Add docs/schemas/execution_plan.json (goal/stages/edges), the
architect_freestyle prompt that instructs emitting the execution pipeline as
JSON, and register the execution_plan llm-emitted kind in both
artifacts.config.toml and sample-config.toml. Validated via
ExecutionPlanSchemaValidationTest (minimal plan passes; missing stages rejected).
Extract a SubagentRunner fun-interface (SubagentRunRequest/SubagentRunResult)
and an InSessionSubagentRunner default that delegates to the existing
executeStage path, with per-run effectives captured in the lambda closure so
the seam stays free of the internal RunEffectives type.
DefaultSessionOrchestrator.enterStage now dispatches through subagentRunner.run
instead of calling executeStage inline; effectives threading is dropped from
run/step/executeMove/enterStage/resume. ReplayOrchestrator supplies its own
runner. Behaviour is identical — RefinementLoopTest and full check stay green.
Stages declaring needs=[...] now receive each needed artifact's validated
JSON (from artifactContentCache) as an L1/USER neededArtifact context entry,
so a one-shot subagent sees its inputs.
Also emit llm-emitted artifact lifecycle events (Created/Validating/Validated)
on successful validation via emitLlmArtifacts, closing a latent gap where
verifyProduces would fail for stages producing only an llm-emitted artifact.
Events are emitted only after validation passes, preserving invariant #7.
The router/narration model behaviour was hardcoded (Main built RouterConfig()
with defaults). Surface it under the [router] section so it's tunable without a
rebuild:
[router] conversation_keep_last, retrieval_k, token_budget
[router.generation] temperature, top_p, max_tokens (chat/steering)
[router.narration] temperature, top_p, max_tokens, max_per_run
ConfigLoader parses the new sections (adds asDouble); Main maps the config-layer
RouterConfig onto the domain RouterConfig and threads narration.max_per_run into
ServerModule. All values default to the previous constants, so behaviour is
unchanged when the sections are absent. Documents the block in sample-config.toml
and adds parser tests for present/absent cases.
Live QA showed two problems with paused-approval narration:
- The narrator was told to "name the stage, outcome, and reason" but the pause
trigger only carried "APPROVAL_PENDING" with no detail — on a first-stage pause
there's no L2 history either, so it had nothing to ground on. NarrationSubscriber
now captures the latest ApprovalRequestedEvent per session (tool name, tier,
preview) and folds it into the pause instruction (preview truncated to 800
chars).
- Narration reused the 512-token chat generation config; a reasoning model spent
the whole budget "thinking" and returned empty content (finishReason=length).
Add a separate narrationGenerationConfig (maxTokens=1024) so thinking can
complete and still leave room for the one or two sentences.
Adds a subscriber test asserting the pause narration names the tool and includes
the preview.
Replace the placeholder "You are a routing assistant" one-liner with two
purpose-written prompts: a conversational prompt that frames the router as
correx's operator-facing interface to the event-sourced engine (grounded in
workflow state, no fabrication, concise, steering-aware), and a dedicated
narrator prompt for buildNarrationContext that asks for one or two present-tense
sentences naming the stage and outcome.
The longer protected frame shifted token budgets, so the two L2 eviction tests
now size their budget from the measured protected-frame + per-entry cost instead
of hard-coded magic numbers, making them robust to prompt length.
Two issues surfaced during live QA once workflows began progressing past the
first stage:
1. Narration prompts went out with no user message, so the chat template
rejected them ("No user query found in messages"). buildNarrationContext
filed the trigger instruction (a user turn) under ContextLayer.L0, and
PromptRenderer folds every L0 entry into the system message regardless of
role. Move the trigger to L1 so it renders as the user turn.
2. TransitionExecutedEvent was emitted after the next stage had already run:
executeMove called enterStage (which executes the stage) before advanceStage
(which emits the transition), so the event log showed a stage running before
the event marking entry into it. Emit the transition before entering.
Adds a regression test asserting a rendered narration prompt carries a user
message.
Tool-calling loops were rendered as all-assistant-calls-then-all-tool-results
with the original task last, causing the model to re-issue the same tool call
indefinitely. Two stacked reorderings caused it:
- DefaultContextPackBuilder grouped entries by sourceType for per-type
compression, discarding a,t,a,t interleaving.
- PromptRenderer forced L1 (the live user turn) to render last, pushing the
task after the entire transcript.
- SessionOrchestrator's tool loop re-fed currentContext.layers.values.flatten()
(grouped by layer) each round, compounding the scramble.
Add a chronological `ordinal` to ContextEntry, stamped by the builder from
input order and restored after grouping/compression (the compressor preserves
entry identity, so ordinals survive). PromptRenderer now orders non-system
messages by ordinal, with the old L1-last layer priority kept only as a
tiebreak so router chat (ordinal 0) is unchanged. The orchestrator keeps a
running ordered accumulator instead of reading back the grouped pack.
Adds builder + renderer ordering regression tests.
Add TypeRouterNarration constant, mark it event-bearing, decode golden test,
applyServer case appending RouterEntry{Role:"narration_llm"} with metrics, and
routerRows render branch showing bright ◆ prefix with FgStrong content + metrics suffix.
- Add NarrationTrigger(kind, instruction) model to core:router
- Add RouterContextBuilder.buildNarrationContext: minimal ContextPack
(system + workflow status + L2 + trigger instruction); excludes
conversationHistory and L3 recalled memory
- Add RouterFacade.narrate: emits RouterNarrationEvent with content,
latencyMs, tokensUsed; skips event on blank inference; never writes
ChatTurnEvent or l3MemoryStore.store
- Add RouterNarrationTest covering all acceptance criteria
- Add narrate default to RouterContextBuilder interface so existing
anonymous test stubs compile without a stub override
ChatTurn wire frame now includes latencyMs: Long? and totalTokens: Int?.
DomainEventMapper reads them from ChatTurnEvent.latencyMs and
tokensUsed?.totalTokens. Positional call site in DomainEventMapperTest
converted to named args. Two new serialization tests cover the ROUTER
(metrics present) and USER (metrics absent) paths.
ChatTurnEvent gains nullable latencyMs and tokensUsed fields (defaults preserve
backward-compat; legacy JSON without them deserializes cleanly). emitChatTurn
accepts optional metrics; the ROUTER emit site passes inferenceResponse values
through; the USER emit site leaves both null.
Stage started/completed/failed events are now injected into the router
transcript (role "stage") and rendered in Faint colour, giving the
operator a chronological in-feed view of workflow progress alongside
user/router/tool turns.
Wires the workspace handshake end to end so a session's workspace is bound
from the client's cwd at connect time and is event-sourced for replay.
- Go TUI sends Hello{workingDir=os.Getwd()} as the first WS frame on connect
(protocol.go encoder + client.go connect path; golden test pins the wire
format against the Kotlin discriminator).
- Server adds ClientMessage.Hello, stashes the per-connection workingDir, and
on session start resolves it through WorkspaceResolver's trust pipeline,
emits SessionWorkspaceBoundEvent (invariant #9), and threads the resolved
workspace into OrchestrationConfig for the live run. A Hello after the first
StartSession is ignored (warn); a rejected path binds the resolver fallback.
- Replay derives the workspace from the recorded event: SessionState gains
boundWorkspace, DefaultSessionReducer fills it from SessionWorkspaceBoundEvent,
and ReplayOrchestrator uses it (Path.of only, no filesystem re-query —
invariant #8) with graceful fallback to config for pre-Phase-B logs.
Absent Hello / null resolver degrades to the prior config-workspace behavior.
A stage's allowedTools only shaped which tool definitions were offered to
the model; the dispatch path resolved returned tool calls straight from the
full registry and executed them without checking stage membership. A
hallucinated, jailbroken, or edited-transcript tool call for any registered
tool would run unchecked (violating invariant #5 / policy-is-absolute).
dispatchToolCalls now rejects any tool call whose name is not in the stage's
allowedTools (STAGE_COMPLETE_TOOL exempt), emitting ToolExecutionRejectedEvent
and feeding an error ContextEntry back to the model instead of executing.
Empty allowedTools means deny-all domain tools, consistent with offering
only STAGE_COMPLETE_TOOL at inference time.
SessionWorkspaceBoundEvent already recorded the bound workspace root but
it was never surfaced. Map it to a new session.workspace_bound frame
(sealed ServerMessage variant, auto-registered; re-emitted on reconnect
via the snapshot replay), carry it on Session.WorkspaceRoot, and show it
home-abbreviated in the status bar (⌂ ~/Programs/correx). Golden test
pins the decode.
The resource gauge was NVIDIA-only — on an AMD/ROCm box Main fell back to
UnavailableProbe, so the TUI showed no VRAM. Add AmdResourceProbe backed
by `rocm-smi --showmeminfo vram --showuse --csv`: it skips the warning
preamble, locates the header by column name (order/version tolerant), and
converts the byte VRAM figures to MiB. Process RSS reuses /proc/<pid>/status
(only populated for a managed-model pid). Main now selects NVIDIA → AMD →
Unavailable. Tests cover the real ROCm 4.0 CSV + reordered columns.
The steer note replaced the action row while editing and vanished once
you esc'd out — so you couldn't see the note or the approve/reject keys,
and it was unclear the note still applied (it did: decide() reads the
buffer). Now the note is a persistent line above the action row (editable
with a caret while typing, a dim reminder once set), and the action row
stays visible. Editing hints clarify enter=approve+send, esc=keep note.
A steering note attached to a tool approval was recorded but only
consumed by the NEXT stage's prompt build (buildSteeringNoteEntries runs
once, before the tool loop), so on a single-stage task it had no visible
effect. Two changes:
- Approve+note: capture userDecision.userSteering at the gate and inject
it as a context entry right after the tool result, so the same stage's
loop re-infers with the note and the model acts on it.
- Reject: buildSteeringNoteEntries now also emits anti-repeat feedback for
REJECTED decisions with no note, so the model does not re-propose the
identical call on the retryable re-run (a bare reject previously fed
back only "approval denied").
Integration test drives an approval with a steering note and asserts the
next inference's context carries it.
- Events were hard-capped to the last 7 per session in addEvent, so the
e-inspector and EVENTS panel could never show more. Raise to 1000
(effectively all, bounded); the EVENTS panel now shows the latest that
fit (tail), and the inspector scrolls a window around the selection
with a position indicator.
- The TUI had no approval.resolved handling at all — the decision frame
was ignored and pending only cleared on session.resumed. Add the
constant + case: clear the gate and record an ApprovalResolved event
(APPROVED/REJECTED/AUTO_APPROVED + reason) into the stream.
- Status bar now shows the current stage and session status alongside the
name. Relabel the cryptic "N bg" badge to "N elsewhere".
Three QA-found issues in the approval gate:
- displayState gated the in-session/approval surfaces on hasApproval
before sessionEntered, so moving the list cursor onto a session with a
pending gate auto-opened it, and `l` back-to-list couldn't escape
(the gate re-popped). Require sessionEntered first.
- The band ran every preview through the two-column diff renderer, so a
shell tool's argv JSON rendered as an identical-column "diff". Detect
real unified diffs (isUnifiedDiff); render other previews as plain
text, and drop the ^x fullscreen hint when there's nothing to expand.
- The footer duplicated the band's approve/reject/steer/diff keys. It now
shows navigation (l back / e events / q quit) instead.
The plane-2 tool-call assessor records verified preconditions
("[PATH_OUTSIDE_WORKSPACE] …") and the server already ships them in
RiskSummaryDto.rationale, but the Go TUI's RiskSummaryDto had no
Rationale field — so the justification was silently dropped at decode
and the gate showed only an opaque tier.
Decode the rationale, carry it on Approval, and render it under the
header in the approval band (warn-marked, above the diff). This closes
the last deferred item of the plane-2 slice-1 plan: the assessment
surfaced to the approval UX. Golden test pins the rationale decode.
Replace the floating approval modal (which blanked the whole UI) with an
opencode-style band that takes the input bar's slot while a gate is
pending: tool/tier/risk header above a side-by-side old|new diff, with
the session output and event stream still visible above it. The band
height adapts to the diff length.
Add a unified-diff parser that aligns removals/additions into split rows
(blank left cell for a create, blank right for a delete) and a
two-column renderer shared by the band preview and the ^x fullscreen
view. Wire plain a/r to approve/reject to match the advertised keys.
The diff overlay incremented diffScrollOffset on every down-key with no
upper bound; the render clamped only the displayed offset, so the stored
value grew unbounded and the up-key appeared dead for N presses. Clamp
the increment against a new diffMaxScroll() (len(lines) - body height),
shared with the render path via diffBodyHeight().
Rebuild the Go TUI transcript from the event stream instead of
optimistic local echoes. ChatTurnEvent now maps to ServerMessage.ChatTurn
(chat.turn); the TUI auto-vivifies sessions and renders user+router turns
from events. SessionStarted is de-special-cased into SessionAnnounced
(still carries workflowId), dropping the router.response shadow and
optimistic echoes.
Delete the unused Kotlin TUI (apps/tui) and add a kotlinx<->Go
golden-fixture test to lock the wire protocol. Gitignore server runtime
logs.
Compute the effective tool registry/executor and plane-2 WorkspacePolicy per run
from config.workspace (effectivesFor), threaded through the orchestrators; a null
workspace falls back to the boot instances byte-for-byte. Wire the concrete
WorkspaceToolRegistryProvider in Main (buildToolConfigForWorkspace). Session undo
unions the session's recorded workspace into its jail roots. (Axis 2 Phase A, tasks 3 + 7.)
kami