+{{range .Candidates}}<tr>
+<td class=text-max>{{.Text}}</td>
+<td class=hint>{{.Source}}{{if .Evidence}} — {{.Evidence}}{{end}}</td>
+<td>{{.Due}}</td>
+<td class=muted>{{.Created}}</td>
+<td><form method=post action=/tasks class=inline-form>
+<input type=hidden name=id value="{{.ID}}">
+<input type=hidden name=action value=confirm>
+<button class=btn>confirm</button></form></td>
+<td><form method=post action=/tasks class=inline-form>
+<input type=hidden name=id value="{{.ID}}">
+<input type=hidden name=action value=drop>
+<button class="btn btn-muted">drop</button></form></td>
+</tr>{{end}}</table></div>
+</section>
+{{end}}
+
+<section class=card>
+<h2 class=card-title>open <span class=badge>{{len .Open}}</span></h2>
+<div class=hint>most pressing first — by the deadlines and the urgency you gave, nothing guessed.</div>
+{{if .Open}}<div class=scroll><table>
+<tr><th>task</th><th>why</th><th>from</th><th>due</th><th>captured</th><th></th><th></th></tr>
+{{range .Open}}<tr>
+<td class=text-max>{{.Text}}</td>
+<td class=hint>{{.Why}}</td>
+<td class=hint>{{.Source}}</td>
+<td>{{.Due}}</td>
+<td class=muted>{{.Created}}</td>
+<td><form method=post action=/tasks class=inline-form>
+<input type=hidden name=id value="{{.ID}}">
+<input type=hidden name=action value=done>
+<button class=btn>done</button></form></td>
+<td><form method=post action=/tasks class=inline-form>
+<input type=hidden name=id value="{{.ID}}">
+<input type=hidden name=action value=drop>
+<button class="btn btn-muted">drop</button></form></td>
+</tr>{{end}}</table></div>
+{{else}}<div class=empty>
+  <svg class=icon width="20" height="20"><use href="/ethos-icons.svg#i-grid"/></svg>
+  <div>no open tasks</div>
+  <div class=hint>add one above, or tell maven "добавь в задачи …"</div>
+</div>{{end}}
+</section>
+
+{{if .Resolved}}
+<section class=card>
+<h2 class=card-title>resolved <span class=badge>{{len .Resolved}}</span></h2>
+<div class=scroll><table>
+<tr><th>task</th><th>status</th><th>when</th></tr>
+{{range .Resolved}}<tr>
+<td class=text-max>{{.Text}}</td>
+<td><span class="badge {{.Status}}">{{.Status}}</span></td>
+<td class=muted>{{.Resolved}}</td>
+</tr>{{end}}</table></div>
+</section>
+{{end}}
+{{template "shellBottom"}}

cmd/mavweb/tasks_test.go

@@ -0,0 +1,225 @@
+package main
+
+import (
+	"context"
+	"net/http"
+	"net/http/httptest"
+	"net/url"
+	"strings"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/ipc"
+	"github.com/kami/maven/internal/tasks"
+)
+
+// fakeTaskCore serves the /tasks handler: a canned list plus a log of the
+// writes the page made.
+type fakeTaskCore struct {
+	ipc.UnimplementedCoreAPI
+
+	tasks   []ipc.Task
+	listErr error
+
+	captured   []ipc.CaptureTaskReq
+	created    bool
+	captureErr error
+
+	statusID  int64
+	statusVal string
+	statusErr error
+}
+
+func (f *fakeTaskCore) ListTasks(_ context.Context, status string) ([]ipc.Task, error) {
+	if f.listErr != nil {
+		return nil, f.listErr
+	}
+	return f.tasks, nil
+}
+
+func (f *fakeTaskCore) CaptureTask(_ context.Context, req ipc.CaptureTaskReq) (ipc.CaptureTaskResp, error) {
+	f.captured = append(f.captured, req)
+	if f.captureErr != nil {
+		return ipc.CaptureTaskResp{}, f.captureErr
+	}
+	return ipc.CaptureTaskResp{ID: 7, Created: f.created}, nil
+}
+
+func (f *fakeTaskCore) SetTaskStatus(_ context.Context, id int64, status string, _ time.Time) error {
+	f.statusID, f.statusVal = id, status
+	return f.statusErr
+}
+
+func TestHandleTasksSplitsCandidatesFromOpen(t *testing.T) {
+	now := time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC)
+	resolved := now.Add(time.Hour)
+	core := &fakeTaskCore{tasks: []ipc.Task{
+		{ID: 1, Text: "купить молоко", Source: "tap:voice", Status: "open", CreatedTs: now},
+		{ID: 2, Text: "продлить страховку", Source: "email:kami", Evidence: "полис истекает", Status: "candidate", CreatedTs: now},
+		{ID: 3, Text: "полить цветы", Source: "tap:web", Status: "done", CreatedTs: now, Resolved: &resolved},
+	}}
+	rec := httptest.NewRecorder()
+	handleTasks(rec, httptest.NewRequest(http.MethodGet, "/tasks", nil), core)
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d", rec.Code)
+	}
+	body := rec.Body.String()
+	for _, want := range []string{
+		"купить молоко", "продлить страховку", "полить цветы",
+		"полис истекает",       // the evidence trail is visible for review
+		"found, not confirmed", // candidates get their own section
+	} {
+		if !strings.Contains(body, want) {
+			t.Errorf("body missing %q", want)
+		}
+	}
+	// The candidate must offer confirm, and the open task must not.
+	if !strings.Contains(body, "value=confirm") {
+		t.Error("candidate row has no confirm action")
+	}
+}
+
+func TestHandleTasksAddCaptures(t *testing.T) {
+	core := &fakeTaskCore{created: true}
+	form := url.Values{"action": {"add"}, "text": {"  позвонить в банк "}, "due": {"2026-08-05"}}
+	req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	rec := httptest.NewRecorder()
+	handleTasks(rec, req, core)
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d", rec.Code)
+	}
+	if len(core.captured) != 1 {
+		t.Fatalf("captured %d requests, want 1", len(core.captured))
+	}
+	got := core.captured[0]
+	if got.Text != "позвонить в банк" {
+		t.Errorf("text = %q, want trimmed", got.Text)
+	}
+	if got.Source != "tap:web" {
+		t.Errorf("source = %q, want tap:web", got.Source)
+	}
+	if got.Status != "open" {
+		t.Errorf("status = %q — a task he typed himself is open, not a candidate", got.Status)
+	}
+	if got.Due == nil || got.Due.Format("2006-01-02") != "2026-08-05" {
+		t.Errorf("due = %v", got.Due)
+	}
+	if !strings.Contains(rec.Body.String(), "added task") {
+		t.Error("no confirmation message")
+	}
+}
+
+func TestHandleTasksAddSaysAlreadyOnTheList(t *testing.T) {
+	core := &fakeTaskCore{created: false}
+	form := url.Values{"action": {"add"}, "text": {"купить молоко"}}
+	req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	rec := httptest.NewRecorder()
+	handleTasks(rec, req, core)
+	if !strings.Contains(rec.Body.String(), "already on the list") {
+		t.Error("a deduped capture must not claim it saved something new")
+	}
+}
+
+func TestHandleTasksStatusActions(t *testing.T) {
+	for _, tc := range []struct{ action, want string }{
+		{"confirm", "open"},
+		{"done", "done"},
+		{"drop", "dropped"},
+	} {
+		core := &fakeTaskCore{}
+		form := url.Values{"action": {tc.action}, "id": {"42"}}
+		req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+		req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+		handleTasks(httptest.NewRecorder(), req, core)
+		if core.statusID != 42 || core.statusVal != tc.want {
+			t.Errorf("%s → SetTaskStatus(%d, %q), want (42, %q)", tc.action, core.statusID, core.statusVal, tc.want)
+		}
+	}
+}
+
+func TestHandleTasksRejectsBadPost(t *testing.T) {
+	core := &fakeTaskCore{}
+	form := url.Values{"action": {"explode"}, "id": {"1"}}
+	req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	rec := httptest.NewRecorder()
+	handleTasks(rec, req, core)
+	// The page still renders, with the error inline — and nothing was written.
+	if rec.Code != http.StatusOK {
+		t.Fatalf("status = %d", rec.Code)
+	}
+	if core.statusVal != "" || len(core.captured) != 0 {
+		t.Error("an unknown action must write nothing")
+	}
+	if !strings.Contains(rec.Body.String(), "unknown action") {
+		t.Error("error not surfaced on the page")
+	}
+}
+
+func TestHandleTasksNoCore(t *testing.T) {
+	rec := httptest.NewRecorder()
+	handleTasks(rec, httptest.NewRequest(http.MethodGet, "/tasks", nil), nil)
+	if rec.Code != http.StatusServiceUnavailable {
+		t.Errorf("status = %d, want 503", rec.Code)
+	}
+}
+
+// The open list is ordered by the ranker, and the reason is shown so the page
+// says why a task is first instead of asking him to trust the order.
+func TestHandleTasksOrdersOpenByRank(t *testing.T) {
+	now := time.Now()
+	due := now
+	core := &fakeTaskCore{tasks: []ipc.Task{
+		{ID: 1, Text: "купить молоко", Status: "open", CreatedTs: now},
+		{ID: 2, Text: "оплатить интернет", Status: "open", CreatedTs: now, Due: &due},
+	}}
+	rec := httptest.NewRecorder()
+	handleTasks(rec, httptest.NewRequest(http.MethodGet, "/tasks", nil), core)
+	body := rec.Body.String()
+	if strings.Index(body, "оплатить интернет") > strings.Index(body, "купить молоко") {
+		t.Error("want the dated task rendered first")
+	}
+	if !strings.Contains(body, "сегодня") {
+		t.Error("want the ranker's reason shown in the why column")
+	}
+}
+
+// A candidate is ranked into place but never carries a priority reason: its due
+// date is Maven's reading of a mail, not something he stated.
+func TestHandleTasksHidesCandidateReason(t *testing.T) {
+	now := time.Now()
+	due := now
+	core := &fakeTaskCore{tasks: []ipc.Task{
+		{ID: 1, Text: "продлить страховку", Status: "candidate", CreatedTs: now, Due: &due},
+	}}
+	rec := httptest.NewRecorder()
+	handleTasks(rec, httptest.NewRequest(http.MethodGet, "/tasks", nil), core)
+	if strings.Contains(rec.Body.String(), "сегодня") {
+		t.Error("a candidate must not be shown with a priority reason")
+	}
+}
+
+func TestApplyTaskPostCarriesWeight(t *testing.T) {
+	core := &fakeTaskCore{created: true}
+	form := url.Values{"action": {"add"}, "text": {"оплатить интернет"}, "weight": {"3"}}
+	req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	handleTasks(httptest.NewRecorder(), req, core)
+	if len(core.captured) != 1 || core.captured[0].Weight != 3 {
+		t.Fatalf("captured = %+v, want weight 3", core.captured)
+	}
+}
+
+// Out of range clamps rather than 400s; a non-number is a real client error.
+func TestApplyTaskPostClampsWeight(t *testing.T) {
+	core := &fakeTaskCore{created: true}
+	form := url.Values{"action": {"add"}, "text": {"что-то"}, "weight": {"99"}}
+	req := httptest.NewRequest(http.MethodPost, "/tasks", strings.NewReader(form.Encode()))
+	req.Header.Set("Content-Type", "application/x-www-form-urlencoded")
+	handleTasks(httptest.NewRecorder(), req, core)
+	if core.captured[0].Weight != tasks.MaxWeight {
+		t.Errorf("weight = %d, want the cap", core.captured[0].Weight)
+	}
+}

deploy/README.md

@@ -36,6 +36,37 @@ present (see `.dockerignore`).
 | `/var/lib/maven` (volume)  | encrypted db at rest                    |
 | `/dev/shm` (tmpfs)         | decrypted db working copy (RAM only)    |
 
+## Reading the outside world (off by default)
+
+`mavend.json` ships without a `feeds` block, which means no RSS/Atom feed is
+fetched and no outbound request is made. Switching it on is adding the block:
+
+```json
+"feeds": {
+  "poll_interval": "30m",
+  "max_items": 5,
+  "max_age": "24h",
+  "sources": [
+    { "name": "habr", "url": "https://habr.com/ru/rss/best/daily/",
+      "category": "технологии", "exclude": ["реклама"] }
+  ]
+}
+```
+
+What it does and does not do:
+
+- items are written as notes with source `rss:<name>`, visible on `/dash`;
+- **nothing is announced.** She reads them back when asked — "что нового в
+  лентах?", "что нового по технологиям?" — and never on arrival. There is no
+  severity or channel knob here on purpose;
+- the fetcher is allowlisted to the hosts of the configured feeds, plus any
+  `allow_hosts`. It refuses non-http(s) schemes and every private address
+  (loopback, the LAN, the `10.42.0.0/24` wg range, cloud metadata). It caps the
+  response at 2 MiB and redirects at 3, and makes at most one request per host
+  per second. See `internal/webfetch`;
+- how far each feed was read is stored as a config fact `rss:latest:<name>`, so
+  a restart does not re-note yesterday's headlines.
+
 ## Not yet verified / host-dependent
 
 This stack is correct-by-construction but has **not been build-tested here**

deploy/ecosystem/docker-compose.yml

@@ -8,6 +8,18 @@
 #
 # Maven's own compose joins this same network (add `ecosystem` as an external
 # network there) to reach nexus:9740 / praxis:8989 / hexis:9741 directly.
+#
+# NO RELEASE PINNING (Vikunja #354): each `build:` below points at a sibling
+# WORKING TREE, so `up --build` ships whatever is checked out there, including
+# uncommitted edits. Before bringing this up, check what you are about to
+# deploy:
+#
+#   for r in nexus praxis hexis; do git -C ../../../$r status --short; \
+#     git -C ../../../$r log -1 --oneline; done
+#
+# The host nginx that fronts these is deploy/ecosystem/nginx.conf — it binds
+# the wg and LAN addresses only, with allow/deny. Keep it that way: none of
+# these containers has auth of its own.
 name: ecosystem
 
 services:

deploy/ecosystem/nginx.conf

@@ -1,13 +1,71 @@
-# Reverse-proxy the three sibling admin UIs. Drop into your nginx sites (or the
+# Reverse-proxy Maven's own web UI plus the three sibling admin UIs. Drop into
-# nginx-panel app) and reload. Assumes the compose publishes each service on
+# your nginx sites (or the nginx-panel app) and reload. Assumes the compose
-# 127.0.0.1:<port>. Add TLS (certbot / your existing cert block) per server.
+# publishes each service on 127.0.0.1:<port>. Add TLS (certbot / your existing
+# cert block) per server.
 #
 # NOTE: hexis.<domain> previously pointed at the MCP tool — repoint that
 # elsewhere first (the app now owns hexis.*).
+#
+# 10.42.0.1 and 192.168.1.104 below are THIS BOX's WireGuard and LAN
+# addresses (homesrv) — these admin UIs have no auth of their own, so the
+# explicit bind + allow/deny below is what keeps them off the open internet.
+# On a different box, replace both addresses with that box's wg and LAN IPs.
+# Do NOT "fix" a failed bind by reverting to `listen 80` (all interfaces) —
+# that removes the only access control these containers have.
+
+# maven.<domain> → mavweb (docker-compose.yml publishes it on 127.0.0.1:9201).
+# Same bind + ACL as the siblings, and for a stronger reason: mavweb serves
+# POST /tools, which defines argv that internal/tool EXECUTES, plus POST
+# /routines, /api/revert and /api/chat (Vikunja #317). Without
+# -webauthn-origin/-webauthn-rpid mavweb has no auth of its own, so this block
+# is the auth. If you add TLS and a basic-auth/oauth2-proxy layer, keep the
+# allow/deny anyway — belt and braces on an RCE surface.
+#
+# WebSocket upgrade matters here: /ws carries push-to-talk audio, so the
+# Upgrade/Connection headers below are required, not decoration. The map keeps
+# `Connection: upgrade` off plain requests; it sits in the http context, which
+# is where sites-available files are included — if your nginx already defines
+# $connection_upgrade, drop this block.
+map $http_upgrade $connection_upgrade {
+    default upgrade;
+    ''      close;
+}
 
 server {
-    listen 80;
+    listen 10.42.0.1:80;
+    listen 192.168.1.104:80;
+    server_name maven.kvmx.ru;
+
+    allow 10.42.0.0/24;
+    allow 192.168.1.0/24;
+    deny all;
+
+    # push-to-talk uploads raw PCM; the default 1m is enough for a short
+    # utterance but not for a long one.
+    client_max_body_size 32m;
+
+    location / {
+        proxy_pass http://127.0.0.1:9201;
+        proxy_http_version 1.1;
+        proxy_set_header Upgrade $http_upgrade;
+        proxy_set_header Connection $connection_upgrade;
+        proxy_set_header Host $host;
+        proxy_set_header X-Real-IP $remote_addr;
+        proxy_set_header X-Forwarded-For $proxy_add_x_forwarded_for;
+        proxy_set_header X-Forwarded-Proto $scheme;
+        proxy_read_timeout 300s;   # an LLM turn can take minutes on the iGPU
+    }
+}
+
+server {
+    listen 10.42.0.1:80;
+    listen 192.168.1.104:80;
     server_name nexus.kvmx.ru;
+
+    allow 10.42.0.0/24;
+    allow 192.168.1.0/24;
+    deny all;
+
     location / {
         proxy_pass http://127.0.0.1:9740;
         proxy_set_header Host $host;
@@ -18,8 +76,14 @@ server {
 }
 
 server {
-    listen 80;
+    listen 10.42.0.1:80;
+    listen 192.168.1.104:80;
     server_name praxis.kvmx.ru;
+
+    allow 10.42.0.0/24;
+    allow 192.168.1.0/24;
+    deny all;
+
     location / {
         proxy_pass http://127.0.0.1:8989;
         proxy_set_header Host $host;
@@ -30,8 +94,14 @@ server {
 }
 
 server {
-    listen 80;
+    listen 10.42.0.1:80;
+    listen 192.168.1.104:80;
     server_name hexis.kvmx.ru;
+
+    allow 10.42.0.0/24;
+    allow 192.168.1.0/24;
+    deny all;
+
     location / {
         proxy_pass http://127.0.0.1:9741;
         proxy_set_header Host $host;

deploy/mavend.json

@@ -6,11 +6,12 @@
   "state_dir": "/var/lib/maven",
 
   "phraser": {
-    "model_path": "/opt/maven/models/llm/qwen3.5/Qwen3.5-0.8B.Q4_K_M.gguf",
+    "model_path": "/opt/maven/models/llm/qwen3/Qwen3-1.7B-UD-Q4_K_XL.gguf",
     "bin_path": "llama-server",
     "n_gpu_layers": 99,
-    "n_ctx": 2048,
+    "n_ctx": 4096,
-    "timeout": "60s"
+    "timeout": "60s",
+    "llm_nudges": false
   },
 
   "telegram": {
@@ -25,6 +26,11 @@
     "severity_ceiling": 2
   },
 
+  "pattern_proposals": {
+    "notify": false,
+    "cooldown": "24h"
+  },
+
   "nexus":  { "url": "http://nexus:9740" },
   "praxis": { "url": "http://praxis:8989" },
   "hexis":  { "url": "http://hexis:9741" },
@@ -36,11 +42,14 @@
     "stt": { "socket": "/run/maven/stt.sock", "lang": "ru" },
     "tts": { "socket": "/run/maven/tts.sock", "lang": "ru" },
     "embedder": {
-      "model_path": "/opt/maven/models/embedder/model.onnx",
+      "model_path": "/opt/maven/models/embedder/multilingual-e5-small/model_quantized.onnx",
-      "tokenizer_path": "/opt/maven/models/embedder/tokenizer.json",
+      "tokenizer_path": "/opt/maven/models/embedder/multilingual-e5-small/tokenizer.json",
       "lib_path": "/opt/maven/lib/libonnxruntime.so"
     },
-    "llm_router": false,
+    "llm_router": true,
+    "query_min_score": 0.55,
+    "query_min_margin": 0.008,
+    "clarify_max_attempts": 3,
     "tool_timeout": "30s",
     "tools": [
       { "name": "status",     "cmd": ["systemctl", "status"],         "scope": "homelab", "destructive": false },

docker-compose.yml

@@ -101,9 +101,42 @@ services:
               "-netdata", "http://127.0.0.1:19999",
               "-kuma", "http://127.0.0.1:3001/metrics",
               "-kuma-key", "uk5_mavpoll-key"]
+              # Money tracking (Vikunja #125) is OFF: it needs a zenmoney token,
+              # which mavpoll reads from a FILE so it never appears in `ps`, in
+              # this file, or in shell history. To enable, mount the token and
+              # append: "-zenmoney-token-file", "/run/secrets/zenmoney.token"
+              # (optionally "-zenmoney-interval", "1h"). Core never sees the
+              # token — the poller writes facts(kind=env, source=poll:zenmoney)
+              # and mavend only reads those back when he asks.
     depends_on: [mavend]
     volumes:
       - sockets:/run/maven
+      # - ./deploy/zenmoney.token:/run/secrets/zenmoney.token:ro
+
+  # The mail reader (Vikunja #246) is OFF and commented out: it needs an IMAP
+  # account, and there is none on this box. mavmaild reads the password from a
+  # FILE so it never appears in `ps`, in this file, or in shell history — the
+  # same rule mavpoll follows for the zenmoney token. Core never sees the
+  # password: the reader hands core message text on one IPC method, and core
+  # writes what the model extracts as task CANDIDATES he reviews on /tasks.
+  # Nothing here can create a reminder, so a misread mail cannot fire.
+  #
+  # To enable: write the password to deploy/imap.password (0600, gitignored),
+  # add an "email": {} block to deploy/mavend.json, and uncomment this service.
+  # mavmaild:
+  #   <<: *image
+  #   command: ["mavmaild", "-socket", "/run/maven/mavend.sock",
+  #             "-imap", "imap.example.org:993",
+  #             "-user", "kami@example.org",
+  #             "-password-file", "/run/secrets/imap.password",
+  #             "-mailbox", "INBOX",
+  #             "-interval", "15m",
+  #             "-state", "/var/lib/maven/mail-seen.json"]
+  #   depends_on: [mavend]
+  #   volumes:
+  #     - sockets:/run/maven
+  #     - dbdata:/var/lib/maven
+  #     - ./deploy/imap.password:/run/secrets/imap.password:ro
 
 volumes:
   dbdata:

docs/plans/03-memory-evaluation.md

@@ -25,3 +25,46 @@
 6. Add `/eval` API method to `ipc.CoreAPI` (or reuse `Chat` with system context) so mavweb can show evaluation history
 7. Add `memory_eval` block to `deploy/mavend.json`
 8. Test with synthetic store state — verify observations match expected patterns
+
+---
+
+## Status 2026-08-01 — foundation shipped (Vikunja #248)
+
+**Shipped:** `internal/memeval` (not `internal/memory/eval.go` — `internal/store`
+imports `internal/memory` for the vector backend, so an evaluator that reads
+`store.Fact` there would close an import cycle). `Evaluator.Evaluate` reads
+`RecentFacts` / `RecentNotes` / `RecentNudges`, prompts the resident model under
+a GBNF grammar for at most three `{observation, confidence, suggested_action}`
+objects, drops anything under `min_confidence`, deduplicates against what earlier
+evaluations wrote, and records the rest as notes with source `infer:memory-eval`.
+Driver: `cmd/mavend/memoryeval.go`, its own goroutine on its own ticker. Config:
+the `memory_eval` block — **absent ⇒ the loop does not run**. Visibility: `/dash`
+already renders notes with their source, so evaluation output is visible with no
+UI change.
+
+**Deliberately not shipped — this is policy, not an unfinished edge:**
+
+- *Dispatching observations as care nudges (plan step 4).* An hourly LLM loop
+  with permission to speak is a machine for generating interruptions, and the
+  content is model-generated text about his own life. The evaluator has no
+  dispatcher reference at all, so it cannot reach a channel by accident. Wiring
+  it to `delivery.Dispatcher` is a separate decision with its own opt-in.
+- *Acting on `suggested_action`.* It is recorded inside the note text and
+  interpreted by nobody. No reminder, routine or fact is created.
+- *Writing observation embeddings.* Notes are written with a nil embedding, so
+  they stay out of the RAG recall pool. Feeding generated text back into the pool
+  it came from is how a small model starts citing its own guesses as evidence.
+
+**Deferred, wants a decision or another capability:**
+
+- *Plan step 6, the `/eval` IPC method and an evaluation-history view.* `/dash`
+  covers reading the output; a dedicated trace surface is worth building once
+  there is real output to look at, and it should probably show the prompt too.
+- *`RecentEvents`.* The plan lists it; the evaluator reads facts, notes and
+  nudges. Detected action/object events already drive pattern proposals (#43), and
+  duplicating them here would mostly re-derive that.
+- *Output quality is unmeasured.* There is no fixture for "did she notice
+  something true". The tests cover the machinery — empty store, confidence floor,
+  dedupe, own-notes exclusion, error handling — not the observations. Until
+  someone reads a week of real output on `/dash`, treat the wording and the
+  `min_confidence` default as unvalidated.

docs/plans/09-behavioral-memory.md

@@ -27,3 +27,36 @@
 6. Wire voice query — `"что я обычно делаю?"` routes to `IntentQuery` → behavior profile lookup → LLM-phrased answer
 7. Add IPC read method `MethodGetBehaviorProfile` so mavweb can display it on `/dash`
 8. Test with synthetic fact history — verify weekly schedule is correctly inferred
+
+---
+
+## Status (2026-08-01) — partially shipped, deliberately narrowed
+
+Shipped on `overnight/behavior-profile`:
+
+- `internal/memory/behavior.go` — `BuildProfile` counts habits per weekday out of
+  self-facts: distinct-day counts (`MinHabitDays = 2`), a median time-of-day, and
+  `FormatWeekdayRU` / `FormatOverallRU` for the spoken answer.
+- `internal/router/habit.go` — `ParseHabitQuery`, which requires a habit marker
+  ("обычно", "каждую", "привычки", …) and parses the weekday deterministically.
+- `cmd/mavend/actions_query.go` — a `habits` query source, so "что я обычно делаю
+  по вторникам?" is answered.
+
+**Not shipped, and not to be shipped as written:**
+
+- *Step 3, LLM-generated profile stored as a fact.* The profile is COUNTED, not
+  generated. A 1.7B asked to summarise a year of habits produces fluent claims
+  about the owner's life that no row supports, and a wrong claim about him is the
+  most expensive kind of wrong maven can be. Counting is verifiable and cheap.
+- *Step 5, incremental updates on fact write.* There is no cache to keep fresh —
+  the profile is recomputed on the question, so a new fact is already in the next
+  answer. A cached profile that can disagree with its own rows is two truths.
+- *Step 4, proactive daily plan proposals via the dispatcher.* Maven is not a nag,
+  and a nudge at 08:00 every day proposing the day is the definition of one. The
+  sanctioned path from "she noticed a pattern" to "she acts on it" already exists:
+  `internal/pattern/detector.go` proposes a routine, and the owner accepts it on
+  `/routines`. It goes through him.
+
+Still open, if wanted later: `MethodGetBehaviorProfile` + a `/dash` panel (step 7).
+The counted profile needs no new IPC method to be *asked* about — the query source
+reads `RecentFacts` over the existing surface — so this is a display concern only.

docs/plans/13-rss-news-feeds.md

@@ -27,3 +27,28 @@
 7. Add voice query handler — `"что нового?"` queries `RecentNotes` filtered by source prefix `rss:` and phrases via `phraser.PhraseQuery`
 8. Add `feeds` block to `config.Config` and `deploy/mavend.json`
 9. Test with a live RSS feed (e.g., `https://news.ycombinator.com/rss`) — verify items appear in notes table
+
+---
+
+## Shipped 2026-08-01 (#258)
+
+`internal/webfetch` (the guarded HTTP door: scheme, allow/deny hosts, private-address
+refusal in the dialer, size cap, redirect cap, per-host rate limit), `internal/rss`
+(RSS 2.0 + Atom parser, poller with durable marks and a keyword filter),
+`cmd/mavend/feeds.go` (ticker, fetcher adapter, `rss:latest:<feed>` fact marks),
+config block `feeds`, and the `feeds` query source with `router.ParseFeedQuery`.
+
+Deviations from the plan above, both deliberate:
+
+- **Step 5 (breaking-news nudges) was not built.** A feed that dispatches is a nag,
+  and the one thing Maven is not is a nag. Items are read when asked and nowhere else.
+  If breaking news is ever wanted, it belongs behind the existing delivery policy
+  (severity, quiet hours, digest), not in the poller.
+- **Step 3 (embedder relevance) is a seam, not an implementation.** `rss.Ranker`
+  exists and is wired nil. Scoring items against an "interest profile" needs a
+  profile, and there is none yet; a threshold with nothing to compare against is a
+  random filter with a confident name. The filter that runs is the per-feed
+  include/exclude keyword list, which he can read and predict.
+
+No new dependency: stdlib `encoding/xml`, no gofeed. Stock deploy config has no
+`feeds` block, so the capability is off.

internal/auth/auth_test.go

@@ -7,7 +7,6 @@ import (
 	"path/filepath"
 	"strings"
 	"testing"
-	"time"
 
 	"github.com/kami/maven/internal/ipc"
 )
@@ -359,8 +358,12 @@ func TestGate_IpcServer_ChatAllowedForEnrolledCaller(t *testing.T) {
 
 // recordingAPI — a no-op CoreAPI that counts WriteFact invocations; the auth
 // check must reject before reaching it, otherwise the refusal leaks into the
-// fake's counts and we fail.
+// fake's counts and we fail. Embeds ipc.UnimplementedCoreAPI so every method
+// this test doesn't exercise returns ipc.ErrNotImplemented loudly instead of
+// being hand-stubbed to a canned value nobody checks.
 type recordingAPI struct {
+	ipc.UnimplementedCoreAPI
+
 	writes int
 	chats  int
 }
@@ -369,89 +372,7 @@ func (r *recordingAPI) WriteFact(_ context.Context, _ ipc.WriteFactReq) (int64,
 	r.writes++
 	return int64(r.writes), nil
 }
-func (r *recordingAPI) LatestFact(_ context.Context, _ string) (ipc.Fact, error) {
-	return ipc.Fact{}, ipc.ErrNoFact
-}
-func (r *recordingAPI) LatestFactBySource(_ context.Context, _, _ string) (ipc.Fact, error) {
-	return ipc.Fact{}, ipc.ErrNoFact
-}
-func (r *recordingAPI) Since(_ context.Context, _ string, _ time.Time) (time.Duration, error) {
-	return 0, ipc.ErrNoFact
-}
-func (r *recordingAPI) Presence(_ context.Context) (ipc.Presence, error) {
-	return ipc.Presence{}, nil
-}
-func (r *recordingAPI) CreateReminder(_ context.Context, _ time.Time, _, _ string) (int64, error) {
-	return 1, nil
-}
-func (r *recordingAPI) MarkReminder(_ context.Context, _ int64, _ string) error { return nil }
-func (r *recordingAPI) ListReminders(_ context.Context, _ int) ([]ipc.Reminder, error) {
-	return nil, nil
-}
-func (r *recordingAPI) TickTrace(_ context.Context) (ipc.TickTrace, error) {
-	return ipc.TickTrace{}, nil
-}
-func (r *recordingAPI) MorningStatus(_ context.Context) ([]ipc.MorningRoutineStatus, error) {
-	return nil, nil
-}
-func (r *recordingAPI) RecordNudge(_ context.Context, _, _, _ string, _ time.Time) (int64, error) {
-	return 1, nil
-}
-func (r *recordingAPI) ResolveNudge(_ context.Context, _ int64, _ string, _ time.Time) error {
-	return nil
-}
-func (r *recordingAPI) RecentOutcomes(_ context.Context, _ string, _ int) ([]string, error) {
-	return nil, nil
-}
-func (r *recordingAPI) RecentFacts(_ context.Context, _ int) ([]ipc.Fact, error) {
-	return nil, nil
-}
-func (r *recordingAPI) CalendarEvents(_ context.Context, _, _ time.Time) ([]ipc.Fact, error) {
-	return nil, nil
-}
-func (r *recordingAPI) RecentNudges(_ context.Context, _ int) ([]ipc.Nudge, error) {
-	return nil, nil
-}
-func (r *recordingAPI) WriteNote(_ context.Context, _ time.Time, _ string, _ []float32, _ string) (int64, error) {
-	return 1, nil
-}
-func (r *recordingAPI) QueryNotes(_ context.Context, _ []float32, _ int) ([]ipc.Note, error) {
-	return nil, nil
-}
-func (r *recordingAPI) RecentNotes(_ context.Context, _ int) ([]ipc.Note, error) {
-	return nil, nil
-}
-func (r *recordingAPI) ProposeTool(_ context.Context, _, _, _ string, _ time.Time) (bool, error) {
-	return false, nil
-}
-func (r *recordingAPI) EnableTool(_ context.Context, _ string, _ []string, _ bool, _ string, _ time.Time) error {
-	return nil
-}
-func (r *recordingAPI) DisableTool(_ context.Context, _ string) error {
-	return nil
-}
-func (r *recordingAPI) DeleteTool(_ context.Context, _ string) error {
-	return nil
-}
-func (r *recordingAPI) LookupTool(_ context.Context, _ string) (ipc.Tool, error) {
-	return ipc.Tool{}, ipc.ErrToolNotFound
-}
-func (r *recordingAPI) ListTools(_ context.Context, _ string) ([]ipc.Tool, error) {
-	return nil, nil
-}
 
-func (r *recordingAPI) RevertFact(_ context.Context, _ string) (int64, error) {
-	return 0, nil
-}
-func (r *recordingAPI) ListProposedRoutines(_ context.Context) ([]ipc.ProposedRoutine, error) {
-	return nil, nil
-}
-func (r *recordingAPI) AcceptProposedRoutine(_ context.Context, _ int64) error {
-	return nil
-}
-func (r *recordingAPI) DismissProposedRoutine(_ context.Context, _ int64) error {
-	return nil
-}
 func (r *recordingAPI) Chat(_ context.Context, text string) (string, error) {
 	r.chats++
 	return "echo: " + text, nil

internal/auth/policy.go

@@ -65,7 +65,21 @@ func Requirement(m ipc.Method) Authority {
 		ipc.MethodCreateReminder,
 		ipc.MethodMarkReminder,
 		ipc.MethodRecordNudge,
-		ipc.MethodResolveNudge:
+		ipc.MethodResolveNudge,
+		// Task capture (Vikunja #130). Listed explicitly rather than left to
+		// the default so the intent is on the record: capturing a task is a
+		// module write, not an allowlist mutation and not a new standing reason
+		// for Maven to speak — nothing in the tick loop reads tasks. It stays
+		// at AuthRead, the same rung as CreateReminder, which is the closest
+		// existing analogue.
+		ipc.MethodCaptureTask,
+		ipc.MethodListTasks,
+		ipc.MethodSetTaskStatus,
+		// Mail ingestion (Vikunja #246). AuthRead because of what the method can
+		// produce: candidate tasks and nothing else. It cannot write a fact, set a
+		// reminder, or touch the tool allowlist, so a compromised mail reader can
+		// at worst put junk on a review page he clears in one click.
+		ipc.MethodIngestMail:
 		return AuthRead
 	}
 	// Unknown method ⇒ AuthRead, but ipc.dispatch returns ErrUnknownMethod

internal/calendar/ambient.go

@@ -0,0 +1,221 @@
+package calendar
+
+import (
+	"strings"
+	"time"
+	"unicode"
+)
+
+// Ambient events — the work calendar read (Vikunja #126).
+//
+// The work calendar is not read by holding a work credential. A corp mail or
+// calendar session living on the homelab ties the box's blast radius to the
+// employer's data, which is the thing the task exists to refuse. What maven
+// reads instead is the SIGNAL: an Android notification-listener on the owner's
+// phone relays meeting notifications over wg/LAN, and maven turns the ones that
+// clearly describe a meeting into calendar events.
+//
+// That makes the provenance honest. A notification is evidence about an event,
+// not a reading of the calendar, so it is stored under SourceAmbient at
+// AmbientConfidence — never indistinguishable from a real CalDAV read, and the
+// query path hedges when it recites one.
+//
+// The parse is deliberately conservative. A notification with no recognisable
+// clock reading produces nothing at all: maven is not a guesser-of-truth, and a
+// mailbox full of noise turned into invented events is worse than a gap. Mail
+// as a notification signal, not a mailbox.
+
+// Notification — one relayed Android notification. Package is the posting app
+// (for the log and for the owner to see where a wrong event came from), Title
+// and Text are the notification's two text lines, Posted is when the phone
+// showed it. Nothing else off the notification is kept.
+type Notification struct {
+	Package string    `json:"package"`
+	Title   string    `json:"title"`
+	Text    string    `json:"text"`
+	Posted  time.Time `json:"posted_at"`
+}
+
+// EventFromNotification turns a notification into the event it describes, or
+// reports false when it does not clearly describe one.
+//
+// It needs two things: a clock reading, and a summary that is not just that
+// clock reading. Everything else is defaulted — the date is Posted's day (a
+// meeting notification is about today or it would not be firing now), and a
+// bare start time gets DefaultReminderDuration.
+func EventFromNotification(n Notification) (Event, bool) {
+	if n.Posted.IsZero() {
+		return Event{}, false
+	}
+	line := strings.TrimSpace(n.Title + " " + n.Text)
+	start, end, ok := parseTimeRange(line)
+	if !ok {
+		return Event{}, false
+	}
+	summary := notificationSummary(n)
+	if summary == "" {
+		return Event{}, false
+	}
+
+	y, m, d := n.Posted.Date()
+	loc := n.Posted.Location()
+	s := time.Date(y, m, d, start.hour, start.min, 0, 0, loc)
+	var e time.Time
+	if end != nil {
+		e = time.Date(y, m, d, end.hour, end.min, 0, 0, loc)
+		// A range that ends before it starts crossed midnight.
+		if !e.After(s) {
+			e = e.AddDate(0, 0, 1)
+		}
+	} else {
+		e = s.Add(DefaultReminderDuration)
+	}
+	return Event{Summary: summary, Start: s, End: e}, true
+}
+
+// notificationSummary picks the text that names the meeting: the title when it
+// carries words, otherwise the body. The clock reading is stripped out — it
+// already lives in the times, and FactValue renders it again.
+func notificationSummary(n Notification) string {
+	for _, cand := range []string{n.Title, n.Text} {
+		s := strings.TrimSpace(stripClock(cand))
+		s = strings.Trim(s, " \t-–—,;:@|·")
+		s = strings.Join(strings.Fields(s), " ")
+		if hasLetters(s) {
+			return s
+		}
+	}
+	return ""
+}
+
+type clock struct{ hour, min int }
+
+// parseTimeRange finds the first clock reading in s, and a second one if the
+// text spells a range. Accepted separators between hours and minutes are ":"
+// and "."; between the two ends of a range, "-", "–", "—" or "до".
+//
+// Bare hours ("в 14") are NOT accepted. Loose digits in a notification are far
+// more often a count, a date or an unread badge than a meeting time, and an
+// invented event is worse than no event.
+func parseTimeRange(s string) (start clock, end *clock, ok bool) {
+	first, _, firstEnd, ok := nextClock(s, 0)
+	if !ok {
+		return clock{}, nil, false
+	}
+	sep := strings.TrimLeft(s[firstEnd:], " \t")
+	for _, p := range []string{"-", "–", "—", "до "} {
+		if !strings.HasPrefix(sep, p) {
+			continue
+		}
+		if second, _, _, ok2 := nextClock(strings.TrimPrefix(sep, p), 0); ok2 {
+			return first, &second, true
+		}
+		break
+	}
+	return first, nil, true
+}
+
+// nextClock scans s from byte offset `from` for the first HH:MM (or HH.MM) and
+// returns it with the byte range it occupied. Digits and separators are ASCII,
+// so byte offsets are safe over Cyrillic text.
+func nextClock(s string, from int) (c clock, start, end int, ok bool) {
+	for i := from; i < len(s); i++ {
+		if !isDigit(s[i]) {
+			continue
+		}
+		j := i
+		for j < len(s) && isDigit(s[j]) {
+			j++
+		}
+		// A run longer than two digits is a year, an id or an unread count.
+		if j-i > 2 {
+			i = j
+			continue
+		}
+		if j >= len(s) || (s[j] != ':' && s[j] != '.') {
+			i = j
+			continue
+		}
+		k := j + 1
+		for k < len(s) && isDigit(s[k]) {
+			k++
+		}
+		if k-(j+1) != 2 {
+			i = j
+			continue
+		}
+		// Reject a group that is a link in a longer dotted or colon chain:
+		// "2026.08.15" would otherwise offer "08.15" as 08:15, and a deadline
+		// date invented as a meeting time is exactly the wrong kind of guess.
+		// A trailing ":ss" is fine — that is a time with seconds.
+		if i > 0 && (s[i-1] == '.' || s[i-1] == ':' || isDigit(s[i-1])) {
+			i = k
+			continue
+		}
+		if k < len(s) && s[k] == '.' && k+1 < len(s) && isDigit(s[k+1]) {
+			i = k
+			continue
+		}
+		hour, min := atoi(s[i:j]), atoi(s[j+1:k])
+		if hour > 23 || min > 59 {
+			i = k
+			continue
+		}
+		return clock{hour, min}, i, k, true
+	}
+	return clock{}, 0, 0, false
+}
+
+func isDigit(b byte) bool { return b >= '0' && b <= '9' }
+
+func atoi(s string) int {
+	n := 0
+	for i := 0; i < len(s); i++ {
+		n = n*10 + int(s[i]-'0')
+	}
+	return n
+}
+
+// stripClock removes every clock reading from a summary candidate, along with
+// the preposition or separator that introduced it.
+func stripClock(s string) string {
+	for {
+		_, start, end, ok := nextClock(s, 0)
+		if !ok {
+			return s
+		}
+		head := trimTrailingPreposition(strings.TrimRight(s[:start], "0123456789:.-–— \t"))
+		s = strings.TrimSpace(strings.TrimSpace(head) + " " + strings.TrimSpace(s[end:]))
+	}
+}
+
+// trimTrailingPreposition drops the word that introduced a clock reading, so
+// "Встреча в 14:00" becomes "Встреча" and "с 11:30 до 12:15 Созвон" does not
+// keep a dangling "с". It repeats, because a range has two of them.
+func trimTrailingPreposition(s string) string {
+	preps := []string{"в", "с", "до", "от", "at", "from", "to"}
+	for again := true; again; {
+		again = false
+		s = strings.TrimRight(s, " \t")
+		for _, p := range preps {
+			if s == p {
+				return ""
+			}
+			if strings.HasSuffix(s, " "+p) {
+				s = s[:len(s)-len(p)-1]
+				again = true
+				break
+			}
+		}
+	}
+	return s
+}
+
+func hasLetters(s string) bool {
+	for _, r := range s {
+		if unicode.IsLetter(r) {
+			return true
+		}
+	}
+	return false
+}

internal/calendar/ambient_test.go

@@ -0,0 +1,148 @@
+package calendar
+
+import (
+	"testing"
+	"time"
+)
+
+func TestEventFromNotification(t *testing.T) {
+	posted := time.Date(2026, 8, 3, 9, 40, 0, 0, time.FixedZone("+04", 4*3600))
+
+	tests := []struct {
+		name        string
+		title, text string
+		wantOK      bool
+		wantSummary string
+		wantStart   string // "15:04"
+		wantEnd     string
+	}{
+		{
+			name:        "range in the body",
+			title:       "Планёрка",
+			text:        "10:00-10:30",
+			wantOK:      true,
+			wantSummary: "Планёрка",
+			wantStart:   "10:00", wantEnd: "10:30",
+		},
+		{
+			name:        "russian preposition and single time",
+			title:       "Встреча с подрядчиком в 14:00",
+			wantOK:      true,
+			wantSummary: "Встреча с подрядчиком",
+			wantStart:   "14:00", wantEnd: "14:30",
+		},
+		{
+			name:        "en dash range",
+			title:       "Sprint review",
+			text:        "Today 16:00 – 17:00, Meet",
+			wantOK:      true,
+			wantSummary: "Sprint review",
+			wantStart:   "16:00", wantEnd: "17:00",
+		},
+		{
+			name:        "до as a range separator",
+			title:       "Созвон",
+			text:        "с 11:30 до 12:15",
+			wantOK:      true,
+			wantSummary: "Созвон",
+			wantStart:   "11:30", wantEnd: "12:15",
+		},
+		{
+			name:        "dotted clock",
+			title:       "Обед 13.00",
+			wantOK:      true,
+			wantSummary: "Обед",
+			wantStart:   "13:00", wantEnd: "13:30",
+		},
+		{
+			name:        "range crossing midnight",
+			title:       "Ночной релиз",
+			text:        "23:30-00:30",
+			wantOK:      true,
+			wantSummary: "Ночной релиз",
+			wantStart:   "23:30", wantEnd: "00:30",
+		},
+		// The conservative half: no clock reading, no event.
+		{name: "no time at all", title: "3 новых письма", wantOK: false},
+		{name: "bare hour is not a time", title: "Планёрка в 14", wantOK: false},
+		{name: "unread count", title: "Входящие", text: "12 непрочитанных", wantOK: false},
+		{name: "a date is not a clock", title: "Отчёт", text: "срок 2026.08.15", wantOK: false},
+		{name: "time but nothing named", title: "10:00-10:30", wantOK: false},
+		{name: "impossible clock", title: "Смена 99:99", wantOK: false},
+		{name: "empty", wantOK: false},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			ev, ok := EventFromNotification(Notification{
+				Package: "com.google.android.gm",
+				Title:   tt.title,
+				Text:    tt.text,
+				Posted:  posted,
+			})
+			if ok != tt.wantOK {
+				t.Fatalf("ok = %v, want %v (event %+v)", ok, tt.wantOK, ev)
+			}
+			if !ok {
+				return
+			}
+			if ev.Summary != tt.wantSummary {
+				t.Errorf("summary = %q, want %q", ev.Summary, tt.wantSummary)
+			}
+			if got := ev.Start.Format("15:04"); got != tt.wantStart {
+				t.Errorf("start = %s, want %s", got, tt.wantStart)
+			}
+			if got := ev.End.Format("15:04"); got != tt.wantEnd {
+				t.Errorf("end = %s, want %s", got, tt.wantEnd)
+			}
+			if !ev.End.After(ev.Start) {
+				t.Errorf("end %v must be after start %v", ev.End, ev.Start)
+			}
+			// The event lands on the day the phone showed it, in the phone's
+			// location — not shifted into UTC.
+			if ev.Start.Location() != posted.Location() {
+				t.Errorf("location = %v, want %v", ev.Start.Location(), posted.Location())
+			}
+			if y, m, d := ev.Start.Date(); y != 2026 || m != time.August || d != 3 {
+				t.Errorf("date = %d-%02d-%02d, want 2026-08-03", y, m, d)
+			}
+		})
+	}
+}
+
+func TestEventFromNotificationNeedsPostedAt(t *testing.T) {
+	if _, ok := EventFromNotification(Notification{Title: "Планёрка 10:00"}); ok {
+		t.Error("a notification with no posted_at has no date to sit on")
+	}
+}
+
+// An ambient event must never be indistinguishable from a calendar read.
+func TestAmbientEventsAreStoredAtReducedConfidence(t *testing.T) {
+	ev, ok := EventFromNotification(Notification{
+		Title:  "Планёрка 10:00-10:30",
+		Posted: time.Date(2026, 8, 3, 9, 0, 0, 0, time.UTC),
+	})
+	if !ok {
+		t.Fatal("expected an event")
+	}
+	if FactKey(ev) == "" || FactValue(ev) == "" {
+		t.Fatal("ambient events must use the shared fact encoding")
+	}
+	if AmbientConfidence >= 1.0 {
+		t.Fatal("ambient confidence must be below a calendar read's")
+	}
+}
+
+func TestStripClock(t *testing.T) {
+	tests := []struct{ in, want string }{
+		{"Встреча в 14:00", "Встреча"},
+		{"Планёрка 10:00-10:30", "Планёрка"},
+		{"с 11:30 до 12:15 Созвон", "Созвон"},
+		{"Ничего", "Ничего"},
+	}
+	for _, tt := range tests {
+		if got := stripClock(tt.in); got != tt.want {
+			t.Errorf("stripClock(%q) = %q, want %q", tt.in, got, tt.want)
+		}
+	}
+}

internal/calendar/calendar.go

@@ -0,0 +1,129 @@
+// Package calendar is the one calendar data model the rest of maven shares:
+// an Event, the iCal text it is parsed from and rendered to, and the fact
+// encoding that puts it in the store.
+//
+// It exists because three separate features read or write the same events and
+// must agree on their shape: the CalDAV read side (cmd/mavcaldav, Vikunja
+// #126/#127), the write-only render target that publishes maven's own
+// reminders as a calendar (#127), and the day plan that recites them (#128).
+// Before this package the parse lived inline in cmd/mavcaldav and the fact key
+// format was a Sprintf in two places.
+//
+// The package is pure: no HTTP, no store, no clock of its own. Callers own the
+// impurity, the way internal/morning and internal/loop do.
+package calendar
+
+import (
+	"fmt"
+	"sort"
+	"strings"
+	"time"
+)
+
+// Fact sources. A calendar event reaches the store as a
+// `facts (kind=env, key=calendar_event_..., source=<one of these>)` row, and
+// the source is the whole provenance story:
+//
+//   - SourcePersonal — maven's own Radicale, read AND rendered to. Canonical
+//     state stays in sqlite; the calendar is a render target (#127).
+//   - SourceWork — a work calendar, read-only by definition (#126). Nothing in
+//     maven ever writes to it: no code path pairs this source with a PUT.
+//   - SourceAmbient — inferred from an Android notification-listener relay
+//     rather than read from a server (#126). Confidence is below 1.0 because a
+//     notification is a signal about an event, not the event.
+const (
+	SourcePersonal = "poll:caldav"
+	SourceWork     = "poll:caldav:work"
+	SourceAmbient  = "ambient:notif"
+)
+
+// AmbientConfidence — the confidence a notification-derived event is stored
+// with. A parsed notification line is evidence, not a reading of the calendar,
+// so it must never be indistinguishable from one (#126).
+const AmbientConfidence = 0.6
+
+// Sources lists every source a calendar event may legitimately carry, for the
+// store query that reads the calendar back out. Ordered from most to least
+// trusted.
+func Sources() []string {
+	return []string{SourcePersonal, SourceWork, SourceAmbient}
+}
+
+// ReadOnlySource reports whether events from this source may never be written
+// back. The work calendar is read-only by definition — see #126: maven holding
+// a credential that can write to an employer's calendar is the thing the task
+// exists to avoid.
+func ReadOnlySource(source string) bool {
+	return source == SourceWork || source == SourceAmbient
+}
+
+// Event — one calendar entry. UID is the iCal UID when the event was parsed
+// from a server and the identity maven renders under when it publishes one;
+// Start/End are instants. All-day events are not modelled: the busy gate and
+// the day plan both need a time of day, and an all-day marker answers neither.
+type Event struct {
+	UID     string
+	Summary string
+	Start   time.Time
+	End     time.Time
+}
+
+// FactKey is the store key for an event: one key per day per summary, stable
+// across polls so re-reading an unchanged calendar rewrites nothing.
+//
+// The date prefix is load-bearing — store.CalendarEvents selects a day range
+// by key prefix, not by a timestamp column.
+func FactKey(e Event) string {
+	return fmt.Sprintf("calendar_event_%s_%s", e.Start.Format("20060102"), safeKey(e.Summary))
+}
+
+// FactValue is the human-readable rendering stored as the fact value, and the
+// string the day plan and the query path read back.
+func FactValue(e Event) string {
+	return fmt.Sprintf("%s @ %s-%s", e.Summary, e.Start.Format("15:04"), e.End.Format("15:04"))
+}
+
+// KeyPrefixForDay is the fact-key prefix covering one calendar day. The store
+// range-scans between two of these.
+func KeyPrefixForDay(day time.Time) string {
+	return fmt.Sprintf("calendar_event_%s", day.Format("20060102"))
+}
+
+// Busy reports whether any event covers the instant now — the read the loop
+// gate uses to suppress nudges during a meeting.
+func Busy(events []Event, now time.Time) bool {
+	for _, e := range events {
+		if !now.Before(e.Start) && now.Before(e.End) {
+			return true
+		}
+	}
+	return false
+}
+
+// Overlapping returns the events intersecting [from, to), sorted by start.
+func Overlapping(events []Event, from, to time.Time) []Event {
+	var out []Event
+	for _, e := range events {
+		if e.End.After(from) && e.Start.Before(to) {
+			out = append(out, e)
+		}
+	}
+	sort.Slice(out, func(i, j int) bool { return out[i].Start.Before(out[j].Start) })
+	return out
+}
+
+// safeKey makes a summary safe to use inside a fact key (ASCII alphanumerics
+// and dashes). Non-Latin summaries collapse to their punctuation, which is why
+// the day prefix carries the identity and this only disambiguates within a day.
+func safeKey(s string) string {
+	var b strings.Builder
+	for _, r := range s {
+		switch {
+		case (r >= 'a' && r <= 'z') || (r >= 'A' && r <= 'Z') || (r >= '0' && r <= '9') || r == '-':
+			b.WriteRune(r)
+		case r == ' ' || r == '_':
+			b.WriteRune('-')
+		}
+	}
+	return b.String()
+}

internal/calendar/calendar_test.go

@@ -0,0 +1,202 @@
+package calendar
+
+import (
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestParseICalDayKeepsOnlyToday(t *testing.T) {
+	now := time.Date(2026, 7, 3, 12, 0, 0, 0, time.UTC)
+
+	body := []byte(`BEGIN:VCALENDAR
+BEGIN:VEVENT
+UID:a@example
+DTSTART:20260703T090000Z
+DTEND:20260703T100000Z
+SUMMARY:Morning standup
+END:VEVENT
+BEGIN:VEVENT
+DTSTART:20260703T140000Z
+DTEND:20260703T150000Z
+SUMMARY:Team sync
+END:VEVENT
+BEGIN:VEVENT
+DTSTART:20260702T140000Z
+DTEND:20260702T150000Z
+SUMMARY:Yesterday retro
+END:VEVENT
+BEGIN:VEVENT
+DTSTART:20260704T090000Z
+DTEND:20260704T100000Z
+SUMMARY:Tomorrow standup
+END:VEVENT
+BEGIN:VEVENT
+DTSTART;VALUE=DATE:20260704
+DTEND;VALUE=DATE:20260705
+SUMMARY:All-day event
+END:VEVENT
+END:VCALENDAR`)
+
+	events := ParseICalDay(body, now)
+	if len(events) != 2 {
+		t.Fatalf("got %d events, want 2 (today only, no all-day/past/future)", len(events))
+	}
+	if events[0].Summary != "Morning standup" || events[0].UID != "a@example" {
+		t.Errorf("events[0] = %+v", events[0])
+	}
+	if !events[0].Start.Equal(time.Date(2026, 7, 3, 9, 0, 0, 0, time.UTC)) {
+		t.Errorf("events[0].Start = %v", events[0].Start)
+	}
+	if !events[0].End.Equal(time.Date(2026, 7, 3, 10, 0, 0, 0, time.UTC)) {
+		t.Errorf("events[0].End = %v", events[0].End)
+	}
+	if events[1].Summary != "Team sync" {
+		t.Errorf("events[1].Summary = %q", events[1].Summary)
+	}
+}
+
+// Regression: "today" is the owner's day, in the owner's location. Taking the
+// day number off a local clock but building the boundaries in UTC made the
+// evening fall outside the window on any box east of Greenwich.
+func TestParseICalDayUsesOwnersDay(t *testing.T) {
+	plus4 := time.FixedZone("+04", 4*60*60)
+	// 01:00 on Aug 1 local is 21:00 on Jul 31 UTC.
+	now := time.Date(2026, 8, 1, 1, 0, 0, 0, plus4)
+	body := []byte("BEGIN:VCALENDAR\nBEGIN:VEVENT\n" +
+		"DTSTART:20260731T195406Z\nDTEND:20260731T235406Z\nSUMMARY:Current meeting\n" +
+		"END:VEVENT\nEND:VCALENDAR")
+
+	events := ParseICalDay(body, now)
+	if len(events) != 1 {
+		t.Fatalf("got %d events, want the in-progress one", len(events))
+	}
+	if !Busy(events, now.UTC()) {
+		t.Error("an event in progress right now must read as busy")
+	}
+}
+
+func TestParseVEVENT(t *testing.T) {
+	block := "DTSTART;TZID=Europe/Moscow:20260703T130000\nDTEND:20260703T140000Z\nSUMMARY:Stand up meeting"
+	e, ok := parseVEVENT(block)
+	if !ok {
+		t.Fatal("expected a parsed event")
+	}
+	if !e.Start.Equal(time.Date(2026, 7, 3, 13, 0, 0, 0, time.UTC)) {
+		t.Errorf("start = %v", e.Start)
+	}
+	if !e.End.Equal(time.Date(2026, 7, 3, 14, 0, 0, 0, time.UTC)) {
+		t.Errorf("end = %v", e.End)
+	}
+	if e.Summary != "Stand up meeting" {
+		t.Errorf("summary = %q", e.Summary)
+	}
+
+	allDay := "DTSTART;VALUE=DATE:20260703\nDTEND;VALUE=DATE:20260704\nSUMMARY:All-day"
+	if _, ok := parseVEVENT(allDay); ok {
+		t.Error("all-day event should be rejected")
+	}
+}
+
+func TestParseDT(t *testing.T) {
+	tests := []struct {
+		name   string
+		line   string
+		want   time.Time
+		wantOK bool
+	}{
+		{"UTC", "DTEND:20260703T100000Z", time.Date(2026, 7, 3, 10, 0, 0, 0, time.UTC), true},
+		{"local", "DTSTART;TZID=Europe/Moscow:20260703T130000", time.Date(2026, 7, 3, 13, 0, 0, 0, time.UTC), true},
+		{"all-day", "DTSTART;VALUE=DATE:20260703", time.Time{}, false},
+		{"garbage", "DTSTART:garbage", time.Time{}, false},
+	}
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got, ok := parseDT(tt.line)
+			if ok != tt.wantOK {
+				t.Errorf("ok = %v, want %v", ok, tt.wantOK)
+			}
+			if !got.Equal(tt.want) {
+				t.Errorf("got %v, want %v", got, tt.want)
+			}
+		})
+	}
+}
+
+func TestSafeKey(t *testing.T) {
+	tests := []struct{ in, want string }{
+		{"Stand up meeting", "Stand-up-meeting"},
+		{"Hello_World", "Hello-World"},
+		{"special@#$chars!!", "specialchars"},
+		{"ALL_CAPS_123", "ALL-CAPS-123"},
+	}
+	for _, tt := range tests {
+		if got := safeKey(tt.in); got != tt.want {
+			t.Errorf("safeKey(%q) = %q, want %q", tt.in, got, tt.want)
+		}
+	}
+}
+
+func TestFactKeyAndValue(t *testing.T) {
+	e := Event{
+		Summary: "Team sync",
+		Start:   time.Date(2026, 7, 3, 14, 0, 0, 0, time.UTC),
+		End:     time.Date(2026, 7, 3, 15, 0, 0, 0, time.UTC),
+	}
+	if got, want := FactKey(e), "calendar_event_20260703_Team-sync"; got != want {
+		t.Errorf("FactKey = %q, want %q", got, want)
+	}
+	if got, want := FactValue(e), "Team sync @ 14:00-15:00"; got != want {
+		t.Errorf("FactValue = %q, want %q", got, want)
+	}
+	if got, want := KeyPrefixForDay(e.Start), "calendar_event_20260703"; got != want {
+		t.Errorf("KeyPrefixForDay = %q, want %q", got, want)
+	}
+	if !strings.HasPrefix(FactKey(e), KeyPrefixForDay(e.Start)) {
+		t.Error("FactKey must start with the day prefix the store range-scans on")
+	}
+}
+
+func TestBusyAndOverlapping(t *testing.T) {
+	base := time.Date(2026, 7, 3, 0, 0, 0, 0, time.UTC)
+	events := []Event{
+		{Summary: "late", Start: base.Add(15 * time.Hour), End: base.Add(16 * time.Hour)},
+		{Summary: "early", Start: base.Add(9 * time.Hour), End: base.Add(10 * time.Hour)},
+	}
+	if !Busy(events, base.Add(9*time.Hour+30*time.Minute)) {
+		t.Error("should be busy inside the early event")
+	}
+	if Busy(events, base.Add(12*time.Hour)) {
+		t.Error("should be free at noon")
+	}
+	// Half-open: the end instant is free.
+	if Busy(events, base.Add(10*time.Hour)) {
+		t.Error("the end instant should not count as busy")
+	}
+	got := Overlapping(events, base.Add(8*time.Hour), base.Add(11*time.Hour))
+	if len(got) != 1 || got[0].Summary != "early" {
+		t.Fatalf("Overlapping = %+v", got)
+	}
+	all := Overlapping(events, base, base.AddDate(0, 0, 1))
+	if len(all) != 2 || all[0].Summary != "early" {
+		t.Fatalf("Overlapping must sort by start: %+v", all)
+	}
+}
+
+func TestSourceTrust(t *testing.T) {
+	if ReadOnlySource(SourcePersonal) {
+		t.Error("the personal calendar is the one maven may render to")
+	}
+	if !ReadOnlySource(SourceWork) {
+		t.Error("the work calendar must be read-only")
+	}
+	if !ReadOnlySource(SourceAmbient) {
+		t.Error("an ambient notification is not a writable calendar")
+	}
+	if AmbientConfidence >= 1.0 {
+		t.Error("ambient events must be less trusted than a calendar read")
+	}
+	if len(Sources()) != 3 {
+		t.Errorf("Sources() = %v", Sources())
+	}
+}

internal/calendar/ical.go

@@ -0,0 +1,145 @@
+package calendar
+
+import (
+	"fmt"
+	"strings"
+	"time"
+)
+
+// ParseICal scans iCal text for VEVENT components and returns the events
+// overlapping [from, to). All-day events are skipped: parseDT reports no time
+// for a VALUE=DATE value, and an event with no clock reading answers neither
+// the busy gate nor the day plan.
+func ParseICal(body []byte, from, to time.Time) []Event {
+	var events []Event
+	text := string(body)
+	for {
+		i := strings.Index(text, "BEGIN:VEVENT")
+		if i < 0 {
+			break
+		}
+		text = text[i+len("BEGIN:VEVENT"):]
+		j := strings.Index(text, "END:VEVENT")
+		if j < 0 {
+			break
+		}
+		block := text[:j]
+		text = text[j+len("END:VEVENT"):]
+
+		e, ok := parseVEVENT(block)
+		if !ok {
+			continue
+		}
+		if e.End.After(from) && e.Start.Before(to) {
+			events = append(events, e)
+		}
+	}
+	return events
+}
+
+// ParseICalDay is ParseICal over the calendar day containing now, in now's own
+// location — the window cmd/mavcaldav polls.
+//
+// The location matters. The old inline version took the day number off a local
+// clock reading but built the boundaries in UTC, so east of Greenwich the
+// window was shifted by the offset and part of the evening fell outside
+// "today": on a +04 box after 20:00 UTC the poller saw an empty calendar. The
+// owner's day is the day the day plan and the busy gate mean.
+func ParseICalDay(body []byte, now time.Time) []Event {
+	y, m, d := now.Date()
+	start := time.Date(y, m, d, 0, 0, 0, 0, now.Location())
+	return ParseICal(body, start, start.AddDate(0, 0, 1))
+}
+
+// parseVEVENT extracts UID, start, end and summary from a VEVENT block.
+// Reports false for all-day events and parse failures.
+func parseVEVENT(block string) (Event, bool) {
+	var e Event
+	for _, line := range strings.Split(block, "\n") {
+		line = strings.TrimSpace(line)
+		switch {
+		case strings.HasPrefix(line, "DTSTART"):
+			if t, ok := parseDT(line); ok {
+				e.Start = t
+			}
+		case strings.HasPrefix(line, "DTEND"):
+			if t, ok := parseDT(line); ok {
+				e.End = t
+			}
+		case strings.HasPrefix(line, "SUMMARY"):
+			e.Summary = afterColon(line)
+		case strings.HasPrefix(line, "UID"):
+			e.UID = afterColon(line)
+		}
+	}
+	if e.Start.IsZero() || e.End.IsZero() {
+		return Event{}, false
+	}
+	return e, true
+}
+
+func afterColon(line string) string {
+	if i := strings.Index(line, ":"); i >= 0 {
+		return strings.TrimSpace(line[i+1:])
+	}
+	return ""
+}
+
+// parseDT parses a DTSTART/DTEND value:
+//
+//   - UTC:      DTEND:20260703T100000Z
+//   - Local:    DTSTART;TZID=Europe/Moscow:20260703T130000
+//   - All-day:  DTSTART;VALUE=DATE:20260703  (rejected)
+//
+// A local time is read as UTC, the behaviour cmd/mavcaldav has always had: the
+// CalDAV server and the poller run in the same timezone, and the busy gate only
+// needs busy/not-busy to be right.
+func parseDT(line string) (time.Time, bool) {
+	if strings.Contains(line, "VALUE=DATE:") {
+		return time.Time{}, false
+	}
+	i := strings.LastIndex(line, ":")
+	if i < 0 {
+		return time.Time{}, false
+	}
+	val := strings.TrimSuffix(strings.TrimSpace(line[i+1:]), "Z")
+	t, err := time.Parse("20060102T150405", val)
+	if err != nil {
+		return time.Time{}, false
+	}
+	return t.UTC(), true
+}
+
+// RenderICal wraps events in a VCALENDAR body suitable for PUTting to a CalDAV
+// collection. One event per file is the CalDAV convention, so callers normally
+// pass a single event.
+//
+// This is the write half of #127 and it only ever renders: the canonical state
+// is sqlite, the calendar is a view of it. Nothing reads a rendered file back.
+func RenderICal(events []Event) string {
+	var b strings.Builder
+	b.WriteString("BEGIN:VCALENDAR\r\nVERSION:2.0\r\nPRODID:-//maven//local calendar//RU\r\n")
+	for _, e := range events {
+		b.WriteString("BEGIN:VEVENT\r\n")
+		fmt.Fprintf(&b, "UID:%s\r\n", escapeText(e.UID))
+		fmt.Fprintf(&b, "DTSTAMP:%s\r\n", e.Start.UTC().Format("20060102T150405Z"))
+		fmt.Fprintf(&b, "DTSTART:%s\r\n", e.Start.UTC().Format("20060102T150405Z"))
+		fmt.Fprintf(&b, "DTEND:%s\r\n", e.End.UTC().Format("20060102T150405Z"))
+		fmt.Fprintf(&b, "SUMMARY:%s\r\n", escapeText(e.Summary))
+		b.WriteString("END:VEVENT\r\n")
+	}
+	b.WriteString("END:VCALENDAR\r\n")
+	return b.String()
+}
+
+// escapeText applies RFC 5545 TEXT escaping and strips the line breaks that
+// would otherwise let a reminder payload inject iCal properties.
+func escapeText(s string) string {
+	s = strings.ReplaceAll(s, "\\", "\\\\")
+	s = strings.ReplaceAll(s, ";", "\\;")
+	s = strings.ReplaceAll(s, ",", "\\,")
+	s = strings.ReplaceAll(s, "\r\n", "\\n")
+	s = strings.ReplaceAll(s, "\n", "\\n")
+	s = strings.ReplaceAll(s, "\r", "\\n")
+	return s
+}

internal/calendar/ical_render_test.go

@@ -0,0 +1,69 @@
+package calendar
+
+import (
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestRenderICalRoundTrips(t *testing.T) {
+	e := ReminderEvent(7, time.Date(2026, 8, 1, 18, 30, 0, 0, time.UTC), "позвонить маме", 0)
+	if e.UID != "maven-reminder-7" {
+		t.Errorf("UID = %q", e.UID)
+	}
+	if got := e.End.Sub(e.Start); got != DefaultReminderDuration {
+		t.Errorf("duration = %v, want %v", got, DefaultReminderDuration)
+	}
+	if got, want := ReminderPath(7), "maven-reminder-7.ics"; got != want {
+		t.Errorf("ReminderPath = %q, want %q", got, want)
+	}
+
+	body := RenderICal([]Event{e})
+	if !strings.HasPrefix(body, "BEGIN:VCALENDAR\r\n") || !strings.HasSuffix(body, "END:VCALENDAR\r\n") {
+		t.Fatalf("not a VCALENDAR body:\n%s", body)
+	}
+
+	back := ParseICal([]byte(body), e.Start.Add(-time.Hour), e.Start.Add(time.Hour))
+	if len(back) != 1 {
+		t.Fatalf("got %d events back, want 1:\n%s", len(back), body)
+	}
+	if back[0].UID != e.UID || back[0].Summary != e.Summary {
+		t.Errorf("round trip lost identity: %+v", back[0])
+	}
+	if !back[0].Start.Equal(e.Start) || !back[0].End.Equal(e.End) {
+		t.Errorf("round trip lost times: %+v", back[0])
+	}
+}
+
+func TestRenderICalIsDeterministic(t *testing.T) {
+	e := ReminderEvent(1, time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC), "выпить воды", 0)
+	if RenderICal([]Event{e}) != RenderICal([]Event{e}) {
+		t.Error("the same reminder must render byte-identically, or every poll re-PUTs it")
+	}
+}
+
+// A reminder payload is owner-supplied text. It must not be able to close the
+// VEVENT and inject properties of its own.
+func TestRenderICalEscapesInjection(t *testing.T) {
+	e := ReminderEvent(2, time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC),
+		"обед\r\nEND:VEVENT\r\nBEGIN:VEVENT\r\nSUMMARY:injected", 0)
+	body := RenderICal([]Event{e})
+	// Count line-initial occurrences: the escaped text still contains the
+	// characters "BEGIN:VEVENT", it just can no longer start a line.
+	if n := strings.Count(body, "\r\nBEGIN:VEVENT\r\n"); n != 1 {
+		t.Fatalf("payload injected a second VEVENT (%d):\n%s", n, body)
+	}
+	if n := strings.Count(body, "\r\nEND:VEVENT\r\n"); n != 1 {
+		t.Fatalf("payload closed the VEVENT early (%d):\n%s", n, body)
+	}
+	if !strings.Contains(body, `SUMMARY:обед\nEND:VEVENT`) {
+		t.Errorf("newlines should be escaped, not dropped:\n%s", body)
+	}
+}
+
+func TestReminderEventEmptyPayload(t *testing.T) {
+	e := ReminderEvent(3, time.Date(2026, 8, 1, 9, 0, 0, 0, time.UTC), "   ", 0)
+	if e.Summary != "напоминание" {
+		t.Errorf("Summary = %q, want the neutral RU fallback", e.Summary)
+	}
+}

internal/calendar/reminder.go

@@ -0,0 +1,45 @@
+package calendar
+
+import (
+	"fmt"
+	"strings"
+	"time"
+)
+
+// ReminderUIDPrefix namespaces every event maven publishes. Two reasons it is a
+// fixed prefix and not a random UUID: the render is idempotent (the same
+// reminder always lands on the same UID, so re-rendering overwrites instead of
+// duplicating), and everything maven owns in the target collection is
+// identifiable at a glance — she never touches a file she did not create.
+const ReminderUIDPrefix = "maven-reminder-"
+
+// DefaultReminderDuration — how long a rendered reminder occupies. A reminder
+// is an instant, a calendar entry is a span, so one has to be invented; 30
+// minutes reads as a block in a calendar app without swallowing the afternoon.
+const DefaultReminderDuration = 30 * time.Minute
+
+// ReminderEvent maps a reminder to the event that represents it. id and fire
+// come from the store; payload is the RU text as the owner said it, rendered
+// verbatim as the summary — the calendar is a view of sqlite, not a place to
+// rephrase.
+func ReminderEvent(id int64, fire time.Time, payload string, dur time.Duration) Event {
+	if dur <= 0 {
+		dur = DefaultReminderDuration
+	}
+	summary := strings.TrimSpace(payload)
+	if summary == "" {
+		summary = "напоминание"
+	}
+	return Event{
+		UID:     fmt.Sprintf("%s%d", ReminderUIDPrefix, id),
+		Summary: summary,
+		Start:   fire,
+		End:     fire.Add(dur),
+	}
+}
+
+// ReminderPath is the collection-relative filename for a rendered reminder.
+// One event per resource, per the CalDAV convention.
+func ReminderPath(id int64) string {
+	return fmt.Sprintf("%s%d.ics", ReminderUIDPrefix, id)
+}

internal/config/config.go

@@ -140,6 +140,27 @@ type Config struct {
 	// item. See internal/morning for the evaluation engine. Empty ⇒ disabled.
 	MorningRoutines []MorningRoutineConfig `json:"morning_routines,omitempty"`
 
+	// PatternProposals — whether a routine the digestion tick inferred on its
+	// own may be announced, and how often. nil / absent ⇒ silent detection
+	// only: proposals are written for /routines and never announced. See
+	// PatternProposalConfig.
+	PatternProposals *PatternProposalConfig `json:"pattern_proposals,omitempty"`
+
+	// MemoryEval — background memory evaluation (internal/memeval). nil /
+	// absent ⇒ no evaluation loop at all. See MemoryEvalConfig.
+	MemoryEval *MemoryEvalConfig `json:"memory_eval,omitempty"`
+
+	// Email — mail ingestion (Vikunja #246). nil / absent ⇒ core refuses
+	// ipc.MethodIngestMail outright, so a mail reader cannot make Maven read a
+	// mailbox by merely existing. See EmailConfig; the IMAP host and credential
+	// live in the reader (cmd/mavmaild), never here.
+	Email *EmailConfig `json:"email,omitempty"`
+
+	// Feeds — RSS/Atom feed reading (Vikunja #258). nil / absent ⇒ no feed is
+	// ever fetched: reading the outside world is off unless configured, like
+	// the weather and telegram. See FeedsConfig.
+	Feeds *FeedsConfig `json:"feeds,omitempty"`
+
 	// Praxis — the ecosystem attention-state service. When configured, maven
 	// calls the Praxis HTTP tools API for attention listing and item lifecycle.
 	// Maven never touches Praxis's database directly (ecosystem invariant: no
@@ -258,18 +279,25 @@ type VoiceConfig struct {
 	RouterThreshold float64 `json:"router_threshold,omitempty"`
 
 	// LLMRouter — route with the resident model instead of the embedding
-	// classifier. Measured on the held-out fixture (ROUTING-EVAL-31-07-2026.md)
+	// classifier. On by default since Vikunja #320.
-	// the model gets 50.0% of intents right against the classifier's 36.8%, but
-	// it costs about 800ms per turn instead of 30ms.
 	//
-	// TODO: the default stays false until two things land.
+	// Measured on the held-out fixture (ROUTING-EVAL-31-07-2026.md): 63.2% of
-	//   1. The LLM router cannot refuse. LLMRouter.Route hardcodes
+	// intents right against the classifier's 50.0%, and no route errors. It
-	//      Confidence: 1.0, so the stage-3 clarify gate never fires and an
+	// costs about 1s per turn instead of 30ms.
-	//      unclear utterance becomes a confident wrong action (Vikunja #359).
+	//
-	//   2. Extractor.Extract never runs on an LLM decision, so acts arrive with
+	// It is safe to leave on. The model can refuse — it answers "unknown" when
-	//      no Fn and reminders with no Time.
+	// it cannot route, and the turn drops to the classifier and its clarify
-	// Turning this on today makes routing more accurate and less safe.
+	// gate. Any LLM error does the same, so a turn never breaks on the model.
-	LLMRouter bool `json:"llm_router,omitempty"`
+	// Slot extraction runs on LLM decisions too, so acts get their Fn and
+	// reminders their Time.
+	//
+	// Set it false to go back to the classifier, e.g. on a box with no
+	// llama-server or when 1s a turn is too slow.
+	//
+	// It is a pointer so that "missing from the file" and "explicitly false"
+	// are different things: missing means on, false means off. Read it with
+	// UseLLMRouter(), not directly.
+	LLMRouter *bool `json:"llm_router,omitempty"`
 
 	// QueryMinScore — the note-recall confidence gate. Top cosine below this
 	// ⇒ "I don't know" instead of a guess. Tuned for the ONNX embedder (0.55);
@@ -277,12 +305,31 @@ type VoiceConfig struct {
 	// default if unset.
 	QueryMinScore float64 `json:"query_min_score,omitempty"`
 
+	// QueryMinMargin — the second half of the recall gate: the top hit must
+	// beat the runner-up by more than this. The absolute score above cannot do
+	// the job on its own, because the e5 embedder puts every cosine in one
+	// narrow high band, so a made-up question scores as high as a real one.
+	// The margin asks whether one note is clearly the best instead.
+	// Negative ⇒ off. 0 ⇒ the default below.
+	QueryMinMargin float64 `json:"query_min_margin,omitempty"`
+
+	// ClarifyMaxAttempts — how many clarifying questions she may ask about one
+	// request before she gives up and says she did not understand. Default 3.
+	ClarifyMaxAttempts int `json:"clarify_max_attempts,omitempty"`
+
 	// Persona — optional prompt prefix that tunes maven's character. Prepended
 	// to every LLM system prompt (nudge phrasing, note queries, general
 	// knowledge). Empty string ⇒ current hardcoded persona (feminine-gendered
 	// Russian self-reference). Example: "Be formal and answer in English only."
 	Persona string `json:"persona,omitempty"`
 
+	// OwnerName / City — optional facts about the owner, added to the shared
+	// context block (internal/persona). Empty is fine: the block still states
+	// who he is grammatically (a man, addressed as "ты") and the current time.
+	// Nothing about correct behaviour may depend on these being filled in.
+	OwnerName string `json:"owner_name,omitempty"`
+	City      string `json:"city,omitempty"`
+
 	// Weather — the weather provider config. nil ⇒ the daemon wires
 	// the stub provider (returns ErrNotConfigured — "погода не настроена").
 	// Set provider to "open-meteo" to use the keyless Open-Meteo API.
@@ -326,6 +373,129 @@ type DigestConfig struct {
 	SeverityCeiling int      `json:"severity_ceiling,omitempty"` // max sev batched
 }
 
+// PatternProposalConfig — announcement policy for routines the digestion tick
+// inferred by itself (Vikunja #247, #43).
+//
+// Detection is always on and always silent by default: the tick writes a
+// proposed_routines row and the /routines page shows it. Notify is what turns
+// "she noticed" into "she said something", and it is OFF unless configured —
+// Maven is not a nag and not autonomous, so a behaviour that speaks without
+// being asked has to be switched on deliberately, like weather and telegram.
+//
+// When Notify is on, the announcement is still heavily restrained:
+//   - at most one proposal per tick, however many were detected;
+//   - at most one per Cooldown across all pairs (not per pair), so a batch of
+//     freshly-detected patterns cannot turn into a queue of interruptions;
+//   - through the ordinary care-class gate (quiet hours / away / snooze), at
+//     sev1 — the lowest severity there is. A proposal is the least urgent
+//     thing Maven can say.
+//
+// A pair is only ever announced once, because it is only ever proposed once:
+// proposed_routines is UNIQUE(action, object) and the row survives dismissal.
+type PatternProposalConfig struct {
+	// Notify — announce newly inferred routines. Default false.
+	Notify bool `json:"notify,omitempty"`
+
+	// Cooldown — minimum spacing between two proposal announcements. 0 ⇒
+	// DefaultProposalCooldown (24h).
+	Cooldown Duration `json:"cooldown,omitempty"`
+}
+
+// AnnounceProposals reports whether inferred routines may be announced. Safe
+// on a nil receiver — an absent config block means silent detection.
+func (p *PatternProposalConfig) AnnounceProposals() bool {
+	return p != nil && p.Notify
+}
+
+// FeedsConfig — the RSS/Atom reader (Vikunja #258, docs/plans/13-rss-news-feeds.md).
+//
+// Absent ⇒ off, and off means no outbound request at all. Present with an empty
+// `sources` list is also off — a poller with nothing to poll is not wired.
+//
+// What a feed may NOT do here: speak. Items are written as notes with source
+// "rss:<name>" and read back when he asks; nothing is dispatched, nudged or
+// announced on arrival. That is the "not a nag" constraint, and it is why there
+// is no severity or channel field in this block to reach for.
+type FeedsConfig struct {
+	// Sources — the feeds to read. Empty ⇒ the reader stays down.
+	Sources []FeedSourceConfig `json:"sources,omitempty"`
+
+	// PollInterval — default per-feed cadence. 0 ⇒ rss.DefaultPollInterval (30m).
+	PollInterval Duration `json:"poll_interval,omitempty"`
+
+	// MaxItems — most items kept from one feed in one poll. 0 ⇒
+	// rss.DefaultMaxItems (5). This is the "не завали мне /dash" knob.
+	MaxItems int `json:"max_items,omitempty"`
+
+	// MaxAge — on a first poll (no saved mark), how far back to take items.
+	// 0 ⇒ rss.DefaultMaxAge (24h), so switching a feed on imports today, not
+	// the archive.
+	MaxAge Duration `json:"max_age,omitempty"`
+
+	// AllowHosts — when set, the reader may only connect to these hosts (and
+	// their subdomains). The feed URLs' own hosts are added automatically, so
+	// this is only needed to be stricter than that.
+	AllowHosts []string `json:"allow_hosts,omitempty"`
+
+	// Timeout — per-request budget. 0 ⇒ webfetch.DefaultTimeout.
+	Timeout Duration `json:"timeout,omitempty"`
+
+	// MaxBytes — response size cap. 0 ⇒ webfetch.DefaultMaxBytes (2 MiB).
+	MaxBytes int64 `json:"max_bytes,omitempty"`
+}
+
+// FeedSourceConfig — one feed.
+type FeedSourceConfig struct {
+	Name     string   `json:"name"`               // note source is "rss:<name>"
+	URL      string   `json:"url"`                // http(s) only
+	Category string   `json:"category,omitempty"` // "технологии" — what "что нового по X?" matches
+	Interval Duration `json:"interval,omitempty"` // 0 ⇒ FeedsConfig.PollInterval
+	Include  []string `json:"include,omitempty"`  // keep only items containing one of these
+	Exclude  []string `json:"exclude,omitempty"`  // drop items containing any of these
+}
+
+// MemoryEvalConfig — the background memory-evaluation loop (Vikunja #248).
+// Absent ⇒ off, like every other capability that costs something the owner did
+// not ask for. Each evaluation is a full LLM round-trip on the one resident
+// model, which is the same model answering him; running it hourly by default
+// would put a multi-second stall in front of an occasional voice turn for a
+// feature he may not want.
+//
+// The loop only ever writes notes (source infer:memory-eval, visible on
+// /dash). It cannot speak — see internal/memeval.
+type MemoryEvalConfig struct {
+	// Interval — how often to evaluate. 0 ⇒ DefaultMemoryEvalInterval.
+	Interval Duration `json:"interval,omitempty"`
+
+	// MaxItems — recent facts / notes / nudges fed into one evaluation.
+	// 0 ⇒ memeval.DefaultMaxItems.
+	MaxItems int `json:"max_items,omitempty"`
+
+	// MinConfidence — observations the model scores below this are dropped.
+	// 0 ⇒ memeval.DefaultMinConfidence.
+	MinConfidence float64 `json:"min_confidence,omitempty"`
+}
+
+// EmailConfig — core's half of the email reader: how many task candidates one
+// message may produce, and how long the extraction call may take.
+//
+// There is deliberately nothing about a mailbox here. Core does not connect to
+// IMAP, does not know an account exists, and holds no mail credential — the
+// reader daemon does, the same split mavpoll uses for the zenmoney token. This
+// block only says "extraction is allowed, with these bounds".
+type EmailConfig struct {
+	// MaxTasks — candidates per message. 0 ⇒ email.MaxCandidates (3).
+	MaxTasks int `json:"max_tasks,omitempty"`
+
+	// Timeout — per-message extraction budget. 0 ⇒ DefaultEmailTimeout. This is
+	// a Thinking model reading a mail; nobody is waiting on the answer, but a
+	// hung llama-server must not pin the reader's connection forever.
+	Timeout Duration `json:"timeout,omitempty"`
+}
+
+// DefaultEmailTimeout — extraction budget per message.
+const DefaultEmailTimeout = 2 * time.Minute
+
 // PhraserConfig — the LLM-backed phraser seam. The daemon spawns llama-server
 // as a managed subprocess and sends chat-completion requests to phrase nudge
 // and reminder messages. nil ⇒ the template-based Stub is used instead.
@@ -343,6 +513,12 @@ type PhraserConfig struct {
 	NGpuLayers int      `json:"n_gpu_layers,omitempty"`
 	NCtx       int      `json:"n_ctx,omitempty"`
 	Timeout    Duration `json:"timeout,omitempty"`
+
+	// LLMNudges — let the model word nudges again. Off by default: nudges are
+	// worded from hand-written Russian templates now (the model broke the
+	// persona and invented units). Chat, query and reminder phrasing always go
+	// through the model regardless. See phraser.Config.LLMNudges.
+	LLMNudges bool `json:"llm_nudges,omitempty"`
 }
 
 // EmbedderConfig — paths for the ONNX multilingual embedder. The daemon
@@ -404,9 +580,27 @@ const (
 	DefaultAutotuneInterval = 10 * time.Minute
 	DefaultRouterThreshold  = 0.55
 	DefaultQueryMinScore    = 0.55
-	DefaultToolTimeout      = 30 * time.Second
+	// Read off the margin sweep in internal/memory/recalleval on the e5
+	// embedder: 0.008 answers 68% of real questions (down from 72%) and cuts
+	// false recall from 5/5 to 1/5. Every larger delta costs real recall
+	// without removing that last one until 0.020, which drops recall to 44%.
+	DefaultQueryMinMargin = 0.008
+	// DefaultClarifyMaxAttempts — see dialogue.DefaultMaxAttempts.
+	DefaultClarifyMaxAttempts = 3
+	DefaultToolTimeout        = 30 * time.Second
+	// DefaultLLMRouter — route with the resident model unless told otherwise.
+	DefaultLLMRouter = true
 
 	DefaultFactEnrichmentInterval = 30 * time.Second
+
+	// DefaultProposalCooldown — one inferred-routine announcement per day at
+	// most. A proposal is never urgent; if two patterns surface in the same
+	// hour, the second one waits, and the /routines page has it either way.
+	DefaultProposalCooldown = 24 * time.Hour
+
+	// DefaultMemoryEvalInterval — the plan's cadence (1h) for the memory
+	// evaluation loop, applied only when the block is present at all.
+	DefaultMemoryEvalInterval = time.Hour
 )
 
 // Load reads the JSON config at path and applies defaults. A missing file is
@@ -479,6 +673,30 @@ func (c *Config) applyDefaults() {
 		c.Digest.SeverityCeiling = 2
 	}
 
+	// Absent block stays nil (⇒ silent detection). Present-but-partial gets the
+	// cooldown default, so `{"notify": true}` is enough to switch it on.
+	if c.PatternProposals != nil && c.PatternProposals.Cooldown <= 0 {
+		c.PatternProposals.Cooldown = Duration(DefaultProposalCooldown)
+	}
+
+	// Same rule: absent stays nil (⇒ no evaluation loop), present gets defaults
+	// so `{}` is a valid "on with the plan's cadence".
+	if c.MemoryEval != nil && c.MemoryEval.Interval <= 0 {
+		c.MemoryEval.Interval = Duration(DefaultMemoryEvalInterval)
+	}
+
+	// Same rule again: absent stays nil (⇒ mail ingestion refused), present gets
+	// the timeout default so `{}` is a valid "on with the defaults".
+	if c.Email != nil && c.Email.Timeout <= 0 {
+		c.Email.Timeout = Duration(DefaultEmailTimeout)
+	}
+
+	// A feeds block with no sources is the same as no block: nothing to poll,
+	// nothing wired. Normalising it to nil keeps that "off" in one place.
+	if c.Feeds != nil && len(c.Feeds.Sources) == 0 {
+		c.Feeds = nil
+	}
+
 	if c.Voice != nil {
 		if c.Voice.RouterThreshold <= 0 {
 			c.Voice.RouterThreshold = DefaultRouterThreshold
@@ -486,9 +704,24 @@ func (c *Config) applyDefaults() {
 		if c.Voice.QueryMinScore <= 0 {
 			c.Voice.QueryMinScore = DefaultQueryMinScore
 		}
+		// Unset ⇒ default. Negative is how you turn the margin off on purpose,
+		// so it is clamped to 0 rather than replaced by the default.
+		switch {
+		case c.Voice.QueryMinMargin == 0:
+			c.Voice.QueryMinMargin = DefaultQueryMinMargin
+		case c.Voice.QueryMinMargin < 0:
+			c.Voice.QueryMinMargin = 0
+		}
+		if c.Voice.ClarifyMaxAttempts <= 0 {
+			c.Voice.ClarifyMaxAttempts = DefaultClarifyMaxAttempts
+		}
 		if c.Voice.ToolTimeout <= 0 {
 			c.Voice.ToolTimeout = Duration(DefaultToolTimeout)
 		}
+		if c.Voice.LLMRouter == nil {
+			on := DefaultLLMRouter
+			c.Voice.LLMRouter = &on
+		}
 	}
 
 	// routines: default severity to care-class (1) — the safe floor: a
@@ -507,6 +740,15 @@ func (c *Config) applyDefaults() {
 	}
 }
 
+// UseLLMRouter reports whether to route with the resident model. Unset means
+// on; only an explicit false in the config turns it off.
+func (v *VoiceConfig) UseLLMRouter() bool {
+	if v == nil || v.LLMRouter == nil {
+		return DefaultLLMRouter
+	}
+	return *v.LLMRouter
+}
+
 func (c *Config) validate() error {
 	if c.Phraser != nil {
 		if c.Phraser.ModelPath == "" {

internal/config/config_test.go

@@ -35,6 +35,27 @@ func TestLoadDefaults(t *testing.T) {
 	}
 }
 
+// Nudges come from templates unless the config says otherwise.
+func TestPhraserLLMNudgesDefaultsOff(t *testing.T) {
+	p := writeConfig(t, `{"phraser":{"model_path":"/tmp/m.gguf"}}`)
+	c, err := Load(p)
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if c.Phraser.LLMNudges {
+		t.Error("llm_nudges defaults on; templates must be the default")
+	}
+
+	p = writeConfig(t, `{"phraser":{"model_path":"/tmp/m.gguf","llm_nudges":true}}`)
+	c, err = Load(p)
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if !c.Phraser.LLMNudges {
+		t.Error("llm_nudges:true did not parse")
+	}
+}
+
 func TestLoadDurationsParse(t *testing.T) {
 	p := writeConfig(t, `{"tick_interval":"90s","repeat_interval":"10m"}`)
 	c, err := Load(p)
@@ -171,14 +192,26 @@ func TestWeatherConfigNilOK(t *testing.T) {
 	}
 }
 
-func TestLLMRouterDefaultsOff(t *testing.T) {
+func TestLLMRouterDefaultsOn(t *testing.T) {
 	p := writeConfig(t, `{"voice":{"enabled":true,"bind":"127.0.0.1:9100"}}`)
 	c, err := Load(p)
 	if err != nil {
 		t.Fatalf("Load: %v", err)
 	}
-	if c.Voice.LLMRouter {
+	if !c.Voice.UseLLMRouter() {
-		t.Error("voice.llm_router absent should mean false")
+		t.Error("voice.llm_router absent should mean on")
+	}
+}
+
+// Missing and explicitly false must not mean the same thing.
+func TestLLMRouterExplicitFalseTurnsItOff(t *testing.T) {
+	p := writeConfig(t, `{"voice":{"enabled":true,"bind":"127.0.0.1:9100","llm_router":false}}`)
+	c, err := Load(p)
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if c.Voice.UseLLMRouter() {
+		t.Error("voice.llm_router false should turn it off")
 	}
 }
 
@@ -188,7 +221,7 @@ func TestLLMRouterRead(t *testing.T) {
 	if err != nil {
 		t.Fatalf("Load: %v", err)
 	}
-	if !c.Voice.LLMRouter {
+	if !c.Voice.UseLLMRouter() {
 		t.Error("voice.llm_router true was not read")
 	}
 }
@@ -210,3 +243,53 @@ func TestDurationRoundTrip(t *testing.T) {
 		t.Errorf("round-trip = %v, want %v", d2, d)
 	}
 }
+
+// Both new opt-in capabilities follow the same rule: absent block ⇒ nil ⇒ the
+// behaviour does not exist. Presence is the enable act, so a bare `{}` block is
+// valid and gets the defaults filled in.
+func TestOptInBlocksAbsentStayNil(t *testing.T) {
+	c, err := Load(writeConfig(t, `{}`))
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if c.PatternProposals != nil {
+		t.Errorf("pattern_proposals absent but got %+v", c.PatternProposals)
+	}
+	if c.PatternProposals.AnnounceProposals() {
+		t.Error("AnnounceProposals() true with no config block")
+	}
+	if c.MemoryEval != nil {
+		t.Errorf("memory_eval absent but got %+v", c.MemoryEval)
+	}
+}
+
+func TestOptInBlocksGetDefaultsWhenPresent(t *testing.T) {
+	c, err := Load(writeConfig(t, `{"pattern_proposals":{"notify":true},"memory_eval":{}}`))
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if !c.PatternProposals.AnnounceProposals() {
+		t.Error("notify:true did not enable announcements")
+	}
+	if time.Duration(c.PatternProposals.Cooldown) != DefaultProposalCooldown {
+		t.Errorf("proposal cooldown = %v, want %v", c.PatternProposals.Cooldown, DefaultProposalCooldown)
+	}
+	if time.Duration(c.MemoryEval.Interval) != DefaultMemoryEvalInterval {
+		t.Errorf("memory eval interval = %v, want %v", c.MemoryEval.Interval, DefaultMemoryEvalInterval)
+	}
+}
+
+// Notify is off even when the block exists — the block is where you tune it,
+// notify:true is the act that lets her speak.
+func TestPatternProposalNotifyDefaultsOff(t *testing.T) {
+	c, err := Load(writeConfig(t, `{"pattern_proposals":{"cooldown":"6h"}}`))
+	if err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if c.PatternProposals.AnnounceProposals() {
+		t.Error("notify defaulted to on")
+	}
+	if time.Duration(c.PatternProposals.Cooldown) != 6*time.Hour {
+		t.Errorf("cooldown = %v, want 6h", c.PatternProposals.Cooldown)
+	}
+}

internal/delivery/dispatcher.go

@@ -137,9 +137,10 @@ func NewDispatcher(cfg Config) *Dispatcher {
 // picks for (severity, presence), sends via the matching sink, and records
 // one nudge row per successful send. returns the dispatches (one per channel).
 //
-// a Drop channel = no send, no record (the nudge was suppressed by routing,
+// a Drop channel = no send (the nudge was suppressed by routing, not by a
-// not by a failure — "a missed water nudge is noise"). a nil sink = channel
+// failure — "a missed water nudge is noise"), but it does leave a 'dropped'
-// not wired, skip silently. a send error stops the dispatch and returns what
+// outbox row so the suppression is visible. a nil sink = channel not wired,
+// skip silently. a send error stops the dispatch and returns what
 // got through — the daemon decides whether to retry.
 func (d *Dispatcher) DispatchNudge(ctx context.Context, pn PhrasedNudge, now time.Time) ([]Dispatch, error) {
 	c := pn.Candidate
@@ -148,6 +149,16 @@ func (d *Dispatcher) DispatchNudge(ctx context.Context, pn PhrasedNudge, now tim
 	for i := 0; i < len(channels); i++ {
 		ch := channels[i]
 		if ch == ChannelDrop {
+			// the routing table suppressed this nudge on purpose (a care nudge
+			// while you're away is noise). that stays — but it must not be
+			// invisible, or "she dropped it" and "the rule never fired" look
+			// the same afterwards. no nudges row: that table feeds the
+			// ignored_rate signal, and a nudge nobody could see must not
+			// count as ignored.
+			id := d.beginOutbox(ctx, "nudge", c.Rule.Name, 0, ch, pn.Summary, now)
+			d.completeOutbox(ctx, id, store.DeliveryDropped, now)
+			log.Printf("dispatcher: dropped %s (sev%d, presence=%s) — routing table suppressed it",
+				c.Rule.Name, c.Severity, c.State.Presence)
 			continue
 		}
 		s := Sendable{
@@ -396,6 +407,13 @@ func messageForChannel(s Sendable) string {
 	if !isAway(s.Channel) {
 		return s.Body
 	}
+	return AwayMessage(s)
+}
+
+// AwayMessage — the only text an off-box channel may ever carry. Exported so
+// the away sinks share this one rule instead of each inventing a fallback: the
+// summary if we have one, otherwise a fixed generic line. Never the body.
+func AwayMessage(s Sendable) string {
 	if s.Summary != "" {
 		return s.Summary
 	}

internal/delivery/durability_test.go

@@ -37,10 +37,12 @@ func TestVoiceNoSessionFallthroughLeavesOutboxTrail(t *testing.T) {
 			[]string{"voice", "ntfy"}, []string{store.DeliveryFailed, store.DeliverySent}},
 		{"sev4 falls through to telegram", loop.Sev4,
 			[]string{"voice", "telegram"}, []string{store.DeliveryFailed, store.DeliverySent}},
-		{"sev1 does not fall through", loop.Sev1,
+		// care severities still don't reach an away channel; since #370 the
-			[]string{"voice"}, []string{store.DeliveryFailed}},
+		// drop itself is a visible row instead of nothing.
-		{"sev2 does not fall through", loop.Sev2,
+		{"sev1 drops instead of falling through", loop.Sev1,
-			[]string{"voice"}, []string{store.DeliveryFailed}},
+			[]string{"voice", "drop"}, []string{store.DeliveryFailed, store.DeliveryDropped}},
+		{"sev2 drops instead of falling through", loop.Sev2,
+			[]string{"voice", "drop"}, []string{store.DeliveryFailed, store.DeliveryDropped}},
 	}
 	for _, c := range cases {
 		t.Run(c.name, func(t *testing.T) {

internal/delivery/ntfysink/ntfysink.go

@@ -2,10 +2,10 @@
 //
 // ntfy is the away-channel for sev3 (ops soft) nudges, sev4 (ops hard)
 // nudges when present (alongside voice), and reminders when away. the
-// message body is the Sendable's Summary — the minimal-body rule from the
+// message body is delivery.AwayMessage — the minimal-body rule from the
 // spec ("disk low on homesrv," not detail; no shoulder-surf exfil through
-// the relay). voice gets Body; away channels get Summary, enforced at the
+// the relay). the dispatcher already strips detail off away sendables; the
-// sink so a phraser bug can't exfil.
+// sink uses the same helper so it can't leak the body on its own either.
 //
 // ntfy runs locally (docker, 127.0.0.1:8085, deny-all auth). maven publishes
 // with a dedicated user (write-only to maven-* topics) — the credential is a
@@ -69,18 +69,14 @@ func New(cfg Config) (*Sink, error) {
 	}, nil
 }
 
-// Send publishes one notification to ntfy. the body is the Sendable's Summary
+// Send publishes one notification to ntfy. the body is the minimal away
-// (minimal body); Title is "maven" (consistent sender identity on the lock
+// message (never the full body); Title is "maven" (consistent sender identity
-// screen — the content is in the body). Priority maps from severity/kind so
+// on the lock screen — the content is in the body). Priority maps from severity/kind so
 // the phone client can ring differently for an alarm vs a soft ops nudge.
 func (s *Sink) Send(ctx context.Context, d delivery.Sendable) error {
-	body := d.Summary
+	// never fall back to d.Body: ntfy leaves the box, so an empty summary gets
-	if body == "" {
+	// a generic line instead of the full detail.
-		body = d.Body // terse full message beats no message
+	body := delivery.AwayMessage(d)
-	}
-	if body == "" {
-		return fmt.Errorf("ntfysink: empty message for %s", d.Channel)
-	}
 
 	req, err := http.NewRequestWithContext(ctx, http.MethodPost, s.topicURL(), strings.NewReader(body))
 	if err != nil {

internal/delivery/ntfysink/ntfysink_test.go

@@ -147,9 +147,9 @@ func TestSendBodyIsSummaryNotFullBody(t *testing.T) {
 	}
 }
 
-func TestSendFallsBackToBodyWhenSummaryEmpty(t *testing.T) {
+func TestSendNeverSendsTheBodyWhenSummaryEmpty(t *testing.T) {
-	// a terse full message is better than no message; the phraser should
+	// #368: this used to fall back to the full body. ntfy leaves the box, so
-	// produce a summary for away-bound severities, but don't silently drop.
+	// an empty summary gets a fixed generic line plus the rule name instead.
 	rs := newRecordingServer(t, 200, "")
 	srv := httptest.NewServer(rs.handler())
 	defer srv.Close()
@@ -160,12 +160,15 @@ func TestSendFallsBackToBodyWhenSummaryEmpty(t *testing.T) {
 		t.Fatalf("Send: %v", err)
 	}
 	_, _, body, _, _, _ := rs.snapshot()
-	if body != s.Body {
+	want := delivery.GenericAwayMessage + ": service_down"
-		t.Fatalf("fallback body: want %q, got %q", s.Body, body)
+	if body != want {
+		t.Fatalf("body: want %q, got %q", want, body)
 	}
 }
 
-func TestSendRejectsEmptyMessage(t *testing.T) {
+func TestSendNeverSendsAnEmptyMessage(t *testing.T) {
+	// with nothing at all to say we still send the generic line — an away
+	// channel can never carry detail, but it also never goes out blank.
 	rs := newRecordingServer(t, 200, "")
 	srv := httptest.NewServer(rs.handler())
 	defer srv.Close()
@@ -173,9 +176,13 @@ func TestSendRejectsEmptyMessage(t *testing.T) {
 	sink, _ := New(Config{BaseURL: srv.URL, Topic: "maven"})
 	s := nudgeSendable(loop.Sev3, "")
 	s.Body = ""
-	err := sink.Send(context.Background(), s)
+	s.RuleName = ""
-	if err == nil {
+	if err := sink.Send(context.Background(), s); err != nil {
-		t.Fatal("want error for empty message")
+		t.Fatalf("Send: %v", err)
+	}
+	_, _, body, _, _, _ := rs.snapshot()
+	if body != delivery.GenericAwayMessage {
+		t.Fatalf("body: want %q, got %q", delivery.GenericAwayMessage, body)
 	}
 }
 

internal/delivery/routing_table_test.go

@@ -208,10 +208,9 @@ func TestAwayChannelsGetMinimalBody(t *testing.T) {
 // TestCareAwayDropIsRecorded — DESIGN.md's drop is a decision ("a missed water
 // nudge is noise, a missed backup failure isn't"), so it should be visible
 // rather than vanish. Today drop is a bare `continue`: no nudge row, no outbox
-// attempt, no log — nothing an operator can see afterwards.
+// attempt, no log — nothing an operator can see afterwards. now it leaves a
+// 'dropped' outbox row.
 func TestCareAwayDropIsRecorded(t *testing.T) {
-	t.Skip("not implemented: dispatcher.go:149-151 skips a Drop channel with no record; there is no 'dropped' outcome in store/delivery.go:16-21")
-
 	ob := &fakeOutbox{}
 	d := NewDispatcher(Config{Voice: &fakeSink{}, Nudges: &fakeNudgeRecorder{}, Outbox: ob})
 

internal/delivery/telegramsink/telegramsink.go

@@ -2,11 +2,12 @@
 //
 // telegram is the away-channel for sev4 (ops hard) nudges — "disk-fire alarm
 // at 2am routes to telegram, repeat til ack." the message body is the
-// Sendable's Summary — the minimal-body rule from the spec ("disk low on
+// delivery.AwayMessage — the minimal-body rule from the spec ("disk low on
-// homesrv," not detail; no shoulder-surf exfil through the relay). voice gets
+// homesrv," not detail; no shoulder-surf exfil through the relay). the
-// Body; away channels get Summary, enforced at the sink so a phraser bug can't
+// dispatcher already strips detail off away sendables; the sink uses the same
-// exfil. additionally, protect_content=true is passed on every send so the
+// helper so it can't leak the body on its own either. additionally,
-// message can't be forwarded out of the chat — locks the minimal body further.
+// protect_content=true is passed on every send so the message can't be
+// forwarded out of the chat — locks the minimal body further.
 //
 // telegram's bot API is region-restricted for this homesrv — direct egress to
 // api.telegram.org is unreliable. the spec's "away channels leave the box —
@@ -140,18 +141,13 @@ type telegramResp struct {
 }
 
 // Send publishes one message to the configured telegram chat. the body is the
-// Sendable's Summary (minimal body); empty Summary falls back to Body (terse
+// minimal away message (never the full body). protect_content=true so even
-// full message beats no message). protect_content=true so a phraser bug (Body
+// that can't be forwarded onward by the user or a chat observer — locks the
-// leaking detail through Summary) can't be forwarded onward by the user or a
+// minimal-body rule at the channel's own last mile.
-// chat observer — locks the minimal-body rule at the channel's own last mile.
 func (s *Sink) Send(ctx context.Context, d delivery.Sendable) error {
-	body := d.Summary
+	// never fall back to d.Body: telegram leaves the box, so an empty summary
-	if body == "" {
+	// gets a generic line instead of the full detail.
-		body = d.Body
+	body := delivery.AwayMessage(d)
-	}
-	if body == "" {
-		return fmt.Errorf("telegramsink: empty message for %s", d.Channel)
-	}
 
 	payload := sendMessageReq{
 		ChatID:              s.cfg.ChatID,

internal/delivery/telegramsink/telegramsink_test.go

@@ -173,9 +173,9 @@ func TestSendBodyIsSummaryNotFullBody(t *testing.T) {
 	}
 }
 
-func TestSendFallsBackToBodyWhenSummaryEmpty(t *testing.T) {
+func TestSendNeverSendsTheBodyWhenSummaryEmpty(t *testing.T) {
-	// terse full message beats none; the phraser should produce a summary for
+	// #368: this used to fall back to the full body. telegram leaves the box,
-	// away-bound severities, but don't silently drop.
+	// so an empty summary gets a fixed generic line plus the rule name.
 	rs := newRecordingServer(t, 200, "")
 	srv := httptest.NewServer(rs.handler())
 	defer srv.Close()
@@ -188,12 +188,14 @@ func TestSendFallsBackToBodyWhenSummaryEmpty(t *testing.T) {
 	_, _, body, _, _ := rs.snapshot()
 	var req sendMessageReq
 	_ = json.Unmarshal([]byte(body), &req)
-	if req.Text != s.Body {
+	want := delivery.GenericAwayMessage + ": service_down"
-		t.Fatalf("fallback text: want %q, got %q", s.Body, req.Text)
+	if req.Text != want {
+		t.Fatalf("text: want %q, got %q", want, req.Text)
 	}
 }
 
-func TestSendRejectsEmptyMessage(t *testing.T) {
+func TestSendNeverSendsAnEmptyMessage(t *testing.T) {
+	// with nothing at all to say we still send the generic line.
 	rs := newRecordingServer(t, 200, "")
 	srv := httptest.NewServer(rs.handler())
 	defer srv.Close()
@@ -201,9 +203,15 @@ func TestSendRejectsEmptyMessage(t *testing.T) {
 	sink, _ := New(sinkCfg(srv.URL))
 	s := nudgeSendable(loop.Sev4, "")
 	s.Body = ""
-	err := sink.Send(context.Background(), s)
+	s.RuleName = ""
-	if err == nil {
+	if err := sink.Send(context.Background(), s); err != nil {
-		t.Fatal("want error for empty message")
+		t.Fatalf("Send: %v", err)
+	}
+	_, _, body, _, _ := rs.snapshot()
+	var req sendMessageReq
+	_ = json.Unmarshal([]byte(body), &req)
+	if req.Text != delivery.GenericAwayMessage {
+		t.Fatalf("text: want %q, got %q", delivery.GenericAwayMessage, req.Text)
 	}
 }
 

internal/dialogue/clarify.go

@@ -17,10 +17,11 @@ const (
 	SlotText  Slot = "text"  // Slots.Text
 )
 
-// MaxAttempts is 1 because Maven is not a nag (DESIGN.md § Non-goals). She asks
+// DefaultMaxAttempts — how many questions she may ask about one request.
-// one clarifying question. If the answer still leaves the slot empty she drops
+// Three, because after three tries the likely problem is that she misheard the
-// the request instead of asking again.
+// whole request, not one slot — so another question about that slot won't help.
-const MaxAttempts = 1
+// Configurable: voice.clarify_max_attempts.
+const DefaultMaxAttempts = 3
 
 // PendingQuestion is what Maven holds while she waits for an answer to an open
 // question. Unlike the yes/no confirms in cmd/mavend/voice.go, the answer here
@@ -32,7 +33,17 @@ type PendingQuestion struct {
 	Utterance string // the user's original raw words
 	Asked     time.Time
 	TTL       time.Duration
-	Attempts  int // questions already asked; capped by MaxAttempts
+	Attempts  int // questions already asked
+	// MaxAttempts caps Attempts. 0 ⇒ DefaultMaxAttempts.
+	MaxAttempts int
+}
+
+// maxAttempts is MaxAttempts with the default filled in.
+func (q *PendingQuestion) maxAttempts() int {
+	if q.MaxAttempts <= 0 {
+		return DefaultMaxAttempts
+	}
+	return q.MaxAttempts
 }
 
 func (q *PendingQuestion) IsExpired(now time.Time) bool {
@@ -41,12 +52,9 @@ func (q *PendingQuestion) IsExpired(now time.Time) bool {
 
 // CanAsk reports whether Maven may ask another question about this request.
 func (q *PendingQuestion) CanAsk() bool {
-	return q.Attempts < MaxAttempts
+	return q.Attempts < q.maxAttempts()
 }
 
-// TODO: the daemon will phrase the question text from Missing (one short ru
-// question per Slot, feminine self-reference) and speak it here.
-
 // ClarifyStore holds the parked questions. Same shape and locking as
 // SessionStore: keyed by dialogue id, expired entries dropped on read.
 type ClarifyStore struct {
@@ -67,8 +75,7 @@ func NewClarifyStore(defaultTTL time.Duration) *ClarifyStore {
 	}
 }
 
-// TODO: the daemon will Put a question here when Decision.Clarify fires, in
+// Put parks a question. Called on a clarify decision (cmd/mavend/clarify.go).
-// place of the flat "не разобрала" reply (cmd/mavend/voice.go).
 func (s *ClarifyStore) Put(id string, q *PendingQuestion) {
 	if q.TTL <= 0 {
 		q.TTL = s.defaultTTL
@@ -78,8 +85,7 @@ func (s *ClarifyStore) Put(id string, q *PendingQuestion) {
 	s.mu.Unlock()
 }
 
-// TODO: the daemon will Get on the next turn, parse that turn into Slots, call
+// Get returns the live parked question, or nil when there is none.
-// Answer, and Delete — the open-question twin of resolveConfirm.
 func (s *ClarifyStore) Get(id string, now time.Time) *PendingQuestion {
 	s.mu.RLock()
 	q, ok := s.questions[id]
@@ -94,6 +100,21 @@ func (s *ClarifyStore) Get(id string, now time.Time) *PendingQuestion {
 	return q
 }
 
+// TakeExpired reports whether a question was parked here but its TTL ran out,
+// and drops it. Get drops such a question silently, which leaves the user
+// thinking his request is still alive — the caller uses this to tell him it is
+// gone before treating his words as a fresh utterance.
+func (s *ClarifyStore) TakeExpired(id string, now time.Time) bool {
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	q, ok := s.questions[id]
+	if !ok || !q.IsExpired(now) {
+		return false
+	}
+	delete(s.questions, id)
+	return true
+}
+
 func (s *ClarifyStore) Delete(id string) {
 	s.mu.Lock()
 	delete(s.questions, id)
@@ -101,44 +122,45 @@ func (s *ClarifyStore) Delete(id string) {
 }
 
 // Answer merges the slots parsed from the user's answer into the parked ones.
-// Only the slots listed in Missing are filled, and an already filled slot is
+// Only the slots listed in Missing are touched. Within those, a value the answer
-// never overwritten — the answer completes the original request, it does not
+// carries WINS over what was parked: she asked about this slot, so «нет, в пять»
-// restate it. Parsing the answer text into `answer` is the caller's job; this
+// after «в три» must replace the time, not be thrown away.
-// package must stay free of internal/router.
+//
+// This is the clarify answer only. A correction in a fresh turn ("вообще-то
+// перенеси на пять") is a different code path (followUpMerge) — not here.
+//
+// Parsing the answer text into `answer` is the caller's job; this package must
+// stay free of internal/router.
 func (q *PendingQuestion) Answer(text string, answer Slots) Slots {
 	out := q.Slots
 	for _, slot := range q.Missing {
 		switch slot {
 		case SlotTime:
-			if !out.HasTime && answer.HasTime {
+			if answer.HasTime {
 				out.Time = answer.Time
 				out.HasTime = true
 			}
 		case SlotKey:
-			if !out.HasKey && answer.HasKey {
+			if answer.HasKey {
 				out.Key = answer.Key
 				out.HasKey = true
 			}
 		case SlotValue:
-			if out.Value == "" && answer.Value != "" {
+			if answer.Value != "" {
 				out.Value = answer.Value
 			}
 		case SlotFn:
-			if !out.HasFn && answer.HasFn {
+			if answer.HasFn {
 				out.Fn = answer.Fn
 				out.HasFn = true
-				if len(out.Args) == 0 {
+				out.Args = append([]string(nil), answer.Args...)
-					out.Args = append([]string(nil), answer.Args...)
-				}
 			}
 		case SlotText:
-			if out.Text == "" {
+			if answer.Text != "" {
-				if answer.Text != "" {
+				out.Text = answer.Text
-					out.Text = answer.Text
+			} else if out.Text == "" {
-				} else {
+				// No parse for a text slot — the raw answer IS the text.
-					// No parse for a text slot — the raw answer IS the text.
+				out.Text = text
-					out.Text = text
-				}
 			}
 		}
 	}

internal/dialogue/clarify_test.go

@@ -1,6 +1,7 @@
 package dialogue
 
 import (
+	"reflect"
 	"testing"
 	"time"
 )
@@ -59,6 +60,28 @@ func TestClarifyStoreGetPutDelete(t *testing.T) {
 	}
 }
 
+// TestClarifyStoreTakeExpired — TakeExpired reports (and drops) only a question
+// whose TTL ran out.
+func TestClarifyStoreTakeExpired(t *testing.T) {
+	s := NewClarifyStore(time.Minute)
+	if s.TakeExpired("voice", base) {
+		t.Fatal("nothing parked ⇒ nothing expired")
+	}
+	s.Put("voice", &PendingQuestion{Missing: []Slot{SlotTime}, Asked: base, TTL: time.Minute})
+	if s.TakeExpired("voice", base.Add(30*time.Second)) {
+		t.Fatal("a live question must not report as expired")
+	}
+	if s.Get("voice", base.Add(30*time.Second)) == nil {
+		t.Fatal("a live question must survive TakeExpired")
+	}
+	if !s.TakeExpired("voice", base.Add(2*time.Minute)) {
+		t.Fatal("a stale question must report as expired")
+	}
+	if s.TakeExpired("voice", base.Add(2*time.Minute)) {
+		t.Fatal("TakeExpired must drop the question, so the second call is false")
+	}
+}
+
 func TestNewClarifyStoreDefaultTTL(t *testing.T) {
 	s := NewClarifyStore(0)
 	q := &PendingQuestion{Asked: base}
@@ -89,12 +112,13 @@ func TestAnswerFillsOnlyMissingSlots(t *testing.T) {
 			want:    Slots{Text: "напомни позвонить", Time: answerTime, HasTime: true},
 		},
 		{
-			name:    "does not overwrite a filled time",
+			// He restated it: «нет, в пять». The new value wins.
+			name:    "a restated time overwrites the parked one",
 			parked:  Slots{Time: other, HasTime: true},
 			missing: []Slot{SlotTime},
-			text:    "в три",
+			text:    "нет, в три",
 			answer:  Slots{Time: answerTime, HasTime: true},
-			want:    Slots{Time: other, HasTime: true},
+			want:    Slots{Time: answerTime, HasTime: true},
 		},
 		{
 			name:    "ignores slots that were not missing",
@@ -121,12 +145,12 @@ func TestAnswerFillsOnlyMissingSlots(t *testing.T) {
 			want:    Slots{Fn: "restart", Args: []string{"nginx"}, HasFn: true},
 		},
 		{
-			name:    "keeps existing args when fn was already known",
+			name:    "a restated fn replaces the fn and its args",
 			parked:  Slots{Fn: "restart", Args: []string{"nginx"}, HasFn: true},
 			missing: []Slot{SlotFn},
 			text:    "останови postgres",
 			answer:  Slots{Fn: "stop", Args: []string{"postgres"}, HasFn: true},
-			want:    Slots{Fn: "restart", Args: []string{"nginx"}, HasFn: true},
+			want:    Slots{Fn: "stop", Args: []string{"postgres"}, HasFn: true},
 		},
 		{
 			name:    "raw answer becomes the text when nothing was parsed",
@@ -157,36 +181,34 @@ func TestAnswerFillsOnlyMissingSlots(t *testing.T) {
 	for _, tc := range cases {
 		t.Run(tc.name, func(t *testing.T) {
 			q := &PendingQuestion{Slots: tc.parked, Missing: tc.missing, Asked: base}
-			got := q.Answer(tc.text, tc.answer)
+			// Whole-struct compare: a new field in Slots is covered for free.
-			if got.Time != tc.want.Time || got.HasTime != tc.want.HasTime ||
+			if got := q.Answer(tc.text, tc.answer); !reflect.DeepEqual(got, tc.want) {
-				got.Key != tc.want.Key || got.HasKey != tc.want.HasKey ||
-				got.Value != tc.want.Value || got.Text != tc.want.Text ||
-				got.Fn != tc.want.Fn || got.HasFn != tc.want.HasFn {
 				t.Fatalf("Answer = %+v, want %+v", got, tc.want)
 			}
-			if len(got.Args) != len(tc.want.Args) {
-				t.Fatalf("Args = %v, want %v", got.Args, tc.want.Args)
-			}
-			for i := range got.Args {
-				if got.Args[i] != tc.want.Args[i] {
-					t.Fatalf("Args = %v, want %v", got.Args, tc.want.Args)
-				}
-			}
 		})
 	}
 }
 
-func TestCanAskCapsAtOneQuestion(t *testing.T) {
+func TestCanAskAllowsThreeQuestionsByDefault(t *testing.T) {
-	if MaxAttempts != 1 {
+	if DefaultMaxAttempts != 3 {
-		t.Fatalf("MaxAttempts = %d, want 1 (Maven asks once, she is not a nag)", MaxAttempts)
+		t.Fatalf("DefaultMaxAttempts = %d, want 3", DefaultMaxAttempts)
 	}
-	q := &PendingQuestion{Asked: base}
+	q := &PendingQuestion{Asked: base} // MaxAttempts unset ⇒ the default
-	if !q.CanAsk() {
+	for i := 0; i < 3; i++ {
-		t.Fatal("a fresh question should be askable")
+		if !q.CanAsk() {
+			t.Fatalf("question %d should be allowed", i+1)
+		}
+		q.Attempts++
 	}
-	q.Attempts = MaxAttempts
 	if q.CanAsk() {
-		t.Fatal("the question should not be asked twice")
+		t.Fatal("a fourth question must not be allowed")
+	}
+}
+
+func TestCanAskHonoursConfiguredMax(t *testing.T) {
+	q := &PendingQuestion{Asked: base, MaxAttempts: 1, Attempts: 1}
+	if q.CanAsk() {
+		t.Fatal("MaxAttempts 1 means one question only")
 	}
 }
 

internal/dialogue/session.go

@@ -1,8 +1,12 @@
 package dialogue
 
 import (
+	"context"
+	"encoding/json"
 	"sync"
 	"time"
+
+	"github.com/kami/maven/internal/store"
 )
 
 type Intent string
@@ -50,10 +54,22 @@ func (s *Session) IsExpired(now time.Time) bool {
 	return now.After(s.Timestamp.Add(s.TTL))
 }
 
+// SessionPersister — the bit of the store the session needs, as an interface
+// so tests can swap it out. Data is an opaque blob: the store never looks
+// inside, we encode the session as JSON here.
+type SessionPersister interface {
+	SaveDialogueSession(ctx context.Context, id string, data []byte, ts time.Time, ttl time.Duration) error
+	DeleteDialogueSession(ctx context.Context, id string) error
+	LoadDialogueSessions(ctx context.Context, now time.Time) ([]store.DialogueSessionRow, error)
+}
+
+// SessionStore keeps the live sessions in a map (the fast path) and mirrors
+// every write to the persister, so a daemon restart can load them back.
 type SessionStore struct {
 	mu         sync.RWMutex
 	sessions   map[string]*Session
 	defaultTTL time.Duration
+	persist    SessionPersister // may be nil: memory only (tests, no-store paths)
 }
 
 func NewSessionStore(defaultTTL time.Duration) *SessionStore {
@@ -66,6 +82,43 @@ func NewSessionStore(defaultTTL time.Duration) *SessionStore {
 	}
 }
 
+// NewPersistentSessionStore — same store, but writes also go to the DB.
+// Call Load once after this to bring back sessions from a previous run.
+func NewPersistentSessionStore(defaultTTL time.Duration, p SessionPersister) *SessionStore {
+	s := NewSessionStore(defaultTTL)
+	s.persist = p
+	return s
+}
+
+// Load — read the saved sessions back into memory. Anything past its TTL is
+// dropped (and deleted from the DB by the store), never revived.
+func (s *SessionStore) Load(ctx context.Context, now time.Time) error {
+	if s.persist == nil {
+		return nil
+	}
+	rows, err := s.persist.LoadDialogueSessions(ctx, now)
+	if err != nil {
+		return err
+	}
+	s.mu.Lock()
+	defer s.mu.Unlock()
+	for _, r := range rows {
+		var sess Session
+		if err := json.Unmarshal(r.Data, &sess); err != nil {
+			// A blob we can't read is not worth failing a startup over.
+			continue
+		}
+		if sess.TTL <= 0 {
+			sess.TTL = r.TTL
+		}
+		if sess.IsExpired(now) {
+			continue
+		}
+		s.sessions[r.ID] = &sess
+	}
+	return nil
+}
+
 func (s *SessionStore) Get(id string, now time.Time) *Session {
 	s.mu.RLock()
 	sess, ok := s.sessions[id]
@@ -87,12 +140,33 @@ func (s *SessionStore) Put(id string, sess *Session) {
 	s.mu.Lock()
 	s.sessions[id] = sess
 	s.mu.Unlock()
+	s.save(id, sess)
 }
 
 func (s *SessionStore) Delete(id string) {
 	s.mu.Lock()
 	delete(s.sessions, id)
 	s.mu.Unlock()
+	if s.persist != nil {
+		_ = s.persist.DeleteDialogueSession(context.Background(), id)
+	}
+}
+
+// save — mirror one session to the DB. Best effort: memory already has it, so
+// a write error costs us the restart safety net, not the current turn.
+func (s *SessionStore) save(id string, sess *Session) {
+	if s.persist == nil {
+		return
+	}
+	data, err := json.Marshal(sess)
+	if err != nil {
+		return
+	}
+	ts := sess.Timestamp
+	if ts.IsZero() {
+		ts = time.Now()
+	}
+	_ = s.persist.SaveDialogueSession(context.Background(), id, data, ts, sess.TTL)
 }
 
 func InheritSlots(prev, cur Slots) Slots {

internal/dialogue/session_persist_test.go

@@ -0,0 +1,118 @@
+package dialogue
+
+import (
+	"context"
+	"path/filepath"
+	"testing"
+	"time"
+
+	"github.com/kami/maven/internal/store"
+)
+
+// openStore — a store on disk, so a second handle can reopen the same file.
+func openStore(t *testing.T, path string) *store.Store {
+	t.Helper()
+	s, err := store.Open(context.Background(), path)
+	if err != nil {
+		t.Fatalf("open store: %v", err)
+	}
+	t.Cleanup(func() { _ = s.Close() })
+	return s
+}
+
+// A session written before a restart comes back and still merges a follow-up.
+func TestSessionSurvivesRestart(t *testing.T) {
+	ctx := context.Background()
+	path := filepath.Join(t.TempDir(), "maven_test.db")
+	now := time.Now().UTC().Truncate(time.Millisecond)
+
+	first := openStore(t, path)
+	before := NewPersistentSessionStore(2*time.Minute, first)
+	before.Put("voice", &Session{
+		Intent:    IntentReminder,
+		Slots:     Slots{Text: "полить цветы", Time: now.Add(time.Hour), HasTime: true},
+		Timestamp: now,
+		TTL:       2 * time.Minute,
+	})
+	if err := first.Close(); err != nil {
+		t.Fatalf("close: %v", err)
+	}
+
+	// fresh handle, fresh in-memory map — as after a daemon restart
+	after := openStore(t, path)
+	reloaded := NewPersistentSessionStore(2*time.Minute, after)
+	if err := reloaded.Load(ctx, now.Add(10*time.Second)); err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	sess := reloaded.Get("voice", now.Add(10*time.Second))
+	if sess == nil {
+		t.Fatal("session did not survive the restart")
+	}
+	if sess.Intent != IntentReminder {
+		t.Fatalf("intent = %q, want reminder", sess.Intent)
+	}
+	// the follow-up carries no text of its own; it must inherit the old one
+	merged := InheritSlots(sess.Slots, Slots{Time: now.Add(2 * time.Hour), HasTime: true})
+	if merged.Text != "полить цветы" {
+		t.Fatalf("merged text = %q, want the earlier turn's text", merged.Text)
+	}
+	if !merged.Time.Equal(now.Add(2 * time.Hour)) {
+		t.Fatalf("merged time = %v, want the follow-up's time", merged.Time)
+	}
+}
+
+// A session past its TTL is dead: a restart must not bring it back.
+func TestExpiredSessionNotResurrected(t *testing.T) {
+	ctx := context.Background()
+	path := filepath.Join(t.TempDir(), "maven_test.db")
+	now := time.Now().UTC().Truncate(time.Millisecond)
+
+	first := openStore(t, path)
+	before := NewPersistentSessionStore(2*time.Minute, first)
+	before.Put("voice", &Session{
+		Intent:    IntentReminder,
+		Slots:     Slots{Text: "полить цветы"},
+		Timestamp: now,
+		TTL:       time.Minute,
+	})
+	if err := first.Close(); err != nil {
+		t.Fatalf("close: %v", err)
+	}
+
+	after := openStore(t, path)
+	reloaded := NewPersistentSessionStore(2*time.Minute, after)
+	later := now.Add(5 * time.Minute) // well past the 1-min TTL
+	if err := reloaded.Load(ctx, later); err != nil {
+		t.Fatalf("Load: %v", err)
+	}
+	if sess := reloaded.Get("voice", later); sess != nil {
+		t.Fatalf("expired session came back: %+v", sess)
+	}
+	// and it is gone from the DB too, not just from memory
+	rows, err := after.LoadDialogueSessions(ctx, later)
+	if err != nil {
+		t.Fatalf("LoadDialogueSessions: %v", err)
+	}
+	if len(rows) != 0 {
+		t.Fatalf("expired row still in the DB: %+v", rows)
+	}
+}
+
+// Delete removes the row as well, so an ended conversation stays ended.
+func TestDeleteRemovesPersistedSession(t *testing.T) {
+	ctx := context.Background()
+	path := filepath.Join(t.TempDir(), "maven_test.db")
+	now := time.Now().UTC().Truncate(time.Millisecond)
+
+	s := openStore(t, path)
+	ss := NewPersistentSessionStore(2*time.Minute, s)
+	ss.Put("voice", &Session{Intent: IntentChat, Timestamp: now, TTL: time.Minute})
+	ss.Delete("voice")
+	rows, err := s.LoadDialogueSessions(ctx, now)
+	if err != nil {
+		t.Fatalf("LoadDialogueSessions: %v", err)
+	}
+	if len(rows) != 0 {
+		t.Fatalf("row survived Delete: %+v", rows)
+	}
+}

internal/email/extract.go

@@ -0,0 +1,226 @@
+package email
+
+import (
+	"context"
+	"encoding/json"
+	"fmt"
+	"strings"
+	"time"
+
+	"github.com/kami/maven/internal/llm"
+	"github.com/kami/maven/internal/persona"
+)
+
+// Extraction — turning one mail into task CANDIDATES, and nothing else.
+//
+// The output of this file can only ever become rows in `tasks` with status
+// "candidate" (store.TaskCandidate), written through the one intake seam
+// (ipc.CaptureTaskReq, Vikunja #130). That bound is the whole design:
+//
+//   - No reminder. A reminder FIRES; it speaks to him unprompted. A 1.7B that
+//     misreads "встреча была в четверг" as a future appointment would then wake
+//     him up about it. A candidate that is wrong is a line on a review page he
+//     dismisses in one click, which is the correct cost of a model being wrong
+//     about someone's mail.
+//   - No fact. A fact is a claim Maven will later recite as true. Nothing read
+//     out of a marketing mail deserves that standing.
+//   - No calendar event, no note, no action. Extraction writes candidates or
+//     writes nothing.
+//
+// The due date the model may return is stored on the candidate (tasks.due_ts),
+// which no scheduler reads — it is there so the review page can sort by it.
+//
+// Privacy: the mail text goes to the resident model on this box and nowhere
+// else. It is never search input (CLAUDE.md: "his notes and facts are never
+// search input" — mail is the same class), and Evidence keeps only the subject
+// line, so the review page shows him where a candidate came from without the
+// store growing a copy of his mailbox.
+
+// MaxCandidates — at most this many candidates per message, enforced by the
+// grammar. A mail with four tasks in it is a mail he has to read himself; a
+// model allowed ten will produce ten.
+const MaxCandidates = 3
+
+// SourcePrefix — provenance for everything this package captures. The mailbox
+// name is appended: "email:INBOX". Same vocabulary as tap:voice / poll:netdata.
+const SourcePrefix = "email:"
+
+// Candidate — one piece of work the model thinks the mail is asking for.
+type Candidate struct {
+	Text string `json:"text"`
+	// Due — "YYYY-MM-DD" or empty. A date the model read out of the text, not a
+	// date it computed: relative wording ("до пятницы") is left in Text, because
+	// a small model resolving "пятница" against today's date gets it wrong often
+	// enough that a stored wrong date is worse than no date.
+	Due string `json:"due"`
+}
+
+// Completer — the llama-server seam, same shape memeval and the router use, so
+// the one resident model serves this caller too.
+type Completer interface {
+	Complete(ctx context.Context, r llm.Req) (string, error)
+}
+
+// Extractor reads a message and returns candidates. It holds no store and no
+// writer on purpose: this type cannot persist anything, so "extraction never
+// acts" is a property of the code, not of a review.
+type Extractor struct {
+	llm Completer
+	// MaxCandidates — 0 ⇒ MaxCandidates.
+	max int
+	// ContextBlock — the shared persona block, optional. Extraction output is
+	// not spoken, so the persona matters less here than in the phraser; it is
+	// wired anyway so a candidate reads in her voice on the review page.
+	contextBlock func() string
+}
+
+func NewExtractor(c Completer, max int, contextBlock func() string) *Extractor {
+	if max <= 0 || max > MaxCandidates {
+		max = MaxCandidates
+	}
+	return &Extractor{llm: c, max: max, contextBlock: contextBlock}
+}
+
+// extractGrammar — GBNF pinning the answer to a bounded array of fixed-shape
+// candidates. Same reasoning as memeval's evalGrammar and the router's
+// routeGrammar: the shape and the length bound are what keep a small model from
+// drifting into prose or spending the token budget repeating one field.
+//
+// The empty array is reachable, deliberately: most mail contains no task, and a
+// model with no way to say "nothing" invents something.
+const extractGrammar = `
+root  ::= "[" ws (item ("," ws item){0,2})? ws "]"
+item  ::= "{" ws "\"text\"" ws ":" ws text "," ws "\"due\"" ws ":" ws due ws "}"
+text  ::= "\"" ([^"\\] | "\\" .){1,120} "\""
+due   ::= "\"\"" | "\"" [0-9]{4} "-" [0-9]{2} "-" [0-9]{2} "\""
+ws    ::= [ \t\n]*
+`
+
+// extractSystem — the extraction prompt.
+//
+// Written around the two failure modes a small model has on this task: it
+// summarises when asked to extract (turning a mail into "письмо от Антона"),
+// and it invents an obligation from any polite closing sentence. Hence the
+// insistence on a verb phrase, and the explicit permission to return [].
+const extractSystem = `Ты читаешь одно письмо из его почты и достаёшь из него дела, которые письмо от него требует.
+
+Правила:
+- Отвечай ТОЛЬКО массивом JSON. Каждый элемент: {"text": "...", "due": "ГГГГ-ММ-ДД" или ""}.
+- text — короткая формулировка дела по-русски, с глаголом: "оплатить счёт за интернет", "отправить акт". Не пересказывай письмо и не описывай его.
+- Дело — это то, что должен сделать ОН. Рассылка, реклама, уведомление, отчёт, письмо «просто к сведению» — дел не содержат.
+- Если письмо ничего от него не требует, верни пустой массив []. Это нормальный ответ, так бывает чаще всего.
+- Ничего не придумывай. Если срока в письме нет — "".
+- due заполняй только когда в письме стоит конкретная дата. Слова вроде «до пятницы» оставь в text, дату не вычисляй.
+- Максимум три дела. Лучше одно точное, чем три общих.`
+
+// Extract returns the candidates in one message.
+//
+// Junk is refused without an LLM call — cheapest possible defence, and the
+// reason the header filter exists. An empty message (no subject, no body) is
+// likewise not worth a round trip.
+//
+// A parse failure is an error the caller logs and moves past. It is never
+// silently turned into zero candidates, because "the model went off the rails"
+// and "the mail contains no task" want different reactions from a human reading
+// the log.
+func (e *Extractor) Extract(ctx context.Context, msg Message) ([]Candidate, error) {
+	if msg.Junk {
+		return nil, nil
+	}
+	user := renderForModel(msg)
+	if user == "" {
+		return nil, nil
+	}
+	raw, err := e.llm.Complete(ctx, llm.Req{
+		System:        persona.Prepend(e.contextBlock, extractSystem),
+		User:          user,
+		Grammar:       extractGrammar,
+		MaxTokens:     512,
+		RepeatPenalty: 1.1,
+	})
+	if err != nil {
+		return nil, fmt.Errorf("email: extract: %w", err)
+	}
+	items, err := parseCandidates(raw)
+	if err != nil {
+		// The raw reply is NOT in the error: it is a transformation of his mail,
+		// and this error reaches the daemon log.
+		return nil, fmt.Errorf("email: extract: unparsable reply (%d bytes)", len(raw))
+	}
+	out := make([]Candidate, 0, len(items))
+	seen := map[string]bool{}
+	for _, it := range items {
+		it.Text = strings.TrimSpace(it.Text)
+		if it.Text == "" {
+			continue
+		}
+		key := strings.ToLower(strings.Join(strings.Fields(it.Text), " "))
+		if seen[key] {
+			continue // the model repeating itself is not two tasks
+		}
+		seen[key] = true
+		if _, ok := ParseDue(it.Due); !ok {
+			it.Due = "" // a date the grammar allowed but the calendar does not
+		}
+		out = append(out, it)
+		if len(out) >= e.max {
+			break
+		}
+	}
+	return out, nil
+}
+
+// renderForModel is the user turn: subject, sender and body, labelled. Only
+// these three fields — no headers, no recipient list, no message-id, nothing
+// that would let the model start reasoning about routing metadata.
+func renderForModel(msg Message) string {
+	var b strings.Builder
+	if msg.From != "" {
+		fmt.Fprintf(&b, "От: %s\n", msg.From)
+	}
+	if msg.Subject != "" {
+		fmt.Fprintf(&b, "Тема: %s\n", msg.Subject)
+	}
+	if msg.Body != "" {
+		fmt.Fprintf(&b, "\n%s\n", msg.Body)
+	}
+	if msg.Subject == "" && msg.Body == "" {
+		return ""
+	}
+	return b.String()
+}
+
+// parseCandidates decodes the grammar-constrained reply, tolerating the
+// wrappers a Thinking model sometimes leaves around it (a fenced block, or
+// leading reasoning before the array).
+func parseCandidates(raw string) ([]Candidate, error) {
+	s := strings.TrimSpace(raw)
+	if i := strings.Index(s, "["); i > 0 {
+		s = s[i:]
+	}
+	if j := strings.LastIndex(s, "]"); j >= 0 {
+		s = s[:j+1]
+	}
+	var out []Candidate
+	if err := json.Unmarshal([]byte(s), &out); err != nil {
+		return nil, err
+	}
+	return out, nil
+}
+
+// ParseDue turns the model's "YYYY-MM-DD" into a time in UTC. Exported because
+// the daemon-side intake stores it on the candidate.
+//
+// The zero-value/empty case returns ok=false rather than an error: no date is
+// the common answer, not a failure.
+func ParseDue(s string) (time.Time, bool) {
+	s = strings.TrimSpace(s)
+	if s == "" {
+		return time.Time{}, false
+	}
+	t, err := time.Parse("2006-01-02", s)
+	if err != nil {
+		return time.Time{}, false
+	}
+	return t, true
+}

internal/email/extract_test.go

@@ -0,0 +1,143 @@
+package email
+
+import (
+	"context"
+	"strings"
+	"testing"
+
+	"github.com/kami/maven/internal/llm"
+)
+
+// fakeLLM returns a canned reply and records the request, so a test can assert
+// on the grammar and on what of the mail was sent.
+type fakeLLM struct {
+	reply string
+	err   error
+	got   llm.Req
+	calls int
+}
+
+func (f *fakeLLM) Complete(_ context.Context, r llm.Req) (string, error) {
+	f.calls++
+	f.got = r
+	return f.reply, f.err
+}
+
+func msgFor(subject, body string) Message {
+	return Message{UID: 1, From: "anton@example.org", Subject: subject, Body: body}
+}
+
+func TestExtractCandidates(t *testing.T) {
+	f := &fakeLLM{reply: `[{"text":"отправить акт","due":""},{"text":"оплатить счёт","due":"2026-08-05"}]`}
+	e := NewExtractor(f, 0, nil)
+	got, err := e.Extract(context.Background(), msgFor("Акт и счёт", "Надо отправить акт и оплатить счёт до 5 августа."))
+	if err != nil {
+		t.Fatalf("extract: %v", err)
+	}
+	if len(got) != 2 {
+		t.Fatalf("got %d candidates, want 2: %+v", len(got), got)
+	}
+	if got[0].Text != "отправить акт" || got[1].Due != "2026-08-05" {
+		t.Errorf("candidates = %+v", got)
+	}
+	if f.got.Grammar == "" {
+		t.Error("extraction must be grammar-constrained")
+	}
+	// The subject and body go to the model; nothing else about the message does.
+	if !strings.Contains(f.got.User, "Акт и счёт") || !strings.Contains(f.got.User, "оплатить счёт") {
+		t.Errorf("user turn = %q", f.got.User)
+	}
+}
+
+func TestExtractEmptyArrayIsNotAnError(t *testing.T) {
+	f := &fakeLLM{reply: "[]"}
+	got, err := NewExtractor(f, 0, nil).Extract(context.Background(), msgFor("FYI", "Просто к сведению."))
+	if err != nil || len(got) != 0 {
+		t.Fatalf("got (%v, %v), want (empty, nil) — no task is the normal answer", got, err)
+	}
+}
+
+// Junk must never reach the model: the header filter exists so the resident
+// model is not spent on newsletters.
+func TestExtractSkipsJunkWithoutCallingModel(t *testing.T) {
+	f := &fakeLLM{reply: `[{"text":"купить всё со скидкой","due":""}]`}
+	msg := msgFor("Скидки", "Sale!")
+	msg.Junk = true
+	got, err := NewExtractor(f, 0, nil).Extract(context.Background(), msg)
+	if err != nil || got != nil {
+		t.Fatalf("got (%v, %v), want (nil, nil)", got, err)
+	}
+	if f.calls != 0 {
+		t.Errorf("model called %d times for junk, want 0", f.calls)
+	}
+}
+
+func TestExtractEmptyMessageIsNotSent(t *testing.T) {
+	f := &fakeLLM{reply: "[]"}
+	if _, err := NewExtractor(f, 0, nil).Extract(context.Background(), Message{UID: 3}); err != nil {
+		t.Fatalf("extract: %v", err)
+	}
+	if f.calls != 0 {
+		t.Errorf("model called %d times for an empty message, want 0", f.calls)
+	}
+}
+
+func TestExtractCaps(t *testing.T) {
+	f := &fakeLLM{reply: `[{"text":"a","due":""},{"text":"b","due":""},{"text":"c","due":""}]`}
+	got, err := NewExtractor(f, 2, nil).Extract(context.Background(), msgFor("s", "b"))
+	if err != nil {
+		t.Fatalf("extract: %v", err)
+	}
+	if len(got) != 2 {
+		t.Errorf("got %d, want the configured cap of 2", len(got))
+	}
+}
+
+func TestExtractDropsRepeatsAndBadDates(t *testing.T) {
+	f := &fakeLLM{reply: `[{"text":"Отправить акт","due":"2026-02-31"},{"text":"отправить  акт","due":""},{"text":"  ","due":""}]`}
+	got, err := NewExtractor(f, 0, nil).Extract(context.Background(), msgFor("s", "b"))
+	if err != nil {
+		t.Fatalf("extract: %v", err)
+	}
+	if len(got) != 1 {
+		t.Fatalf("got %d candidates, want 1 (repeat and blank dropped): %+v", len(got), got)
+	}
+	if got[0].Due != "" {
+		t.Errorf("due = %q, want empty — 2026-02-31 is not a date", got[0].Due)
+	}
+}
+
+// A Thinking model sometimes wraps the array; and when it emits something
+// unparsable the caller must hear about it rather than see "no tasks".
+func TestParseCandidatesTolerance(t *testing.T) {
+	got, err := parseCandidates("думаю... [{\"text\":\"x\",\"due\":\"\"}] всё")
+	if err != nil || len(got) != 1 || got[0].Text != "x" {
+		t.Fatalf("got (%+v, %v)", got, err)
+	}
+	if _, err := parseCandidates("нет никакого JSON"); err == nil {
+		t.Error("unparsable output must be an error")
+	}
+}
+
+func TestExtractParseErrorHidesMailText(t *testing.T) {
+	f := &fakeLLM{reply: "он просил отправить акт, вот такой ответ"}
+	_, err := NewExtractor(f, 0, nil).Extract(context.Background(), msgFor("Акт", "секретный текст"))
+	if err == nil {
+		t.Fatal("want an error")
+	}
+	if strings.Contains(err.Error(), "акт") || strings.Contains(err.Error(), "секретный") {
+		t.Errorf("error text leaks mail content: %v", err)
+	}
+}
+
+func TestParseDue(t *testing.T) {
+	if _, ok := ParseDue(""); ok {
+		t.Error("empty due must be (zero, false)")
+	}
+	if got, ok := ParseDue("2026-08-05"); !ok || got.Year() != 2026 || got.Month() != 8 || got.Day() != 5 {
+		t.Errorf("ParseDue = (%v, %v)", got, ok)
+	}
+	if _, ok := ParseDue("05.08.2026"); ok {
+		t.Error("a non-ISO date must not parse")
+	}
+}

internal/email/fetch.go

@@ -0,0 +1,98 @@
+package email
+
+import (
+	"fmt"
+	"time"
+)
+
+// FetchSince is the whole read path in one call: connect, log in, examine the
+// mailbox read-only, list what arrived since a date, fetch and parse the ones
+// the caller has not seen, log out.
+//
+// It is a function rather than a long-lived object because a mail poller should
+// not hold an authenticated session (and therefore his credential in a live TLS
+// state) between polls. Connect, read, drop.
+//
+// skip decides which UIDs are already known — the poller's seen-set. max bounds
+// one poll: a mailbox that received 400 messages overnight must not turn into
+// 400 LLM calls, and the newest max are the ones a task could still be hiding
+// in. Junk messages are returned too, flagged, so the caller can mark them seen
+// without a second protocol round.
+type FetchSince struct {
+	Addr    string // host or host:993
+	User    string
+	Mailbox string // e.g. "INBOX"
+	Timeout time.Duration
+	Since   time.Time
+	Max     int
+	Skip    func(uid uint32) bool
+}
+
+// Run performs one read. password is passed here, not stored in the struct, so
+// the configuration of a mailbox and the secret for it are never the same value
+// sitting in the same place.
+func (f FetchSince) Run(password string) ([]Message, error) {
+	return f.RunWith(password, nil)
+}
+
+// RunWith is Run with an explicit connection function, which is how the reader
+// daemon and the tests substitute an in-process server. nil ⇒ Dial, i.e.
+// implicit TLS with certificate verification; there is no configuration path
+// that reaches this, so no deployment can end up talking cleartext IMAP.
+func (f FetchSince) RunWith(password string, dial func(addr string, timeout time.Duration) (*Conn, error)) ([]Message, error) {
+	if f.Addr == "" || f.User == "" || f.Mailbox == "" {
+		return nil, fmt.Errorf("email: mailbox not configured (addr/user/mailbox)")
+	}
+	if dial == nil {
+		dial = Dial
+	}
+	c, err := dial(f.Addr, f.Timeout)
+	if err != nil {
+		return nil, err
+	}
+	defer c.Close()
+	if err := c.Login(f.User, password); err != nil {
+		return nil, err
+	}
+	defer c.Logout()
+	if err := c.Select(f.Mailbox); err != nil {
+		return nil, err
+	}
+	uids, err := c.SearchSince(f.Since)
+	if err != nil {
+		return nil, err
+	}
+
+	// Newest UIDs first — IMAP hands them back ascending, and when Max clips the
+	// list the recent mail is what matters.
+	wanted := make([]uint32, 0, len(uids))
+	for i := len(uids) - 1; i >= 0; i-- {
+		if f.Skip != nil && f.Skip(uids[i]) {
+			continue
+		}
+		wanted = append(wanted, uids[i])
+		if f.Max > 0 && len(wanted) >= f.Max {
+			break
+		}
+	}
+
+	out := make([]Message, 0, len(wanted))
+	for _, uid := range wanted {
+		raw, err := c.Fetch(uid)
+		if err != nil {
+			// One unreadable message does not abandon the poll; the rest of the
+			// mailbox is still worth reading. The error names the UID, not the
+			// message.
+			return out, fmt.Errorf("email: fetch uid %d: %w", uid, err)
+		}
+		if len(raw) == 0 {
+			continue // vanished between SEARCH and FETCH
+		}
+		msg, err := ParseMessage(uid, raw)
+		if err != nil {
+			continue // unparsable headers — nothing to review, skip silently
+		}
+		out = append(out, msg)
+	}
+	return out, nil
+}

internal/email/fetch_test.go

@@ -0,0 +1,49 @@
+package email
+
+import (
+	"net"
+	"strings"
+	"testing"
+	"time"
+)
+
+func TestFetchSinceRun(t *testing.T) {
+	mk := func(subject string) string {
+		return "Subject: " + subject + "\r\nContent-Type: text/plain; charset=utf-8\r\n\r\nbody\r\n"
+	}
+	f := &fakeIMAP{
+		uids: []uint32{1, 2, 3},
+		msgs: map[uint32]string{1: mk("one"), 2: mk("two"), 3: mk("three")},
+	}
+	fs := FetchSince{
+		Addr: "mail.example:993", User: "kami", Mailbox: "INBOX",
+		Timeout: 5 * time.Second,
+		Since:   time.Date(2026, 7, 30, 0, 0, 0, 0, time.UTC),
+		Max:     2,
+		Skip:    func(uid uint32) bool { return uid == 3 },
+	}
+	msgs, err := fs.RunWith("secret", func(addr string, timeout time.Duration) (*Conn, error) {
+		cli, srv := net.Pipe()
+		go f.serve(t, srv)
+		return NewConn(cli, timeout)
+	})
+	if err != nil {
+		t.Fatalf("run: %v", err)
+	}
+	// Newest first, the already-seen UID skipped, Max respected.
+	if len(msgs) != 2 {
+		t.Fatalf("got %d messages, want 2: %+v", len(msgs), msgs)
+	}
+	if msgs[0].Subject != "two" || msgs[1].Subject != "one" {
+		t.Errorf("subjects = %q,%q, want two,one (newest first)", msgs[0].Subject, msgs[1].Subject)
+	}
+	if strings.Contains(strings.Join(f.cmds, " "), "UID FETCH 3") {
+		t.Error("a skipped UID must not be fetched again")
+	}
+}
+
+func TestFetchSinceRequiresConfig(t *testing.T) {
+	if _, err := (FetchSince{}).Run("secret"); err == nil {
+		t.Fatal("an unconfigured mailbox must not be read")
+	}
+}