orchestra/AUDIT.md

Orchestra — spec conformance audit & remediation plan

Audited 2026-07-27 against orchestra-spec (1).md at commit 325c684. Method: read every non-test file in internal/ and cmd/, traced each spec section to its call site, and checked whether the live path (main.go → router → coordinator → adapter) actually reaches it.

---

Verdict

The substrate (Layer 1) is genuinely built. Layers 2–4 are shaped but not wired: the packages exist, have tests, and compile — but the code paths that would make an unattended run work are either dead, mis-keyed, or built against herdr methods that appear to be invented.

Concretely: a task can be created, routed, and leased. It can never complete, and it can never rotate. After 3 lease expiries the router marks it TaskFailed. That is the whole of "connection works but features are underdeveloped."

progress.md overstates completion in several places (see §Corrections).

Spec layer	State
L1 substrate (§3, §4)	Built and correct in the main path. Some defects (below), no missing mechanism.
L2 harness (§5)	Not functional. Occupancy broken, rotation unreachable, adapter protocol unverified.
L3 continuity (§6)	Dead code. Handoff schema, pickup validation, scratch branches, TASK.md — none reached from the live path.
L4 surfaces (§7)	Partial. Brief/standup/delivery exist; quota projection has no producer; authz has a bypass.

---

Blocking defects — in dependency order

B1. Occupancy is measured from the wrong value (§5.2.1)

herdr/adapter.go:191

u, e := a.Usage(s.PaneID)     // ClaudeUsage(path string) wants a transcript file

ClaudeUsage/CodexUsage/OpenCodeUsage all take a filesystem path to session state. They are handed a herdr pane ID. Every call returns open <pane-id>: no such file, so Coordinator.rotate hits if err != nil { continue } on line 430 and never rotates anything.

The spec is unusually explicit here — §5.2.1: "Build this measurement and verify it against a live session before wiring any trigger — the whole rotation system rests on this number." That step was skipped, and the verification gap is exactly where the bug is.

Compounding: CodexActiveUsage (the discovery function that resolves the active rollout via state_*.sqlite) and ClaudeStopHookUsage and OpenCodeStatus are written but called from nowhere outside tests.

B2. The adapter lookup key is wrong in three of four call sites (§5.3, §5.4)

AdapterFactory.Herdrs is keyed by herdr instance id (homesrv-claude). Session.Harness is set by CLIAdapter.Lease to the harness kind (claude). Then:

Call site	Key used	Resolves?
refreshSessionHealth (orchestrator.go:221)	HerdrID, falls back to lease	yes
Reconcile (:291)	session.Harness	no
expire (:376, :397)	session.Harness	no
rotate (:419)	session.Harness	no

So: orphaned panes are never killed on restart, pane.exited fast-path never fires, expired leases never kill their pane, and rotation exits before it begins. All silently — every one is a bare continue.

B3. Nothing emits TaskCompleted (§4, §5.2)

Grep for producers: there is exactly one, POST /v1/tasks/{id}/complete, which a human has to call. The spec assigns this to the stop-hook/wrapper ("the plane emits events, not the agent" — Invariant 2). There is no stop hook, no wrapper, no completion detection of any kind.

Consequence: an agent that finishes its work sits idle until the 30-minute lease TTL expires → TaskReleased → re-leased → repeat → TaskFailed at attempt 3. Unattended runtime, the one metric in §0, is currently bounded at 90 minutes per task and always ends in a false failure.

B4. The router counts rotation as a retry (§5.3, §5.4)

router.go:127 increments attempts[TaskID] on every TaskReleased, and :169 increments again on every lease. :149 fails the task at MaxAttempts (3, from main.go:64).

Rotation is TaskReleased (§5.3: "rotation = intra-task lease transfer"). So a task healthy enough to rotate twice is killed by the retry limit. The spec's retry policy is for failures (§5.4), not lease transfers. These need separate counters — attempts should only advance on expiry/crash releases, never on a release carrying a valid handoff_ref.

B5. Herdr protocol methods are unverified and at least partly invented

adapter.go calls pane.release, pane.kill, pane.rotation_signal, pane.read, agent.get, agent.start, worktree.create/open, agent.prompt. The file itself documents that a previous method (pane.status) was not a valid protocol method — so this surface has a history of being written against an imagined API.

pane.rotation_signal is near-certainly not real: herdr is a generic multiplexer with no concept of Orchestra rotation. And pane.release returning a handoff_ref is architecturally wrong regardless of whether the method exists — herdr does not write handoffs; the agent does (§6.1, the handoff is a CAS artifact produced under the Face-B stop hook).

Nothing here can be settled from source. It needs herdr api schema --json run against your protocol-17 instance, diffed against every method the adapter calls.

Related: §5.1 requires agent.prompt with inline wait for bootstrap injection, because that's what closes the send-into-a-half-rendered-prompt race. CLIAdapter.Lease calls Prompt(..., wait=0) — no wait. Only Bootstrap passes one.

B6. Layer 3 is entirely dead code (§6)

Never called from anything but tests: continuity.ValidatePickup, ScratchCommit, ScratchPush, ScratchPull, ScratchSync, MarkdownChanges, Save, Load, Encode, Decode.

VerifyTaskFile is called — but only when GitWorktrees.TaskFileSHA is non-empty, and main.go:112 constructs the worktrees without ever setting it. Nothing writes a TASK.md into a worktree in the first place.

So the entire §6.2 pickup contract — the part the spec calls "the make-or-break" and "the defense against the telephone game" — does not execute. Rotation, if B1/B2 were fixed, would hand the next agent a handoff_ref that was never schema-validated, never anchor-checked, and never accompanied by an immutable spec.

B7. Quota projection has no producer (§7.2)

QuotaAvailability.sumSince folds QuotaReported events. AggregateQuota does the same for the brief. No code appends a QuotaReported event. Your config sets quota_limit_5h: 50 / quota_limit_weekly: 500 on all six herdrs; those limits are compared against a permanent 0. The availability filter is a no-op and the brief's quota_consumed is always {}.

B8. Surface: system is an unauthenticated full-control bypass (§7.1)

authz.CapabilityFor grants System → FullControl. authz.HTTP reads the surface straight from the X-Orchestra-Surface header, and the token map in main.go:760 has no entry for system — so tokens[System] is "" and the token check is skipped entirely. Any request on the LAN with X-Orchestra-Surface: system can emit any event on any task.

The surface is also absent from ParseSurface's intent: System is supposed to mean "the plane itself, in-process", but it's reachable over HTTP. It should never be accepted from a request header.

(Also: unset ORCHESTRA_*_TOKEN means that surface is unauthenticated; your .orchestra-config/orchestra.env sets none of them. Defensible on WireGuard, worth a deliberate decision rather than an accident.)

---

Secondary defects

#	Where	Issue
S1	operations.go:16,25	go vet fails: Brief.From/To both serialize as "from"; GitSync.Branch/Head/Status all as "branch". The brief's git state is unreadable by any client.
S2	main.go:238	Brief's git state is read from ORCHESTRA_DATA (the event-log dir), not from the project worktrees. §7.4 wants "what pushed, what's on which branch, what workpc still needs to pull" — per project.
S3	operations.BuildBrief	Counts completions but never surfaces report_ref/receipt, which §7.4 names as the proofs the brief exists to carry.
S4	delivery.Fanout.Run:110	A single send error returns and permanently kills the notification goroutine — silently, since main.go:729 only logs. One ntfy hiccup at 2am = no notifications for the rest of the night. Cursor is also in-memory, so a restart re-notifies the entire log from seq 0.
S5	store.Lease:341, ExpireLeases:350	Event.ID is set to the task id, so event IDs collide across every lease of a task. ApplyAdvisory and any future ID-based lookup are unsound.
S6	store.Append:191	Ingest dedup returns nil (success) without appending; main.go:175 then returns s.Events(0)[len-1] — an unrelated event — with 201 Created.
S7	store.apply:154	TaskAmended applies only title. §4 lists title, due, description, inherent_priority. Amendments to the others are accepted, logged, and silently ignored by the projection.
S8	§3.1	No compensation-event mechanism exists. Append-only holds, but the spec's correction path ("a compensating event is appended") has no implementation.
S9	main.go:482	Two independent lease-expiry loops (main.go 1s ticker and Coordinator.expire 30s) race on the same reclaim. Harmless today only because version CAS rejects the loser.
S10	federation	federation.Registry.Register accepts a self-declared id + self-chosen token from any caller — registration is admission-control-free.
S11	§5.3	Thrash detection, the soft ~55% threshold, milestone rotation, and agent-initiated ROTATE are all absent. Only the hard threshold exists, and it's unreachable (B1).

---

The architectural fork — decided 2026-07-27

There are two incompatible federation designs in the tree, and only one is deployed.

Design A — "drive the remote socket" (deployed)

clients/herdr-bridge.go runs on workpc and proxies its local herdr Unix socket to 192.168.1.105:9245. .orchestra-config/config.jsonc registers workpc-claude / workpc-codex / workpc-opencode against that machine, so the homesrv coordinator resolves a workpc herdr as a candidate and calls worktree.create, agent.start, agent.prompt on it over TCP as if it were local. workpc is a dumb pane host; all orchestration logic and all state live on homesrv, and live state crosses the machine boundary continuously.

Design B — "workers pull tasks" (/v1/federation/*, unused)

workpc would run its own Orchestra process that registers itself (POST /v1/federation/workers), heartbeats, polls the event log with a cursor (GET /v1/federation/events + /ack), claims its own lease (POST .../claim), runs a local coordinator against its local herdr socket, and reports a handoff carrying an anchor_sha it computed from its own checkout (POST .../handoff). homesrv stays authoritative for the log; nothing but git and validated artifacts crosses the wire.

This is what the spec describes — §2.1: "Everything crossing a machine boundary is git + a validated artifact — never live state over the wire." It is fully written on the server side and has zero clients. There is no worker binary in this repo; every one of those endpoints is unreachable in the current deployment.

Why keeping both is harmful, not merely untidy

Design A breaks a specific correctness property Design B was written to hold.

The anchor is validated on the wrong machine. Coordinator.rotate (orchestrator.go:460) calls herdr.HeadSHA(session.Worktree), which shells out to git -C <path> rev-parse HEAD. The coordinator runs on homesrv; session.Worktree is a path on workpc. Two outcomes, both bad:

• the path doesn't exist on homesrv → HeadSHA errors → rotation silently skips forever (another bare continue); or
• the path does exist on homesrv — likely, since every project shares the /var/lib/orchestra/worktrees/<project>/<task> layout — → it returns homesrv's HEAD for an unrelated checkout, and the emitted TaskReleased certifies a commit the agent never worked on.

The second is the dangerous one: it passes validation, looks correct, and hands the successor an anchor describing another repo's state. That is exactly the failure §9 item 8 names — "worktree/anchor validation when a lease's git checkout lives on a different host than the router" — against the degrade-safe default the spec already supplies: "validate against the local checkout wherever the harness runs." Design B satisfies this by construction; Design A cannot satisfy it without moving validation out of the coordinator.

The same class of bug applies to per-host quota accounting (§9 item 8) and to cleanupCompleted, which runs git worktree remove on homesrv for a worktree that lives on workpc.

Latent defects in the unused half

Never surfaced because nothing exercises them:

• Offline detection only runs inside Snapshot() (federation.go:103-118), called solely from GET /v1/federation/workers. Nothing polls it, so OnOffline — the hook at main.go:131 that releases leases held by a vanished worker — fires only if a human hits that endpoint. The TTL backstop §5.4 relies on does not tick on its own.
• The registry is in-memory with no persistence. Every restart forgets all workers and their cursors; a reconnecting worker re-reads the entire log from seq 0.

Decision

Keep Design A through Phase 5; commit to Design B in Phase 6. Phases 1–4 (occupancy, Face B, rotation, continuity) are single-host concerns, provable on homesrv alone with workpc's herdr as just another pane host. Resolving the federation question first would block the fixes that actually unblock unattended runtime.

Two guardrails land immediately, so Design A cannot corrupt state in the meantime:

1. Anchor computation happens where the checkout is. Under the bridge that means having herdr run the rev-parse in the pane's worktree rather than shelling out locally — or, if the protocol schema won't support it, refusing to rotate any lease held by a non-local herdr until Phase 6. A loud refusal beats a false anchor.
2. Same treatment for cleanupCompleted's git worktree remove.

Phase 6 then builds the worker binary against the endpoints that already exist, adds server-issued tokens (S10), persists the registry, and drives offline detection from a ticker rather than a request handler.

What is explicitly not acceptable is leaving both designs in place unmarked. If the worker binary is ever abandoned, delete /v1/federation/* and record the deviation — as it stands the repo reads as though the spec-conformant path is implemented when nothing can reach it.

---

Remediation plan

Ordered so each phase is verifiable on its own and nothing depends on an unproven layer below it — the spec's own build discipline (§8).

Phase 0 — Ground truth (half a day, blocks everything)

Nothing below is safe to write until the herdr surface is known.

1. On workpc: herdr api schema --json > herdr-schema.json, commit it to deploy/.
2. Write internal/herdr/schema_test.go: for every method string in adapter.go, assert it exists in the committed schema with the params we send. This is the test that would have caught pane.status and will catch the next one.
3. Manually drive one pane end to end against real herdr — create worktree, start agent, prompt with inline wait, read status, kill — and record the actual request/response shapes. Fix adapter.go to match.
4. Delete pane.rotation_signal and the RotationSignal interface unless the schema proves it exists.

Done when: schema_test.go passes against the committed schema, and a scripted manual run starts and stops a real Claude Code pane on workpc.

Phase 1 — Occupancy, measured for real (§5.2.1) — fixes B1

The spec says build this first and verify against a live session. Do exactly that, and nothing else in this phase.

1. Add SessionFile string to herdr.Session. Resolve it at lease time, per harness:
   • claude — the transcript path. Cleanest source is the Stop hook's stdin (transcript_path), which you need for Phase 2 anyway; until then resolve ~/.claude/projects/<enc-worktree-path>/<session>.jsonl by newest-mtime under the encoded worktree dir.
   • codex — CodexActiveUsage already does the state_*.sqlite → threads.rollout_path discovery. Call it. Filter by the rollout whose cwd matches the worktree.
   • opencode — OpenCodeStatus against :4096 as fast path, ~/.local/share/opencode/storage/message/ as backstop, per §5.2.1's "the SSE stream is not rock-solid".
2. Change CLIAdapter.Occupancy to use s.SessionFile, and make a missing session file a hard error surfaced in MonitorHealth, never a silent continue.
3. Keep the §5.2.1 trap explicit: assert in a test that a fixture with a large cumulative total but small last-turn usage yields low occupancy. This is the failure mode the spec singles out.
4. Add GET /v1/tasks/{id}/occupancy returning the raw numerator, the window, and the fraction — so you can eyeball it against a live session before trusting it to drive rotation.

Done when: a real Claude Code session at a known context fill reports a fraction within a few points of what /context says.

Phase 2 — Face B and completion (§5.2, §4) — fixes B3, B5

1. Claude Code Stop hook (deploy/hooks/orchestra-stop.sh, bash, no deps — §5.6). Reads hook JSON on stdin, POSTs transcript_path + task id to a new POST /v1/harness/turn endpoint. Exit 2 with stderr when the plane says "rotate but no valid handoff yet" — that's the spec's refuse-the- turn mechanism, and ClaudeStopHookUsage already parses the input shape.
2. POST /v1/harness/turn computes occupancy, evaluates the rotation triggers, and returns a decision: continue | prepare_handoff (soft) | rotate_now (hard) | refuse (over threshold, handoff missing/invalid).
3. Completion: POST /v1/harness/complete — the wrapper/hook path that uploads the report to CAS and emits TaskCompleted with report_ref + receipt. Wire the receipt from Phase 1's occupancy reader, summed across lease intervals per §4.
4. Codex/opencode Face B: rollout-tail and SSE session.status respectively, polled by the coordinator rather than hook-pushed. Same decision endpoint.

Done when: a task given a trivial goal on a real harness reaches TaskCompleted with a report_ref resolvable from CAS, with no human action.

Phase 3 — Make rotation reachable and correct (§5.3, §5.4) — fixes B2, B4

1. Key sessions by HerdrID everywhere. Add a single Coordinator.adapterFor(session) (herdr.Adapter, error) and route all four call sites through it. Add a regression test that registers an adapter under homesrv-claude, leases with Harness: "claude", and asserts rotation still fires.
2. Replace every silent continue in rotate/expire with a recorded reason on MonitorHealth. The class of bug in B1/B2 is only invisible because of those bare continues.
3. Split the router's counters: releases (informational) vs failures (drives MaxAttempts). A TaskReleased carrying a valid handoff_ref must not advance failures. Test: a task that rotates 5 times is still alive.
4. Implement the missing triggers (§5.3): soft 55% → prepare_handoff; milestone; thrash (N failed test runs / same file M times / identical tool calls) as a circuit breaker with reason=thrash and populated dead_ends; agent-initiated ROTATE.
5. Make TTL and the retry policy config, not the hardcoded 30*time.Minute at router.go:165 and 1800 in main.go (§5.4: "N, ttl, retry backoff = config").
6. Land the two Design-A guardrails (see the federation decision above): compute the anchor where the checkout is, or refuse to rotate a lease held by a non-local herdr; same for cleanupCompleted's worktree removal. These are the fixes that make it safe to defer federation to Phase 6.

Done when: a long-running task crosses 75% occupancy, rotates at a turn boundary, and the successor continues — twice in a row, unattended.

Phase 4 — Wire Layer 3 into the live path (§6) — fixes B6

This is where the dead code becomes load-bearing.

1. Write TASK.md into the worktree at creation from the TaskCreated payload; hash it; store the hash on the task; pass it as TaskFileSHA into PerProjectGitWorktrees (main.go:112). Re-inject it on every rotation bootstrap — §6.2 is explicit that this is what defeats the telephone game.
2. Handoff production: the rotating agent writes the §6.1 TOML/JSON handoff; the stop hook uploads it via POST /v1/artifacts; the plane continuity.Decodes it, rejecting free-form [knowledge] (Invariant 4), and only then emits TaskReleased with the hash. Never accept a handoff_ref the plane hasn't validated.
3. Scratch-branch commit before release (ScratchCommit), so §6.2 step 3 collapses to one sha compare. ScratchCommit already refuses to commit a dirty TASK.md — good, keep that.
4. Pickup: ValidatePickup runs in Coordinator.Start before the successor's bootstrap prompt. Fail → do not start; emit TaskBlocked.
5. Replace CLIAdapter.Bootstrap's freeform prose with the §6.2 procedure (~200 tokens: read handoff, run validate-handoff, re-read TASK.md, proceed), and use agent.prompt's inline wait.
6. Wire MarkdownChanges to the §6.3 adjacent-task notice, or delete it and record the deferral. Dead code that looks implemented is what produced this audit.

Done when: a rotation whose anchor has drifted is refused, visibly, and one whose anchor is clean proceeds without the successor re-deriving context.

Phase 5 — Close the surfaces (§7) — fixes B7, B8, S1–S4

1. Emit QuotaReported. Source: the same per-harness session state as Phase 1, on a timer per harness+window (§7.2 — "a projection keyed by harness+window, fed by the same per-harness session state"). Until then, your quota limits are decorative.
2. Reject X-Orchestra-Surface: system at the HTTP boundary unconditionally. System must be constructible in-process only. Then decide explicitly whether the remaining surfaces require tokens on your LAN, and set them.
3. Fix the go vet json-tag collisions (S1) — the brief is currently unparseable for git state.
4. Brief: per-project git sync state from the actual worktrees (S2), and include report_ref/receipt in the rollup (S3).
5. Delivery: never return on send error — log, back off, continue. Persist the cursor next to the event log (S4). Deliver the morning brief itself.

Done when: /v1/brief for an overnight window shows real quota per harness, real per-project git state, and the receipts for every completion.

Phase 6 — Federation, decided (§2.1, §9 item 8) — fixes the fork, S10

Only after Phases 1–5 run clean on a single host.

1. Anchor validation and HeadSHA execute where the checkout is. Today they run on homesrv against a workpc path.
2. Either build the worker binary against /v1/federation/* (git-only transport, per spec) or delete that API and document the bridge as a deliberate deviation. Do not keep both.
3. Add admission control to worker registration (S10): server-issued tokens, not self-declared.
4. Then run the real two-machine overnight batch that §9 item 8 asks for.

---

Corrections to progress.md

Worth fixing, because the next session will otherwise trust it:

• "Continuity: strict handoff schema/validation, CAS save/load, pickup validation, scratch-branch commit/push/pull helpers" — accurate as a description of the package, misleading as a description of the system. None of it is reachable at runtime.
• "herdr adapters with native occupancy readers" — the readers exist; they are called with the wrong argument and always fail.
• "Router: ... quota availability at conservative 80% threshold" — the threshold logic is correct, but no QuotaReported event is ever produced, so it evaluates against zero.
• The rotation fix described in "Verified fixed this pass" is real and correct — but rotate cannot reach that code, because the adapter lookup above it (B2) fails first. The test passes because it registers the adapter under the harness kind rather than the herdr id, which the production config never does.
• "go build ./... and go test ./... both pass" — true; go vet ./... does not.

---

Suggested order of attack

If you want one thing to do today: Phase 0, then B2 (a ~20-line fix that makes three subsystems reachable), then B1. Those three turn a system that cannot rotate into one that can, and everything else in the plan is building on top rather than repairing underneath.

---

Phase 0 — done, 2026-07-27

B1, B2, B4, B8, S1, S5, S6 were already fixed and landed as of this session (confirmed by reading the current code, not just trusting progress.md — see adapterFor in internal/orchestrator/orchestrator.go:190 and CLIAdapter.Occupancy in internal/herdr/adapter.go:197).

This box (homesrv) turned out to have live TCP reachability to the real herdr instance at 192.168.1.105:9245 (workpc) the whole time — the unavailable: connection refused lines in journalctl -u orchestra.service are for homesrv-* herdrs dialing 192.168.1.104:9245, which has no local herdr running; workpc-* herdrs were connecting fine but main.go never logs a success, only a failure, so there was no positive signal either way. Also found: a real task is stuck live right now — workspace wA, task 06FT6CKD9Y98AZRX6X8K3QXFZG, opencode agent, pane wA:p1, agent_status: "blocked" — almost certainly stuck because Release/rotation could never reach it (see below).

Ran the actual Phase 0 steps against this live instance (raw JSON-RPC probes over TCP, params-omitted/empty-object tricks to read Rust serde's missing-field errors — no herdr CLI available locally, so herdr api schema --json itself wasn't run, but the equivalent info was extracted this way). Full method list and findings committed to deploy/herdr-schema.json.

Confirmed, with a real server response, not just static reading of adapter.go:

• pane.release, pane.kill, pane.rotation_signal — none of these exist in the real protocol. Confirms B5's suspicion exactly.
• Real replacement for pane.kill is pane.close({pane_id}) — same shape, drop-in. Fixed in internal/herdr/adapter.go.
• Real replacement for pane.release is pane.release_agent({pane_id, source, agent}) — structurally different, and per B5's own analysis it cannot return a handoff_ref regardless (herdr doesn't write handoffs, the agent does, §6.1). Wiring this for real needs Phase 4's handoff-production path first. CLIAdapter.Release now returns a loud error naming exactly that instead of calling a method that doesn't exist. Not a full fix — Phase 4 still owns making Release do something real.
• pane.rotation_signal doesn't exist and never will (herdr has no rotation concept) — deleted RotationSignal interface, its CLIAdapter method, and the call site in Coordinator.rotate, per this doc's own instruction ("Delete ... unless the schema proves it exists").
• agent.get, pane.read, agent.prompt, worktree.create, worktree.open, agent.start — all confirmed real, no changes needed there.
• Protocol version confirmed live: 17, matching config.jsonc's "protocol": "17" (returned as a bare JSON number by the server; the existing string-fallback parse in CheckProtocol happens to handle that correctly already).

go build ./..., go vet ./..., go test ./... all pass after these changes.

B3 — partial fix, 2026-07-27

Added POST /v1/harness/complete (cmd/orchestra/main.go) — the first automatic TaskCompleted producer. Design: a Claude Code Stop hook (deploy/hooks/orchestra-stop.sh) runs on every turn boundary but only POSTs when the agent has written .orchestra-report.md at the worktree root — that file is the explicit "I'm done" signal, since Stop fires on every pause and treating every stop as completion would be wrong (this is exactly the distinction Phase 2 items 1–2, the continue/prepare_handoff/rotate_now/ refuse turn-decision endpoint, are meant to own — that endpoint is still unbuilt, so there is currently no plane-side signal telling the agent when to rotate vs. finish; the marker-file convention is a stopgap that only covers the completion half).

The handler builds the receipt server-side from herdr.ClaudeUsage against the transcript path the hook supplies (same local-filesystem assumption as CLIAdapter.Occupancy — doesn't hold for a session hosted on a herdr that isn't local to the machine running orchestra, i.e. the federation-fork caveat applies here too) rather than trusting a self-reported number, and uploads the report body via Store.PutArtifact for report_ref. Event is appended with Surface: string(authz.System) hardcoded in Go — not read from a request header — consistent with B8 (system must never be header-controlled); gated instead by an optional ORCHESTRA_HARNESS_TOKEN bearer check.

Not done: Codex/opencode completion producers (only Claude wired), the turn-decision endpoint itself, and no test — cmd/orchestra/main.go has zero handler test coverage of any kind (everything is inline in main()), so this follows the existing gap rather than introducing an isolated test harness for one handler.

Fixed 2026-07-27 (later same day): CLIAdapter.Lease's bootstrap prompt now passes time.Minute instead of wait=0, matching Bootstrap's inline-wait pattern — closes the race B5 named (send-into-a-half-rendered-prompt). internal/herdr/adapter.go. go build/vet/test all still pass; no existing test asserted the old wait=0 value.

Still open from B5 (not attempted this pass — larger, needs design, not just a method-name swap):

• Release's real implementation, which depends on Phase 4 (§6) handoff production existing at all.
• The stuck live task (06FT6CKD9Y98AZRX6X8K3QXFZG) was deliberately not manipulated directly (no pane.close/pane.release_agent call against it) — killing or releasing a real running agent from an audit session without the user present is exactly the kind of action that warrants asking first.

B5 — closed, 2026-07-27 (later same day)

CLIAdapter.Release now does something real instead of refusing. Design mirrors the .orchestra-report.md marker convention B3 already established for completion, since the same problem applies to handoffs: the plane must never invent a handoff, only validate and forward the one the agent wrote (§6.1). Concretely:

1. The agent is expected to write .orchestra-handoff.json (herdr.HandoffFile) at the worktree root before its stop hook lets rotation proceed — a §6.1 handoff schema, not free prose.
2. Release reads that file, decodes it with continuity.Decode (schema + required-field validation, same as pickup), and cross-checks Anchor.GitSHA against herdr.HeadSHA(session.Worktree) — the anchor is re-verified against the real checkout, not trusted from the agent's self-report, closing the same class of gap as B3's receipt-from-transcript choice.
3. Only if both checks pass does it upload the handoff via continuity.Save (CLIAdapter.CAS, wired to the same *store.Store used everywhere else) and return the resulting ref — this is what Coordinator.rotate puts in TaskReleased.handoff_ref.
4. Only then does it call the real pane.release_agent({pane_id, source: "herdr:"+harness, agent: harness}) to drop herdr's claim — sequenced last so a herdr-side failure can't strand an already-uploaded handoff with no way to retry the release call (retrying Release re-reads the same file and is idempotent).

A missing or invalid handoff file, an anchor mismatch, or a pane.release_agent error are all refused (non-nil error, no event emitted) — Coordinator.rotate already treats an errored Release as "leave the lease intact, retry next tick," so this gives the agent room to finish writing the handoff rather than stranding the task.

herdr.Claude/Codex/OpenCode constructors now take a continuity.CAS parameter; cmd/orchestra/main.go passes the existing *store.Store (which already implements PutArtifact/Artifact).

New tests in internal/herdr/adapter_test.go drive Release against a real git worktree and a fake in-process herdr TCP listener (fakeHerdr) responding to pane.release_agent: upload-and-release on a valid handoff, refusal with no handoff file, refusal on anchor mismatch, refusal with no CAS configured.

Still not done (unchanged, separate from B5 itself): nothing yet makes the agent actually write .orchestra-handoff.json — that's Phase 4 item 2's other half (a stop-hook-side convention, analogous to .orchestra-report.md/deploy/hooks/orchestra-stop.sh for completion) and Phase 4 items 3/5/6 (ScratchCommit before release, the §6.2 bootstrap-prompt rewrite, MarkdownChanges wiring). go build ./..., go vet ./..., and go test ./... all still pass.

B6 — partial fix, 2026-07-27 (Phase 4 items 1 and 4)

Two of Phase 4's six items landed; the rest are unchanged (still open, listed below).

1. TASK.md is now actually written. continuity.RenderTaskFile(t) produces the §6.2 immutable spec content from domain.Task; GitWorktrees.Create (internal/orchestrator/orchestrator.go) writes it into every freshly created worktree and commits it immediately — it must be committed, not left dirty, both so ScratchCommit's "TASK.md is immutable" check (which inspects git status) sees it as clean, and so its hash is stable across whatever the agent does afterward. A worktree that already has a TASK.md (recreation on restart) is left untouched. New continuity.TaskFileHash(root) reads it back and hashes it — this is what makes w.TaskFileSHA/VerifyTaskFile, previously dead because nothing ever set TaskFileSHA, actually reachable.
2. Pickup validation now gates bootstrap. Coordinator.Start (internal/orchestrator/orchestrator.go) computes the new worktree's TaskFileHash, and — whenever the lease carries a handoff_ref (i.e. this is a rotation continuation, not a fresh lease) — loads the handoff from CAS and runs continuity.ValidatePickup(worktree, handoff, taskFileSHA) before calling Adapter.Bootstrap. A validation failure kills the just-started session and emits TaskBlocked instead of hand­ing the successor an unverified anchor. This is exactly the gap B6 named: ValidatePickup had no caller outside its own tests. TestStartBlocksOnInvalidPickup (rotation_test.go) drives this against a handoff whose anchor SHA doesn't exist in the repo and asserts the task ends Blocked, never bootstrapped. TestGitWorktreesCommitsTaskFile (worktrees_test.go) asserts the written/committed content matches RenderTaskFile and survives recreation.

Caveat recorded in code: TASK.md hashing is best-effort — if a worktree came from the WorktreeCreator (herdr-hosted, potentially remote) path rather than GitWorktrees, TaskFileHash fails silently and pickup validation runs with an empty taskFileSHA (so it still checks anchor SHA and dirty-file hashes, just not TASK.md). No current adapter or test exercises WorktreeCreator with a real handoff_ref, so this is unverified, not proven safe — same cross-host caveat as the federation-fork section.

Still open from B6/Phase 4 (unchanged, larger and needs live-agent cooperation):

• Item 2: handoff production — the rotating agent writing the §6.1 handoff and a stop-hook path uploading it via POST /v1/artifacts before Coordinator.rotate calls Adapter.Release. Release still just refuses (see B5 above) — there is nothing yet to validate-and-mint a ref from.

Phase 4 items 3, 5, 6 — landed 2026-07-27

1. ScratchCommit wired into Release, not into rotate. Rather than calling it from Coordinator.rotate (which only has a herdr.Session, not the handoff), CLIAdapter.Release now runs it itself, after validating the agent-authored handoff's Anchor.GitSHA against the worktree's real HEAD and re-verifying every Anchor.Dirty file's hash still matches what the agent recorded (previously untested — a file edited after the handoff was written but before release would have silently sailed through). If the handoff has dirty entries, Release commits them atomically onto orchestra/scratch/<handoff-meta-id> via continuity.ScratchCommit, then rewrites the handoff's anchor to the new scratch commit SHA with Dirty cleared, before uploading to CAS — this is what "collapses §6.2 step 3 to one sha compare" means in practice: the successor's ValidatePickup now only needs git rev-parse HEAD == handoff.anchor.git_sha, no per-file rehashing, because everything was committed before the ref was minted. ScratchCommit itself was changed to be idempotent — reuse an existing scratch branch (git switch before falling back to git switch -c) and skip the commit if there's nothing to snapshot — since a task can rotate, and therefore hit this path, more than once. Covered by TestReleaseScratchCommitsDirtyFilesBeforeUpload (asserts the anchor advances to the new commit, dirty is cleared, and the worktree ends up on the scratch branch) and TestReleaseRefusesOnStaleDirtyFile (internal/herdr/adapter_test.go).
2. Item 5 — Bootstrap prompt rewritten. CLIAdapter.Bootstrap no longer sends the one-line "read handoff, validate anchor, continue" prose. It now tells the agent the plane has already validated anchor/TASK.md (true, per B6's ValidatePickup gate in Coordinator.Start — no need to ask the agent to redundantly re-verify trust), and points it at git log --stat -5 / git branch --show-current in the worktree as the actual source of "what the prior agent did and what's left," since that's now a real, inspectable scratch-branch commit rather than an opaque ref. Deliberately does not claim a GET /v1/artifacts/<ref> fetch path — no such HTTP route exists (/v1/artifacts is POST-only, upload only, confirmed by reading cmd/orchestra/main.go); an earlier draft of this prompt invented that endpoint and was corrected before landing, which is exactly the class of bug this audit exists to catch.
3. Item 6 — MarkdownChanges deleted, not wired. Confirmed zero callers anywhere (including its own tests — there were none, despite being listed as "believed accurate" in a prior progress.md snapshot). Wiring it for real needs a design for what "adjacent task" means and where the notice surfaces (brief? a new event type?), which is a real feature, not a wiring fix — AUDIT.md explicitly allows "delete it and record the deviation" as the alternative to half-implementing that. Taking that option rather than bolting on an undesigned notification path.

Phase 4 item 2 — closed 2026-07-27. The remaining gap after B5 was not "which harness" (Release was always harness-agnostic) — it was that nothing, for any harness, ever told the agent the .orchestra-handoff.json convention existed. Coordinator.rotate (internal/orchestrator/orchestrator.go) now checks whether the resolved adapter implements a new optional herdr.HandoffRequester capability; if HandoffFile isn't present in the worktree yet, it calls RequestHandoff once (herdr.Session.HandoffRequested guards against re-prompting every tick) and skips Release for that tick, leaving the lease intact — exactly the same "ask, don't invent" shape as B3's .orchestra-report.md convention. CLIAdapter.RequestHandoff (internal/herdr/adapter.go) sends a prompt naming the exact §6.1 JSON shape (meta.id, anchor.git_sha/branch/dirty[].{path,sha256}) and explicitly tells the agent not to fabricate the SHA/hashes. Once the file exists, the existing Release path (validate → scratch-commit dirty files → upload → CAS → pane.release_agent) is unchanged. Covered by TestRotationRequestsHandoffBeforeReleasing (internal/orchestrator/rotation_test.go), which asserts Release is never called while the file is absent and fires once it's written. go build, go vet, go test ./... all pass.

Not attempted here (separate, deployment-level question, not a code gap): whether Codex's/opencode's own turn-boundary mechanism actually surfaces this in-pane prompt to the agent before it exits the way Claude Code's Stop hook does — that's Phase 2 item 4 territory (native Face B per harness), not Phase 4.