fix(bin): prevent receiver wake failures from blocking remote handoffs - #4033
Merged
Merged
Conversation
Owner
Author
|
Validation note (documented limitation, accepted at the no-mistakes test gate) The four remote-transport scenarios this change covers are exercised by executable integration tests over the deterministic fake SSH boundary in tests/fm-remote-backlog-handoff.test.sh, driving the real scripts and functions:
They were not driven against a live remote secondmate inside the pipeline, because no live remote target exists there. |
 The PR appears safe to merge, with the changed retry and release behavior aligned with the accepted durable-receipt contract. |
A remote backlog handoff holds its outbox until the backlog receipt and the receiver wake are both confirmed, and retries the wake under the same pending-reply correlation on every resume. When that wake's remote transport was lost, the correlation stayed undelivered in delivery_unknown and the watcher's next pending-reply tick escalated it. Both the reuse predicate and the known-undelivered reset refused an escalated record, so the resume refused to resend the wake forever and every later handoff to that mate jammed behind the outbox. Treat an escalated record with no confirmed delivery as the undelivered correlation it is: fm_pending_reply_corr_reusable accepts it for its own task and fm_pending_reply_reset_known_undelivered returns it to awaiting_report for the idempotent remote resend, while a delivered record is still never reset and a missed-report escalation keeps its meaning. The published delivery-unknown decision stays open until the record resolves, so a repeat loss neither re-notifies nor strands it. Reproduce the deadlock end to end in the remote handoff test (lost wake transport, watcher escalation, resume) and pin the predicate contract in the pending-reply suite; the fm-send fixture that pinned the refusal now uses a genuinely stale delivered escalation.
kunchenguid
force-pushed
the
fm/fm-handoff-wake-escalation-deadlock-r1
branch
from
September 10, 2026 05:26
9f9856a to
dba3175
Compare
kunchenguid
changed the title
msylergy
added a commit
to sylergydigital/firstmate
that referenced
this pull request
Sep 14, 2026
…y fork (#1) * fix(bin): suppress false Claude long-turn supervision alarms (#4119) * fix(bin): stop false watcher-down alarms on long Claude turns A healthy Stop auto-arm rewake or open claim already explains a mid-turn beacon that has aged past grace, because turn-end will re-arm. Keep the supervision-off banner for a missing, failed, or exhausted generation. * no-mistakes(review): Bind Claude rewakes to active recovery generation * no-mistakes(document): Document Claude long-turn supervision exception * no-mistakes(lint): Fix empty ShellCheck assignment * no-mistakes(ci): Fixed all reported CI failures: quoted the hyphenated recovery-delivery value to satisfy ShellCheck SC2100, and updated the session-lock auto-arm fixture to emit the recovery marker and watcher beacon now required for a valid rewake. Verified fm-session-lock-ancestry, fm-test-run, stale-banner, Claude auto-arm, targeted lint, ShellCheck, and workflow lint checks pass * fix(herdr): recover gone and drifted worker endpoints (#4120) * fix(herdr): classify a gone session's endpoint as recoverable A task whose Herdr endpoint could not be read was classified `unreadable`, which blocks recovery by design. The commonest reason that read fails is that the recorded session's server is not running at all - a host reboot, a server exit, a session never restored - and that is authoritative absence for every pane in that session, not an ambiguous answer about one of them. Tasks in that state had no sanctioned way back. The recovery-grade read now settles an uninterpretable pane read with the session server's own `.server.running` state: positively stopped reads `missing`, while a running server, or a server state that cannot itself be read, still reads `unreadable`. Resting the verdict on that field rather than on the `server_not_running` error code is what keeps it working across Herdr 0.8.x and 0.9.0, since the field is present on both and the code is not. Only that one boundary is widened. The husk classifier under it stays strict, so duplicate prevention, rollback, and teardown - the paths that can destroy something - keep refusing on exactly the reads they refused on before. Separately, a relaunch refused outright when the endpoint's shell had drifted out of the recorded worktree. An agent's own exit routinely leaves its shell somewhere else, so that refusal stranded tasks whose work was sitting untouched on disk. The shell is now told once to return, and only a shell that will not go refuses; the replacement still never starts outside the copy holding the work. Herdr 0.8.x is not installed on this host, so protocol-20 coverage is structural plus the adapter fixture exercising both response shapes, and is recorded as such rather than as a live result. Fixes #4091. * no-mistakes(review): Restrict drift recovery to Herdr endpoints * no-mistakes(review): Correct Herdr recovery verification coverage * no-mistakes(document): Document Herdr endpoint recovery boundaries * fix(bin): prevent receiver wake failures from blocking remote handoffs (#4033) * fix(bin): keep an escalated undelivered handoff wake retryable A remote backlog handoff holds its outbox until the backlog receipt and the receiver wake are both confirmed, and retries the wake under the same pending-reply correlation on every resume. When that wake's remote transport was lost, the correlation stayed undelivered in delivery_unknown and the watcher's next pending-reply tick escalated it. Both the reuse predicate and the known-undelivered reset refused an escalated record, so the resume refused to resend the wake forever and every later handoff to that mate jammed behind the outbox. Treat an escalated record with no confirmed delivery as the undelivered correlation it is: fm_pending_reply_corr_reusable accepts it for its own task and fm_pending_reply_reset_known_undelivered returns it to awaiting_report for the idempotent remote resend, while a delivered record is still never reset and a missed-report escalation keeps its meaning. The published delivery-unknown decision stays open until the record resolves, so a repeat loss neither re-notifies nor strands it. Reproduce the deadlock end to end in the remote handoff test (lost wake transport, watcher escalation, resume) and pin the predicate contract in the pending-reply suite; the fm-send fixture that pinned the refusal now uses a genuinely stale delivered escalation. * no-mistakes(review): Decouple durable outboxes from best-effort wake retries * no-mistakes(review): Align handoff documentation with durable receipt release policy * no-mistakes(review): Handle unrecordable wake state as dropped * no-mistakes(review): Prevent stale wake markers blocking handoffs * no-mistakes(review): Prevent stale delivered markers suppressing new wakes * no-mistakes(document): Clarify retry escalation decision lifecycle * no-mistakes(document): Document pending receiver wake retries * fix: restrict captain address rule to user chat (#4075) * docs: bound the mandatory captain address to the chat channel AGENTS.md's opening address rule said "address the user as captain at least once in every response" and never said what a response is. The artefact exclusion two lines below governed only the optional nautical seasoning, not the mandatory address. An agent that reads this file without being the first mate - a pipeline corrector agent running inside a copy of this repo - therefore read the obligation as applying everywhere and the exclusion as applying only to flavour, and opened its delivery message with "Captain,". That reading was correct. Patch the existing owner rather than adding a rule elsewhere: - bound the obligation to chat messages sent to the captain; - state the artefact exclusion once, explicitly binding every agent that reads this file whether or not it is the first mate, and naming commit messages, PR and issue descriptions, briefs, code and comments; - fold the seasoning under the same bound instead of carrying a second, narrower copy of the exclusion. The obligation itself is unchanged: the captain is still addressed in every chat message. AGENTS.md goes from 603 to 602 lines: the redundant "never send a response with zero direct address" clause and the duplicated seasoning exclusion pay for the new bound. The two cross-references that paraphrased the unbounded wording (bin/fm-parent-channel-lib.sh's header and docs/secondmate-parent-channel.md's problem statement) now match the owner; neither restates the rule. * fix(review): Limit address exclusions to artifacts while preserving public replies * fix(document): Consolidate captain address guidance * fix(herdr): allow detached teardown of persisted-focused tabs (#4131) * fix(herdr): close persisted-focused tabs when no live client is attached The teardown active-tab guard treated Herdr's last-focused pointer as a live viewer, so detached sessions could not close panes on that tab. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(herdr): allow detached seeded-tab prune after live-client gate Projection create still restored the persisted focused tab after a successful prune, so a detached last-focused seeded tab still quarantined the spawn. Co-authored-by: Cursor <cursoragent@cursor.com> * fix(herdr): probe live client after seeded prune only when that tab was focused The extra title-clear read after every prune shifted canned CLI fixtures and failed projection create. Co-authored-by: Cursor <cursoragent@cursor.com> * no-mistakes(review): Tighten Herdr active-tab close guard * no-mistakes(review): Guard Herdr mutations with fresh target focus * no-mistakes(document): Document Herdr live-viewer teardown guard --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(bin): escalate decision-owned wakes once as the decision (#4169) * fix(bin): escalate decision-owned wakes once as the decision The away-mode daemon treated a needs-decision: queued payload as an unknown wake, so suppression markers never committed and the same open decision re-escalated on every poll. Classify that payload through the existing signal path so it escalates once, labelled as the decision, and an unchanged repeat is suppressed on the same terms as any other signal. Fixes #4096 * no-mistakes(review): Escalate captain-held decision-owned rows once as the decision * no-mistakes(review): Self-handle captain-held decision-owned rows instead of escalating them * no-mistakes(document): Name away daemon as needs-decision payload reader * test(herdr): cover agent exit-to-shell liveness (#4172) * test(herdr): pin leftover-shell vs live-idle via agent get Herdr 0.9.0 already distinguishes a Pi that exits to a surviving pane shell from a sibling live idle occupant. Pin that pair through agent get and the recovery classifier so a lagged pane-get status cannot silently reclaim the leftover shell as alive. Co-authored-by: Cursor <cursoragent@cursor.com> * no-mistakes(document): Document Herdr leftover-shell liveness regression * no-mistakes(ci): Fixed Lint failure SC2034 by replacing the unused wait-loop variable with `_`. Verified with the pinned project lint command, Bash syntax check, and git diff check --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(bin): rebalance portable parallel test lanes using CI timings (#4151) * ci: rebalance the portable parallel lanes on measured runner durations Both portable parallel lanes are capped at 10 minutes. Lane 1 was cancelled at that cap on every request raised on 2026-09-10 while lane 2 finished in about 3.5 minutes, so no request could go green. CONTRACT CLASS: RESTORE. The workflow already promises two duration-balanced lanes and the shard documentation already claims a measured wall; this re-establishes both against what the lanes now cost, and changes no lane count, no cap, and no scope of what runs. The counter-argument, so nobody has to take that on trust: two pieces here are genuinely new rather than restored, and either could be argued to make this a NEW-behavior change. `--list-scheduled` now ranks a parallel lane on measured durations where it previously handed every parallel script the serial default weight and returned an alphabetical order; and `--check-coverage` gains three reported fields. I classify the change RESTORE because both exist only to make the already-promised property checkable, but they are named here rather than folded into the restoration. === PART 1: THE TOTAL, AND HOW IT WAS OBTAINED === This section stands on its own. It establishes what the parallel set costs. It derives no packing; Part 2 does that, from this number. THE TOTAL: 828568 ms, about 13 min 49 s of serial work across the 24 scripts. Lane 1 held 624299 ms of it and lane 2 held 204269 ms, a 3.06:1 split. HOW IT WAS OBTAINED. The difficulty was that lane 1 had never finished, so its duration did not exist as a recorded figure anywhere and no timing artifact was expected for it. It turned out to be recoverable from the real lane without estimating, by two routes, across six CI runs on 2026-09-10 (34459949083, 34460760299, 34462530836, 34462758357, 34466966385, 34470382458): - Run 34462758357's lane-1 job finished its suite 18 s BEFORE the wall and uploaded a complete fm-test-timing-portable-parallel-1 artifact carrying all 11 scripts, FM_TEST_SUMMARY total=11 failed=0 duration_ms=598225. The upload step is if: always(), so the cancellation did not suppress it. This is one full, untruncated lane-1 measurement. - The five other lane-1 jobs were cancelled mid-suite, but each logs every script that had already finished as an FM_TEST_END duration_ms= marker. Those per-script records are complete measurements of completed scripts; only the script in flight at cancellation is lost, and it differs by run. Lane 2 completed in all six runs, so its scripts come from the six uploaded fm-test-timing-portable-parallel-2 artifacts. Every one of the 24 scripts therefore carries at least one untruncated measurement: 20 of them measured in all six runs, two in three or four runs, and two (fm-brief, fm-transition-lib, the tail of lane 1) in the single complete run. Each hint is the SLOWEST value that script reached, so the total is an upper envelope rather than an average. NO FIGURE IN IT IS DERIVED FROM A TRUNCATED LANE, and no lower bound was ever extrapolated into a total. THE ENVIRONMENT, AND WHETHER IT TRANSFERS. Every hint is a serial run of the real portable parallel lane on a GitHub ubuntu-latest runner, produced by the lane's own CI job. It transfers because it is not a proxy for the lane; it is the lane. Nothing in the total came from this machine or from any harness of mine. That mattered, and here is what it would have cost. A same-day macOS cross-check of the same scripts ran 1.7x to 5.0x slower with the ratio varying per script (fm-test-run 157420 ms against 92944 ms, fm-x-mode 67217 ms against 31870 ms, fm-composer-ghost 10521 ms against 2120 ms). Local timings therefore do not scale the lane, they REORDER it, so a packing derived from them would have balanced the wrong thing while looking clean. WHAT IT REPLACES, which is the root cause. The lanes were packed from the 2026-08-20 concurrent isolation proof: 24 candidates across four LOCAL workers. That record answers whether the candidates are isolation-safe, not how long a SERIAL CI lane runs, so it was structurally incapable of representing lane wall clock even when it was fresh. It was also never refreshed while the set grew about 3.2x. Both the wrong instrument and the staleness are fixed here: the hints now come from the lane itself and carry their run ids and date. === PART 2: THE SPLIT DERIVED FROM THAT TOTAL === Longest-processing-time assignment over those hints gives 414269 ms and 414299 ms, 30 ms apart, against 624299/204269 before. tests/fm-pi-primary-types.test.sh stays in lane 1 because that is the job which installs the Pi package, so ci.yml needs no step changes. === PART 3: DOES THE MARGIN SURVIVE MACHINE VARIANCE === Stated explicitly, because 6.90 min against a 10 min cap is 69% of cap before any variance is applied, and the cap covers the whole job rather than the suite. worst lane, script time 414299 ms 6.90 min job overhead, measured on the real lane ~18 s (see below) expected healthy job ~432300 ms 7.21 min x1.29 on the script time, plus overhead ~552400 ms 9.21 min cap 600000 ms 10.00 min room left after the multiplication ~47.6 s 7.9% of cap The 1.29x is the runner variance measured today on the SIBLING SERIAL lane, as supplied; it is not this lane's own figure. This lane family does have its own, and it is tighter: the six full lane-2 sums today span 192939 ms to 203451 ms, a spread of 1.054x. At that figure the worst lane lands near 7.58 min with about 2.4 min of room. I have used the LARGER, borrowed 1.29x for the verdict rather than the tighter one this lane actually shows, and note that the hints are already per-script maxima, so 1.29x on top is conservative twice over. THE MARGIN SURVIVES THE MULTIPLICATION, so this proceeds rather than stopping. The 18 s overhead is measured, not assumed: in run 34462758357 the lane-1 job ran 10 min 16 s against a 598.2 s suite, and lane 2 ran 3 min 21 s against a 192.9 s suite, a ~10 s difference that matches lane 1's extra Pi package install. The cap is unchanged, the lane count is unchanged, and nothing in the serial lane, its shard count, its guard or its hint table is touched. === PART 4: THE RECORDED FACT === The workflow comment no longer restates the shard wall as a literal, which is how "~1 min of serial sum" survived a 10x change without announcing it. It now points at bin/fm-test-run.sh --check-coverage, which prints parallel_max_ms, parallel_imbalance_ms and parallel_unhinted derived from the hint table, so the current number is computed on demand. The shard documentation carries the dated run ids, which route it was taken by, and the local cross-check that shows why local numbers are not admissible as hints. Two regressions pin what rotted: lane membership must be stored longest-measured-first, and the lanes must be fully hinted and packed within 5% of each other. Both were run against the old composition and both fail on it (420030 ms imbalance against a 624299 ms worst lane). The ordering assertion they replace named a specific script by hand and had itself gone stale. === PART 5: NAMED AND LEFT, OUTSIDE THIS REBALANCE === tests/fm-captain-hold-lifecycle.test.sh alone is 296481 ms, 36% of the whole set, so it is the floor of any two-lane split: no repacking can put a lane below it. After this rebalance the cap is about 1.45x the healthy lane where the sibling serial lane keeps roughly 2x. Nothing refuses a stale parallel hint the way PORTABLE_SERIAL_MAX_UNHINTED_PERCENT bounds the serial lane. parallel_unhinted is reported, not enforced, which is what let this drift for three weeks unnoticed. * fix(review): Restrict parallel scheduling hints to portable parallel lanes * fix(document): Clarify parallel lane scheduling and timing evidence * fix(bin): stop claiming prose-mentioned PR URLs as a task's delivered PR (#4148) pr_for_task fell back to scraping the whole status log with tail -1, so any PR URL a worker ever mentioned in prose - including a scout citing someone else's PR - became the task's delivered PR in the parent-channel terminal report. Recorded meta pr= is now the only authoritative source, the fallback scrape accepts only a preferred terminal line in a mode's ready-signal shape (done: PR <url> or done: PR <url> checks green), and a scout never carries pr= at all. * fix(procevent): confirm reconcile launches and reclaim provably dead claims instead of counting a dead drop as started (#4212) * fix(procevent): stop a dead runner owning a source and reconcile reporting it The captain answered ten calls on a bearings board, the board accepted them, and nothing collected them. He had to answer all ten again in chat. A surface that presents as armed while being a dead drop is worse than one that visibly fails, because the answers looked recorded. Two independent defects, reproduced together in an isolated home where reconcile reports started=1 on every run while ownership never moves and no runner ever attaches. 1. reconcile counted a launch it never verified. detach_runner is fire-and-forget and discards the child's stderr, so a runner that died before it could claim was counted exactly like one that is listening. Launches are now confirmed - the source observed owned, or its runner record moved - before being reported as started; the rest are reported as failed= with a non-zero exit. The runner-record clause is what keeps a fast-completing source from being reported as a failure when it finished between two polls. One bounded window covers a whole cycle's launches, so a home full of broken sources costs the same wait as one. 2. A claim whose whole generation is provably gone could be refused forever. Reclaiming it ran cleanups over that dead generation's own leftovers, and any failure vetoed the claim - permanently, because none of those conditions clears on its own. Every one of those leftovers is keyed by the dead generation's claim token and a replacement always claims a fresh one, so none can collide with what replaces it. fm_procevent_claim_capture_reservation_reclaim_locked already said this for the reservation record; the staging file and the shape check on the registry directory recorded to hold it now take the same rule. Removing the claim record itself stays a hard precondition: two owners is the one outcome worse than none. Two smaller repairs to the same "registered is not listening" confusion: - `list` reported OWNER=none for a source nothing can claim. A reused PID whose process group survives reaches that state through the stale branch rather than the leaderless one, so it read as an idle source waiting to be started - the reassuring answer this surface gave while a board collected nothing. It now reports the orphaned state it shares. - reconcile relaunched into that same unclaimable state on every cycle, spawning a runner that could only die on the claim. docs/configuration.md already promised it preserves such a claim without starting a replacement; the code now does that and reports it as uncertain. This is NOT a third instance of today's two lock-identity defects (4e1bf9aa and its replayed predecessor). Those were wrong liveness predicates: a reused PID read as a live holder, then an exec'd holder read as dead. Here the predicate is right - the code correctly proves the owner dead and refuses the claim anyway, on a condition unrelated to liveness. Regression coverage, each failing on the parent commit for its own reason: - tests/fm-procevent.test.sh: a source that cannot start is reported as failed rather than started; a dead generation whose leftovers cannot be tidied no longer keeps owning its source (the parent reports a start while nothing ever runs); the existing reused-PID fixture now also asserts the orphaned listing and that no doomed relaunch is reported. - tests/fm-captain-hold-lifecycle.test.sh: a board answer reaches the keyed-answer intake through the runner end to end - durable capture, the wake, and the closed task carrying the captain's selection. This one passes on the parent, because that chain was never what broke. fm-procevent 100, fm-bearings-board 18, fm-captain-hold-lifecycle 50, fm-procevent-when 13 and fm-procevent-quota 18 pass; bin/fm-lint.sh and bin/fm-doc-audience-check.sh clean. tests/fm-extension-binding.test.sh has two failures identical on the parent commit (EACCES on package install in this sandbox) and unrelated to this change. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_016gxgshn5jkWJ3GEYWy7vTG * no-mistakes(review): confirm reconcile launches on durable launch stamps * no-mistakes(review): announce stranded sources and refuse bad confirm windows * no-mistakes(review): announce leaderless strands, bound confirm window, fix recovery docs * no-mistakes(review): announce unconfirmed launches once per episode, qualify start reclaim * no-mistakes(review): nonce launch-failed keys, refuse bad window at arm * no-mistakes(review): state only observed launch outcome, shorten episode nonce * no-mistakes(test): assert launch-failed headline not re-delivered, allow recovery wake * no-mistakes(document): docs: cover strand and launch-failure wakes in skill trigger and verification record * no-mistakes(lint): restructure SC2015 chain into explicit if-block * test(watch-triage): fix two timing-exposed defects the pipeline found Both surfaced in the no-mistakes test step on this branch, each failing one full run of tests/fm-watch-triage.test.sh; neither was accepted as a flake to retry past. 1. The new launch-failed delivery test assumed an already-surfaced key never wakes the watcher again. That is false: a fresh watcher legitimately re-surfaces any unacknowledged queue row through its downtime-recovery path ("check: rearm-resurface"), so the assertion failed whenever a re-arm landed between its two checks. The pipeline's own fix tolerated any wake lacking the repeated key's headline; this tightens it to exactly one tolerated reason, by its exact line, with a failure message that names the expectation so a reworded path reads as "the tolerated recovery path changed" rather than as a mystery - and so nobody restores the strict silence check. The positive assertion (a fresh-suffix key is delivered under its own headline) is unchanged. 2. seed_captured_procevent_result retired its source in the gap between the runner publishing its wake and releasing its claim, so retire read the exiting runner's ownership as uncertain and refused ("cannot confirm runner identity"). The fixture and retire path pre-date this branch; the confirm window returns reconcile closer to the moment of capture, which made the gap easier to hit. The fixture now waits, bounded, for the claim release the publish promises, with the reason at the wait. Verified on this head with tasks-axi on PATH: fm-watch-triage 113/113 with no skips, fm-procevent 106/106, fm-captain-hold-lifecycle 50/50, fm-watch-arm 15/15, fm-bearings-board 18/18, fm-procevent-when 13/13, fm-procevent-quota 18/18; bin/fm-lint.sh and bin/fm-doc-audience-check.sh exit 0. First attempt, no retries. * no-mistakes(document): docs: route stranded and launch-failed wakes in skill handling --------- Co-authored-by: Claude Opus 5 <noreply@anthropic.com> * fix(herdr): verify agent liveness at process level before trusting registration (#4191) * fix(herdr): verify agent registrations at process level before trusting them Herdr keeps a Pi registration (`agent get` -> agent=pi, agent_status=idle) after the Pi process has exited to a plain shell whenever a nested interactive shell sits under the pane's top shell, which is the crew shape `treehouse get` leaves behind. The pane classifier trusted that registration alone, so `fm-control.sh <id> relaunch`, `fm-spawn.sh --relaunch`, and the crew-state recovery read all treated a shell-only pane as a live agent and refused recovery for as long as the record lived. The Herdr adapter now reads `pane process-info` plus the real process table through a shared harness-process classifier (bin/fm-agent-process-lib.sh, moved verbatim out of the tmux adapter so both backends mean the same thing by agent, shell, and other) before a registered agent counts as live. A registration over a shell-only pane is the new explicit `stale-agent` pane state, which the recovery-grade read maps to `dead`; husk detection, reclaim, presentation recovery, and session cleanup keep refusing it, so recovery reuses the pane and nothing gains close authority. A working record is verified the same way before the native busy verdict reports busy, so the recovery classifier never reports a shell-only pane as working. An unreadable process view reads unknown, trusting neither the registration nor its absence. Reproduced and measured on Herdr 0.9.0 with Pi 0.85.1 in an isolated lab; the new default-on live guard tests/fm-herdr-pi-stale-registration-live-e2e.test.sh exercises the real stale record, tests/fm-control-herdr-smoke.test.sh proves exit and relaunch through the control plane, and the portable suites pin the classifier over real processes. Fixes #4115. Duplicates: #3639, #3487, #2908, #3545. * no-mistakes(review): settle transient prompt helpers before trusting herdr process state * no-mistakes(review): drop stray codegraph file; read spaced comm whole in descendant walk * no-mistakes(review): untrack stray .codegraph/.gitignore * no-mistakes(review): untrack codegraph file; make spaced-path walk test discriminating * no-mistakes(review): untrack stray .codegraph/.gitignore * no-mistakes(review): untrack stray .codegraph/.gitignore re-added by fix round * no-mistakes(review): untrack stray .codegraph/.gitignore * no-mistakes(review): untrack codegraph file, drop dead control case, record process-info floor * no-mistakes(review): refuse stale-agent on fresh herdr spawn preflight Documented non-goal: fresh-spawn, reclaim, and presentation-recovery auto-recovery for a stale-agent pane is a separate design change, out of scope here, to be proposed upstream as its own issue if wanted. * no-mistakes(test): Fix herdr flake: don't misread transient empty foreground as unreadable * no-mistakes(document): Add fm-agent-process-lib.sh to scripts inventory * no-mistakes(fix): update remote herdr fixture to the real pane process-info shape The shared remote-secondmate herdr fixture still returned the old flat process-info body ({"result":{"process":{"name":...}}}). The process-level liveness classifier added for #4115 requires the real {"result":{"type":"pane_process_info","process_info":{...foreground_processes}}} shape and treated the old body as unreadable, so an already-launched remote endpoint's agent-state read failed and any relaunch attempt against it died with "remote endpoint state is unreadable; refusing duplicate launch" instead of reaching the state it was actually exercising (tests/fm-remote-secondmate-parent-binding.test.sh, tests/fm-remote-secondmate-lifecycle-e2e.test.sh). * no-mistakes(review): test: add empty-foreground regression test for herdr flake fix * no-mistakes(document): docs: register new stale-registration live-e2e test in herdr entry points * feat: add live-head merge gates and away task grants (#4199) * Bind GitHub merges to a live green head and require an away-task grant. A GitHub merge now re-reads the pull request and passes --match-head-commit, so a red or moved head cannot land the way GitLab already refused. While an away record exists, only yolo or a named grant may merge, so hold-for-return cannot ship an ungated PR. Co-authored-by: Cursor <cursoragent@cursor.com> * no-mistakes(review): Harden away merge authorization and grant parsing * no-mistakes(review): Restrict fallback outcomes to proved GitHub merges * no-mistakes(document): Refresh merge safety documentation * no-mistakes(ci): Fixed all three CI failures by updating legacy GitHub merge fixtures for live-head verification/direct gh merges and removing a process-event runner cleanup race. Verified fm-pr-check-security, fm-captain-hold-lifecycle, and fm-watch-triage pass locally; shell syntax and git diff checks also pass * no-mistakes(document): Document attended red-check exception --------- Co-authored-by: Cursor <cursoragent@cursor.com> * fix(bin): bound the Claude turn-end re-block against a frozen auto-arm epoch (#4221) The --claude guard's re-block budget charged the auto-arm ledger epoch, not the re-block: `budget_account_current_epoch` advanced the session count only when `state/.claude-autoarm-epoch` named a different generation than the previous accounting. The epoch advances only inside the auto-arm hook's generation claim, so a hook kept inert before that claim - a session lock held by a live harness outside its ancestry, a hook that never fires, or an identity or write failure ahead of `fm_autoarm_claim_next` - left the ledger frozen at its last outcome and the count frozen with it. Reproduced in a fixture: twelve consecutive Stops re-blocked with the count at 0 and the attended fail-open never fired, leaving only Claude's silent 8-block override, the blind end the bounded alarm exists to prevent. The budget now charges a re-block against an epoch the previous re-block already charged, while still charging each epoch at most once per Stop so the wait loop's repeated observations of one fresh terminal outcome and the same invocation's block decision cannot double count. The advancing-epoch progression is unchanged: three re-blocks, then one attended fail-open for a verified failure episode, and a frozen epoch now follows the same shape. Budget exhaustion without a verified failure still blocks, by the existing contract, and positive watcher recovery still clears the whole episode. Regression coverage drives the real auto-arm hook against a foreign session lock holder, asserts the ledger itself stays frozen, and fails before the fix in both the verified and unverified shapes; the existing unverified budget test now proves its budget actually ran out. * feat(herdr): add guarded foreground viewer for live validation (#4242) * feat(herdr): attach a real foreground viewer so the live-client teardown cases can be driven PR #4131 gated the Herdr active-tab close refusal on a live foreground client instead of the persisted `.focused` pointer, but only its two detached scenarios could be validated live. Every pseudo-terminal the runner built started at a zero-sized window grid, so Herdr registered no foreground client and `terminal title clear` kept answering `no_foreground_client`, leaving the four attached-client scenarios untested. That was a harness limit, not a product one. Add `fm-herdr-lab.sh viewer start|stop <session>`, backed by `bin/fm-herdr-lab-viewer.py`. The launcher sets the pty window size on the master fd BEFORE the fork, so the TUI cannot read the grid until it is already non-zero, and scrubs the inherited `HERDR_*` variables so Herdr's nested-viewer refusal does not fire when the helper runs inside one of its own panes. Attach and detach are both confirmed against the session's own foreground-client reason rather than assumed from a signal. The viewer inherits the lab's isolation contract: it attaches only to a session carrying this lab's ownership tripwire, never to `default`, and it signals only the processes it recorded, so a client someone else attached is never touched. Teardown now refuses while an owned viewer is still attached. Turn the reproduction into the regression with `tests/fm-herdr-attached-viewer-live-e2e.test.sh`, which drives #4131's scenarios 3, 4, 5, and 7 live against real Herdr and asserts the close refusal fires. Scenarios 4 and 5 need a focus change at one exact product boundary, so a PATH shim performs the real `tab focus` when the close helper issues its planning `pane get`. Removing either half of the recipe from the launcher makes the guard fail with the same `no_foreground_client` symptom #4131 reported. * test(herdr): fail loudly when an attached-viewer fixture cannot be created The fixture helpers run inside command substitutions, where fail() exits only the subshell and leaves the script running with empty ids. Return non-zero instead and carry the message at each call site. * fix(herdr): stop the viewer launcher's kill timer from raising on an exited child The SIGALRM escalation called os.kill unguarded, so a viewer that exited during the grace window turned an ordinary shutdown into a traceback inside the signal handler. * docs: list the lab viewer's pty engine in the bin toolbelt * no-mistakes(review): Harden Herdr viewer ownership and live CI coverage * no-mistakes(review): Validate viewer startup timeout and process ownership * no-mistakes(document): Document Herdr viewer safety contracts * no-mistakes(review): Fix viewer timeout to two seconds * no-mistakes(review): Cancel timed-out viewers and fix PTY grid * no-mistakes(review): Serialize viewer transitions and verify process parentage * no-mistakes(review): Harden viewer ownership locks and deduplicate CI * no-mistakes(review): Release interrupted locks and preserve viewer escalation * no-mistakes(review): Remove viewer locks and cancel interrupted launches * no-mistakes(review): Close viewer launch signal races * no-mistakes(document): Document attached Herdr viewer regression * fix(bin): let nonvisual work proceed when lavish-axi is unavailable (#3766) * fix(bootstrap): allow nonvisual work without Lavish * no-mistakes(review): Gate scout brief Lavish line on bootstrap version floor * test: isolate fixture Git config from host global and system settings (#3825) * fix(tests): isolate fixture Git configuration from host preferences Ignore global and system Git configuration in the shared test library, which all four fixture helper entry points source. Keep local config, command-line overrides and explicitly supplied test config usable without changing the caller's environment or real project signing preferences. Exercise global and system signing inputs through all four helpers, real fixture and child commits, explicit signing overrides, unchanged input files, and signing refusal outside fixture subprocesses. Verification evidence for issue #3770: On pristine upstream f09de8a3, all 12 reported suites failed and each logged "No secret key" using a private GIT_CONFIG_GLOBAL containing commit.gpgsign=true and gpg.format=openpgp, GIT_CONFIG_NOSYSTEM=1, and an empty private GNUPGHOME (GIT_CONFIG_COUNT and GIT_CONFIG_PARAMETERS unset). With this change, all 12 pass in the identical environment through bin/fm-test-run.sh --per-script-timeout-secs 900: fm-backlog-atomicity, fm-bootstrap-network-parallel, fm-bootstrap, fm-crew-state, fm-fleet-sync, fm-gate-refuse, fm-grok-harness, fm-session-start, fm-sessionstart-nudge, fm-tangle-guard, fm-test-run, and fm-update (all tests/<name>.test.sh). The new fm-test-fixtures regression failed before the library change and passes after it. Canonical bin/fm-lint.sh passes. Additional verification exposed fm-teardown's herdr-preflight-missing-adapter assertion on both this branch and an unchanged f09de8a3 archive with signing neutralized. That pre-existing failure needs separate disposition; it is not repaired or skipped here. The separately owned Muse and composer fixture defects remain untouched. Fixes #3770 * no-mistakes(review): Complete fixture Git isolation and scope config assertions * no-mistakes(review): Share Git isolation across standalone fixture entry points * no-mistakes(review): Map git-config helper changes to lib.sh dependents * no-mistakes(review): Select fixture-isolation regression on runner change; halve config matrix * no-mistakes(review): Scope fixture-isolation regression selection to the runner alone * no-mistakes(document): Give fixture Git isolation helper its owning header * no-mistakes(document): Record fixture Git-isolation coverage in fixtures suite header * no-mistakes(review): Fix linked-worktree fixtures and remove redundant Git isolation * no-mistakes(document): Correct stale runner-selection documentation * no-mistakes(document): Clarify family antecedent in isolation-proof runner evidence * feat(bin): add config/claude-permission-mode to launch Claude workers in auto mode (#4239) * feat(spawn): add config/claude-permission-mode to launch Claude workers in auto mode Every Claude worker launched with --dangerously-skip-permissions, and a captain who refuses bypass mode had no way to select Claude Code's classifier-reviewed auto mode instead. A new one-token local config, config/claude-permission-mode, selects the permission flag for every Claude launch: absent or `bypass` keeps today's launch byte-for-byte, `auto` swaps in --permission-mode auto, and any other value refuses the spawn before any endpoint, worktree, or record exists and names the accepted values. fm-spawn resolves the file on every spawn and relaunch, threads the flag through the Claude launch template for crewmates, scouts, and secondmates alike, and records claude_permission_mode=auto in the task meta only under auto so the default meta stays unchanged; a relaunch re-resolves rather than preserving the line. The file is a captain-wide safety preference, so it joins the inherited local material pushed into secondmate homes. The Claude adapter reference records the verified auto launch shape on Claude Code 2.1.269 and that it never meets the once-per-machine bypass confirmation dialog; docs/configuration.md owns the schema. * no-mistakes(review): drop unread claude_permission_mode meta line and its assertions * fix(teardown): leave a Treehouse pool slot reassigned to another task untouched (#4243) * fix(teardown): refuse to return a Treehouse pool slot reassigned to another task A pool slot is reused across tasks, so a finished task's worktree= line can name a slot a different, live task now holds. Teardown already refused when a second task record named the same live path, but that scan cannot prove the record it is tearing down is the current owner: the task that took the slot next may leave no record the scan can reach - its own worker may have exited and its record been cleaned up, or it may live in a home this machine does not register. Teardown then killed every process under the path, hard-reset it and returned it, and its unlanded-work refusal never fired because it was inspecting a directory that no longer belonged to the task being torn down (observed 2026-09-07). Treehouse's own state file cannot answer the ownership question. It records a slot's owner as a live process lease (owner_pid plus owner_started_at, with `treehouse status` reporting in-use from the processes actually running under the path), which names no task and is released by the very event that makes a record stale - the worker exiting. An unleased slot therefore reads identical whether it is still this task's or has since been handed on, and a slot whose new holder has also exited but left uncommitted work reads as free. So the identity source is Firstmate's own claim, not Treehouse's lease. fm-spawn writes that claim - the task id - into the slot at the moment it takes it, under the same project lock that allocates the slot, and fm-teardown drops it only after the slot is genuinely returned. It lives at <pool>/<slot>/.fm-slot-owner, a sibling of the repo checkout rather than a file inside it, so claiming a slot can never dirty the copy the landed-work checks inspect. A claim naming another task, or one that cannot be read, refuses; --force does not lift either refusal, because --force authorizes discarding this task's unlanded work, never another task's live work. A slot that cannot be claimed refuses the spawn instead. An absent claim proceeds on exactly the record-scan protection it had before: slots taken before claims existed, and slots already returned, carry none, and refusing those would strand every task in flight across this change on no evidence at all. The refusal is deliberately all-or-nothing rather than partially completing the task's own cleanup. state/<id>.meta is the only durable record naming the worktree and endpoint, so removing it would destroy the evidence needed to reconcile which record is wrong, and its removal is one step with the backlog transition. Nothing is stranded: clearing the stale worktree= line leaves a record with no slot to release, which then tears down normally, and the refusal names that remedy. Repairing the previous claimant's stale worktree= line at spawn time is left for separate work. It would have the new owner write another task's record - the same class of cross-task mutation this bug is - and would need that record's own meta lock; with the claim in place teardown refuses on evidence rather than depending on the stale pointer having been scrubbed. For the same reason the relaunch path writes no claim: it holds no allocation lock, and a record whose worktree= is already stale would stamp the wrong task's claim onto a live sibling's slot. The regression reproduces the reuse sequence with only one discoverable record, including a clean, fully landed ship copy torn down without --force - the shape of the real incident, which the previous code returned to the pool - and fails against the previous code; the existing two-record, cross-home, own-slot and no-claim cases still pass unchanged. This builds ON upstream b028e8b1 (#3837), which is already in this branch's base (origin/main 40c50ea8) and owns the record-exclusivity scan. Nothing here replaces that scan; the claim is the positive proof it cannot supply. Claude-Session: https://claude.ai/code/session_01JTBmuqKugaPUj7k9TXQwFS * no-mistakes(review): teardown leaves reassigned slot; spawn abort drops claim * no-mistakes(review): narrow Treehouse lease evidence; gate abort claim release on lock * no-mistakes(review): pin spawn-side slot claim; narrow abort-release header * no-mistakes(document): docs: point slot-claim rationale at fm-wake-lib owner * docs(AGENTS): keep brief-fill from widening the captain's ask (#4247) The reviewer treats Captain's intent as acceptance criteria, so a widened ask there drives over-built work; the spec should carry only what the ask requires. * fix: identify underway tasks and sort charted work (#4245) * feat(bearings): name the Underway rows and order Charted Next newest filed first The fleet board's Underway rows led with the run status alone, so a scan told the captain where a pipeline stood but never which task the row was, and Charted Next rendered in backlog order rather than by when work was filed. The snapshot now projects the durable task name onto every in_flight row - from this home's backlog title, and from a secondmate home's own ledger for an active child - and the durable filed date onto every gate. The board's Underway row leads with that name and keeps the run status on its second line, and Charted Next renders newest filed first, with rows carrying no comparable date keeping their payload order after every dated row. The payload validator requires an explicit name marker on every Underway row and refuses a filed value that is not an ISO date, so the board can never sort on garbage or invent a label. * no-mistakes(review): Fix Bearings labels, bounds, and filed validation * no-mistakes(review): Fix Bearings identifiers and eligible queue bounds * no-mistakes(document): Document Bearings labels and newest-first bounds * no-mistakes(ci): Updated the stock macOS Bash CI expectation from 56 to 59 Bearings tests. Verified the suite under /bin/bash 3.2: all 59 tests pass. git diff --check also passes * fix(bearings): surface return catch-up without blocking snapshots (#4248) * fix(bearings): report the away-return catch-up instead of refusing A captain returning from away and asking for bearings got zero bytes and an error: fm-bearings-snapshot.sh ran the away-return guard with `|| exit $?` before reading any fleet state, so the mere existence of the catch-up gate killed every bearings mode (and /ahoy with them). Bearings now consults that guard rather than obeying it. fm-afk-return.sh separates its two refusal branches by exit status, so an ACTIVE away window still refuses exactly as before - the right answer there is to run the return first - while return catch-up (exit 4) lets collection and projection proceed and is disclosed as one action-free `(return-catchup)` gate row, following the existing `(main-inventory)` precedent. It stays out of decisions_open: these blockers are firstmate-actionable, not the captain's own call, and the per-task blockers already project as their own Underway rows. The guard's refusal text also stops promising a blocker list it cannot produce: a gate retained for a lifecycle reason alone now names that retention reason, and bearings carries the same reason in the gate row's title. Reporting is not ordinary work. AGENTS.md already scopes the return hold to work rather than reporting, so only the /afk and bearings skills needed the correction. * no-mistakes(document): Refresh away-return Bearings verification * no-mistakes(review): Reserve catch-up gate outside Bearings truncation * no-mistakes(review): Preserve filed dates in catch-up gate output * no-mistakes(document): Document reserved catch-up gate projection * fix(bin): address the home's backlog from any directory and detect a forked code-root copy (#4223) * fix(backlog): address the home's backlog from any directory and detect a forked code-root copy A home outside the code root forks its queue: the tracked .tasks.toml names data/backlog.md relative to tasks-axi's working directory, so a bare tasks-axi call from the code root writes the code root's data/ while session start, spawn, and teardown use $FM_HOME/data. Linking the code-root copy into the home does not hold, because tasks-axi 0.2.4 writes by renaming a temp file over its target and rename(2) replaces a symlink: add, start, hold, and done from the code root each turn the link back into a regular file. The archive path is resolved against the working directory too, even with --file. bin/fm-tasks-axi.sh runs tasks-axi against this home's backlog from any directory, using the lifecycle transitions' existing addressing (run from the data directory's parent, pin <data>/backlog.md through TASKS_AXI_FILE). It keeps relative --to/--*-file arguments meaning the caller's paths, and refuses a caller --file, an unresolvable home, and a symlinked home backlog. The fm-send hold lookup, fm-public-followup, and the fm-decision-hold shim, which relied on cwd discovery, now go through it with an explicit FM_HOME and a cleared data override, so they keep addressing exactly $FM_HOME/data and an ambient TASKS_AXI_FILE cannot divert them; every agent-facing backlog command names it instead of bare tasks-axi. Bootstrap gains a detect-only BACKLOG_RECONCILE check, also run read-only: when the home's data directory is not the code root's, a code-root data/backlog.md or data/done-archive.md that is not the home's own file is reported as a fork, with the merge procedure in bootstrap-diagnostics. * test(teardown): assert the completion hint names bin/fm-tasks-axi.sh ready The completion hint now points at the home-addressed command instead of a bare tasks-axi call, so the dependency-cleared follow-up assertion checks for that command. * no-mistakes(test): clear ambient tasks-axi env in tests/lib.sh * no-mistakes(document): drop bare tasks-axi example from cd-guard doc * no-mistakes(lint): replace ls -A decoy listing with find for SC2012 * no-mistakes: apply CI fixes * revert: keep the compliance gate unchanged; the synchronize race is filed separately * fix: pre-register Claude trust for secondmate homes (#4262) * fix(spawn): pre-register Claude workspace trust for secondmate homes A claude --secondmate launch skipped workspace-trust registration entirely, so a standalone-clone secondmate home (an explicit ~/fm-homes/<id> path) had no store entry and its pane wedged on the "Is this a project you trust?" dialog before it read its charter. The step was gated on the task kind rather than on the harness, so the spawn's fail-closed guard had nothing to run against and reported a launch that could never start work. fm-claude-trust.sh gains a secondmate-home mode. A secondmate home is a whole firstmate instance, produced either as a leased worktree or as a standalone clone, so the linked-worktree test cannot decide it and the seed is the evidence instead: the .fm-secondmate-home marker must be a regular file this user owns naming exactly the id being spawned, the home must hold AGENTS.md and bin/, and each operational directory must resolve inside the home. That is the set fm-home-seed.sh writes and fm-spawn.sh's own home validation re-checks, so nothing wider than a home a secondmate spawn would launch into can earn home-level trust. The worktree path is unchanged, and still refuses a home. fm-spawn.sh now runs the registration for every claude launch and keeps refusing the spawn when it fails, rather than launching an agent that would wedge. * no-mistakes(document): Correct Claude secondmate trust guidance * fix: ignore superseded failed GitHub check runs (#4258) * fix(pr-merge): judge each required check by its current run When the base branch advances, GitHub cancels a pull request's in-flight run and re-triggers it. The cancelled run stays in statusCheckRollup beside the passing re-run, so the rollup can hold several runs of one check name at the same head while GitHub itself reports the pull request CLEAN. github_checks_not_green judged every run independently, so that superseded failure refused a genuinely mergeable pull request and pushed the operator toward a needless --allow-red. Group the rollup by the reported name and judge each check by its current run. Supersession is proven, never assumed: a name leaves the red set only when every one of its non-green runs is strictly older than one of its green runs, dated by the forge's own settled timestamp - a check run's completedAt once its status is COMPLETED, or a status context's createdAt - and only in the whole-second UTC form GitHub emits, which is the one spelling that orders correctly as plain text. A run with no such timestamp is never superseded, so a still-running, queued or undated run keeps its check red, and a name with no green run at all stays red. An unnamed entry is grouped alone so two unrelated unnamed checks are never treated as one. Every comparison is one-directional: it can only clear a failure a later success provably replaced, and never clears a check whose current run failed, is pending, or is missing. No other guard moves - the pull request must still be open, undrafted, mergeable, conflict-free and head-bound, and --allow-red still waives exactly its named check with every other check green. Live reproduction: PR #4224 read CLEAN with an old FAILURE and a newer SUCCESS for one check name and was refused; it now verifies, while #4208 and #4210, whose latest runs failed, still refuse. * no-mistakes(review): Use check-run start times for safe supersession * no-mistakes(document): Clarify GitHub check-rollup documentation * fix(bin): persist merge authority for poll-detected outcomes (#4266) * fix(merge): persist the merge authority on poll-detected merge outcomes The merge ledger tags a merge with the authority that permitted it while the away-posture record existed, but only the direct attended merge in bin/fm-pr-merge.sh recorded it. A merge the forge queued, or one the merge poll detected after the fact, published an untagged row, so exactly the merges no agent watched were the least auditable. bin/fm-merge-authority-lib.sh now owns that answer, read from the same structured sources the merge gate already used: the task's recorded yolo posture and the away-posture record's mechanical grant list, never prose. bin/fm-pr-merge.sh keeps its own refusal wording and gates on that answer; bin/fm-watch.sh only records it on the row its poll publishes, so reading the authority never becomes a second path to a merge. An unresolved answer records an untagged row rather than dropping the outcome or inventing an authority. * no-mistakes(review): Persist canonical merge authority for queued poll outcomes * no-mistakes(review): Harden merge authority persistence against lifecycle races * no-mistakes(review): Serialize poll authority publication with teardown * no-mistakes(document): Clarify persisted merge authority lifecycle * no-mistakes(ci): Added targeted SC2034 suppressions for the two public result assignments in bin/fm-merge-authority-lib.sh. Verified successfully with `CI=true bin/fm-lint.sh` * ci: supersede superseded PR CI and bound unbounded jobs (#4281) The 2026-09-12 Actions starvation incident found firstmate CI with no concurrency deduplication, so every superseded PR head kept its full 13-job fan-out, and four jobs with no timeout at all. Add per-PR supersession keyed on the PR number for pull_request events and on the unique run id for push events, cancelling only pull_request runs, so a new PR head replaces its own in-flight CI while every main push keeps its own group and is never cancelled. Add hang tripwires to the four previously unbounded jobs: 25 minutes for lint (measured at 14-16 minutes) and 5 minutes each for the coverage guard, the timing aggregate, and the repo invariants. Measured lane bounds are unchanged. tests/fm-ci-workflow.test.sh resolves the workflow's concurrency expressions against simulated pull_request and push contexts and holds every job's finite timeout. * test(watch): gate backlog-hold away-record fixture on tasks-axi (#4288) Every other make_hold_home caller in this file skips when tasks-axi is absent; this test was the one unguarded call, so hosts without tasks-axi hard-fail the fixture build instead of skipping. * fix(backlog): bound per-item backlog row reads so a wedged backend cannot blind a session start (#4027) * fix(bin): bound each backlog row read so one wedged backend cannot blind a session start bin/fm-bootstrap.sh's reconcile and close-replay sweeps read the backlog backend once per item through fm_backlog_row_show, and that read was unbounded. A single wedged `tasks-axi show` therefore consumed the whole FM_SESSION_START_TIMEOUT and truncated the digest before the wake queue, supervision instructions, fleet state, and context sections ever printed, leaving the fleet unsupervised with no live watcher. The harm was a blind startup, not a slow one. Bound the read with the existing shared timeout primitive (bin/fm-timeout-lib.sh), so a wedged backend degrades to a loud partial reconcile: the sweep's existing BACKLOG_RECONCILE diagnostic names the item it could not read and the loop continues to the next one. The first bound hit also latches FM_BACKLOG_ROW_SHOW_WEDGED, so a sweep over many items pays one bound rather than one per item and still names every item it skipped, which is what keeps the digest whole on a home carrying a large fleet. The bound holds regardless of any particular tasks-axi install, so it does not depend on the 0.2.5 `show` hang being resolved separately. * fix(bin): set the wedged-backend latch where it survives, and prove it The latch added with the read bound was inert. fm_backlog_row_show runs inside a command substitution in both of its status-capturing callers, so the subshell read the inherited value correctly but its write died with the subshell. Every item still paid a full bound and reported `exceeded`, never `skipped`, which left the large-fleet case the latch existed to cover completely uncovered. Move the write to the two callers that capture the read's status and own the surviving shell, and leave fm_backlog_row_show reading the latch only. Correct the comments that claimed an ownership the function never had. The test that was supposed to cover this asserted only that the second read finished under a generous ceiling, which is true whether or not the latch works. Assert instead that a latched read is strictly faster than one bound and that it reports its own item as skipped, so an inert latch fails the test. * test: cover every item the wedged-backend latch skips The latch assertion exercised a single skipped item, so "every skipped item is still named" was inferred rather than tested. Probe three items instead and assert each skipped one names itself and costs less than a bound. Verified as a real guard by removing both latch writes: the suite then fails on the first skipped item instead of passing. * no-mistakes(review): distinguish backlog read-bound hits from absent rows * no-mistakes(review): preserve read-bound status through the captain verify gates * no-mistakes(review): Preserve backlog read-bound hits through resolve_entry and reconcile instead of spending them as absent rows * no-mistakes(review): Preserve backlog read-bound 124 through migrated-prefix scan and remaining task_show call sites * no-mistakes(document): Document bounded backlog row reads and FM_BACKLOG_ROW_TIMEOUT_SECS * no-mistakes(ci): Fixed all four failing CI checks with one root-cause fix plus one test-heredity fix. (1) bin/fm-captain-hold.sh: task_show carries the row in TASK_SHOW_OUTPUT and emits no stdout, but four call sites still used the stale command-substitution convention show=$(task_show ...), leaving show empty: task_show_or_fail (every captain hold failed with 'did not retain its hold-set stamp' - broke fm-captain-hold-lifecycle in parallel 1 and fm-bearings-board in serial 3), resolve_migrated_entry (migrated-prefix resolution could never match), reconcile-requests (existing rows were refused as absent), and command_open --identity (printed a constant '#0' identity, so fm-watch-triage's re-held captain call inherited the previous call's silence in serial 1). This is also the Greptile P1. Fixed by invoking task_show in the current shell and reading show=$TASK_SHOW_OUTPUT, the convention the other eight call sites already use; read-bound hits still stop loudly by name. (2) tests/fm-backlog-read-bound.test.sh (serial 4, unclassified family): the new e2e half implicitly relied on the author's process tree containing a harness process so fm-lock.sh would grant the fleet lock; on CI runners the lock is refused, the reconcile sweep is skipped, and the final BACKLOG_RECONCILE assertion fails. Reproduced by simulating a CI ancestry via a ps shim, fixed by pinning the lock evidence with the established fake-ps harness fixture pattern from tests/fm-session-start.test.sh. Verified: shellcheck clean; parallel-1, serial-3, and serial-4 lanes fully green locally (failed=0); serial-1 lane green except fm-gemini-harness, which fails only under local Node v26 (comm=node-MainThread); CI's default Node 22 reports comm=node, the branch that test passes on, so it is not a CI failure * no-mistakes(document): Verified bounded backlog read docs accurate across branch * fix(merge): serialize away authority with synchronous merges (#4285) * fix(merge): serialize the away-authority check with a synchronous merge bin/fm-pr-merge.sh read the away-posture record for merge authority (the per-task merge grant and the yolo/away-grant decision) and handed the merge to the forge afterwards. An archive at the captain's return or a grant revoked by a replacement record could land in between, so a merge could proceed on away authority that no longer held. The away record now carries a cross-subsystem lock, built on the existing bounded lock primitive rather than a new lock format: the record-mutating subcommands hold it across their mutation, and the merge holds it across both its authority read and the forge command. Because a queued or auto merge returns before the pull request lands, and would therefore outlive the lock, an away merge is now refused whenever it could land asynchronously: a requested --auto, a base branch whose merge-queue state does not prove an immediate merge, and GitLab's asynchronous flags and configuration. What remains permitted while away is the synchronous merge that lands inside the lock. This closes the common away-record/merge race against a live lock owner. It does not make the merge atomic in every case, and two narrow races are accepted and documented at their sites rather than hidden, both confused-agent-grade in the sense bin/fm-lease-lib.sh already uses: - A merge-queue rule change or a PR base change in the window between the queue-free preflight and the forge call can still enqueue the merge, which can then land after its grant lapses. - Killing the lock-owning shell while its gh or glab child is still running lets stale-owner recovery reclaim the lock and the record be archived or replaced, after which the orphaned child can complete the merge on lapsed authority. Closing either one needs landing verification or an ownership handoff, which is deliberately out of scope here. No existing gate is relaxed. The lock is taken after the live green-at-head verify and the captain-hold check, the in-lock authority read is unchanged, and a lock that cannot be taken refuses the merge rather than proceeding unlocked. The away grant stays a structured field; no prose is parsed. * no-mistakes(review): Fix GitHub rollup fixture base branch * no-mistakes(document): Document atomic away-authority merge locking * no-mistakes(ci): Updated two executable GitHub API fixtures to include the required baseRefName. Both previously failing test suites now pass: fm-captain-hold-lifecycle.test.sh and fm-pr-check-security.test.sh. git diff --check also passes * feat(bin): add Antigravity CLI (agy) as third worker/scout adapter (#4200) * feat(agy): verify Antigravity CLI as third worker/scout adapter Detection by anchored ancestry in fm-harness.sh (no marker of its own); bootstrap harness and effort validation; launch template with model and effort mapping plus reachable-catalog model validation; rendered-tail busy fallback in fm-busy-lib.sh with delivery footer in fm-composer-lib.sh; control mechanics with crewmate/scout-only refusal; tmux liveness naming; router entry with concise adapter reference; dated verification record; portable regression plus opt-in live drift guard. Verified live on agy 1.2.0: supervised spawn, durable steering, same-copy relaunch, and exit, with Herdr-native busy agreement. * no-mistakes(review): bound agy model probe, gate trust dialog, narrow busy signature * no-mistakes(review): pre-register agy workspace trust, make readiness gate strict * no-mistakes(review): Close Orca terminal on gate failure; isolate live-guard HOME; tighten agy matching * no-mistakes(document): Document agy adapter in stale harness enumerations * no-mistakes(review): Clamp non-positive FM_AGY_MODELS_TIMEOUT to the default bound * no-mistakes(document): Fix stale test-shard snapshots after agy lane additions * no-mistakes(ci): Fixed ci-3 (tests/fm-agy-harness.test.sh:519). Root cause: the agy spawn fixture's default base PATH (/usr/bin:/bin:/usr/sbin:/sbin) omits node's directory, but the spawn drives the real bin/fm-agy-trust.sh (which hard-requires node to record trust) and the fixture's fake tmux trust lookup (node -e) under that PATH. On the ubuntu-latest CI runner node lives in the toolcache (/usr/local/bin), so trust pre-registration failed on portable serial 2; on typical Arch hosts node is in /usr/bin, masking the defect. Fix (smallest, following the existing tests/fm-kimi-harness.test.sh precedent of carrying the interpreter's resolved directory): resolve node from the invoking environment (failing the test with 'test needs node' if absent, as kimi does f…
friesentius
pushed a commit
to friesentius/firstmate
that referenced
this pull request
Sep 21, 2026
kunchenguid#4033) * fix(bin): keep an escalated undelivered handoff wake retryable A remote backlog handoff holds its outbox until the backlog receipt and the receiver wake are both confirmed, and retries the wake under the same pending-reply correlation on every resume. When that wake's remote transport was lost, the correlation stayed undelivered in delivery_unknown and the watcher's next pending-reply tick escalated it. Both the reuse predicate and the known-undelivered reset refused an escalated record, so the resume refused to resend the wake forever and every later handoff to that mate jammed behind the outbox. Treat an escalated record with no confirmed delivery as the undelivered correlation it is: fm_pending_reply_corr_reusable accepts it for its own task and fm_pending_reply_reset_known_undelivered returns it to awaiting_report for the idempotent remote resend, while a delivered record is still never reset and a missed-report escalation keeps its meaning. The published delivery-unknown decision stays open until the record resolves, so a repeat loss neither re-notifies nor strands it. Reproduce the deadlock end to end in the remote handoff test (lost wake transport, watcher escalation, resume) and pin the predicate contract in the pending-reply suite; the fm-send fixture that pinned the refusal now uses a genuinely stale delivered escalation. * no-mistakes(review): Decouple durable outboxes from best-effort wake retries * no-mistakes(review): Align handoff documentation with durable receipt release policy * no-mistakes(review): Handle unrecordable wake state as dropped * no-mistakes(review): Prevent stale wake markers blocking handoffs * no-mistakes(review): Prevent stale delivered markers suppressing new wakes * no-mistakes(document): Clarify retry escalation decision lifecycle * no-mistakes(document): Document pending receiver wake retries
friesentius
pushed a commit
to friesentius/firstmate
that referenced
this pull request
Sep 21, 2026
kunchenguid#4033) * fix(bin): keep an escalated undelivered handoff wake retryable A remote backlog handoff holds its outbox until the backlog receipt and the receiver wake are both confirmed, and retries the wake under the same pending-reply correlation on every resume. When that wake's remote transport was lost, the correlation stayed undelivered in delivery_unknown and the watcher's next pending-reply tick escalated it. Both the reuse predicate and the known-undelivered reset refused an escalated record, so the resume refused to resend the wake forever and every later handoff to that mate jammed behind the outbox. Treat an escalated record with no confirmed delivery as the undelivered correlation it is: fm_pending_reply_corr_reusable accepts it for its own task and fm_pending_reply_reset_known_undelivered returns it to awaiting_report for the idempotent remote resend, while a delivered record is still never reset and a missed-report escalation keeps its meaning. The published delivery-unknown decision stays open until the record resolves, so a repeat loss neither re-notifies nor strands it. Reproduce the deadlock end to end in the remote handoff test (lost wake transport, watcher escalation, resume) and pin the predicate contract in the pending-reply suite; the fm-send fixture that pinned the refusal now uses a genuinely stale delivered escalation. * no-mistakes(review): Decouple durable outboxes from best-effort wake retries * no-mistakes(review): Align handoff documentation with durable receipt release policy * no-mistakes(review): Handle unrecordable wake state as dropped * no-mistakes(review): Prevent stale wake markers blocking handoffs * no-mistakes(review): Prevent stale delivered markers suppressing new wakes * no-mistakes(document): Clarify retry escalation decision lifecycle * no-mistakes(document): Document pending receiver wake retries
sctru
added a commit
to sctru/firstmate
that referenced
this pull request
Sep 29, 2026
* feat(bin): add IMAP/SMTP mail plane with standing poll (#3765)
Opt-in IMAP/SMTP mail plane (fm-mail.sh / fm-mail-check.sh). Absent FM_MAIL_* stays off.
Speaking as Kun's firstmate: this is merged. Thank you @feilipu — really appreciate you taking the time on this.
* fix: launch remote Herdr through the user login shell (#4061)
* fix(remote): start the fm-remote Herdr agent through a login shell
Launchd was exec-ing herdr directly, so the Aqua agent inherited a background session without login-keychain access. Start it via /bin/zsh -lc exec so panes keep login env and can refresh OAuth tokens after reboot.
* fix(remote): start fm-remote Herdr via the account login shell
Resolve UserShell from Directory Services and invoke it with separate -l and -c so bash, fish, and zsh all get login-keychain access. Fall back to SHELL, then /bin/zsh, then /bin/sh without failing the render.
* no-mistakes(review): Fix launch-agent shell fallback resolution
* no-mistakes(review): Preserve and escape Directory Services shell paths
* no-mistakes(document): Document login-shell LaunchAgent behavior
* no-mistakes(ci): Updated the trusted fm-remote-doctor SHA-256 identity in bin/fm-remote-entrypoint.sh. Verified with tests/fm-on.test.sh, tests/fm-remote-doctor.test.sh, bash syntax checks, and git diff --check
* no-mistakes(ci): Resolved the login shell exactly once per doctor invocation and threaded it through plist rendering, installed/loaded contract validation, repair reporting, and post-repair checks. Added a regression test proving repeated repair remains healthy and performs no reload when a hypothetical second Directory Services lookup would differ. Updated the trusted doctor hash. Verified doctor, fm-on, remote-entrypoint, lint tests, ShellCheck, syntax, and diff checks
* no-mistakes(ci): Made Darwin shell resolution hermetic with executable injection and a 2-second Directory Services timeout. Updated tests to inject shells by default, isolate dscl-specific cases, parse plists semantically, and verify stalled dscl fallback. Updated the trusted doctor hash. Doctor, fm-on, entrypoint, syntax, hash, and diff checks pass
* no-mistakes(ci): Raised portable serial CI timeout from 20 to 30 minutes, refreshed the specified timing hints, added missing hints, and recomputed shard documentation. Verified coverage, runner behavior tests, workflow lint tests, shell syntax, requested timing maxima, and diff checks
* fix: distinguish landed deliveries from resolved captain calls (#3710)
* fix(bearings): keep captain-approved deliveries in Recently Landed
A closed task is never held: tasks-axi clears the held flag when a task
closes and keeps hold-kind and the hold reason as the record of the call
that was made. Recently Landed excluded every Done row whose hold-kind was
captain, so the marker it treated as "closed while still waiting on the
captain" was in fact the proof that the captain had approved the work. Every
merge routed through a captain decision disappeared from the list of what
shipped, including under --all-landed.
The selector now asks whether the closed row delivered something. Recently
Landed is merged PRs, completed scouts, and finished local-only merges, so a
row carrying one of those artifacts belongs there whoever approved it. A
captain question closes with an answer and no artifact of its own, and that
is what still stays out, so an answered question is never rendered as
shipped work.
The same rule was written twice - the bearings projection selects this
home's Done rows and the fleet snapshot selects each secondmate home's Done
rows into the roll-up the same section merges in - which is why one defect
hid deliveries in every home. Both now share bin/fm-landed-lib.sh.
* fix(review): Normalize landed evidence and exclude answered captain questions
* fix(review): Normalize captain delivery evidence across relocated data
* fix(review): Record authoritative delivery provenance with legacy fallback
* fix(review): Harden delivery provenance across forced and pruned completions
* fix(review): Replace premature merge closure with existing release contract
* fix(review): Document provenance-based Recently Landed selection
* fix(document): Align documentation with completion provenance
* fix(lint): Fix targeted ShellCheck warnings
* fix(ci): order the pinned tasks-axi install before its stock-Bash consumers
In `.github/workflows/ci.yml` the pinned tasks-axi install now precedes both
stock-Bash consumers, and the Bearings expectation is updated from 49 to 50
tests.
Verified with macOS Bash 3.2: snapshot 16/16, Bearings 50/50, public-followup
1/1. Full repository lint and all three workflow validations pass, and
`git diff --check` is clean.
* fix(review): Make completion provenance unambiguous
* fix(review): Make completion verdict authoritative over quoted provenance
* fix(review): Preserve retained artifacts through resumed captain closes
* fix(review): Unify completion provenance ordering across writer and reader
* fix(review): Preserve artifacts across failed captain closes
* fix(review): Refresh v1 assertions; provenance authority remains unresolved
* fix(review): Remove unreliable provenance while preserving landed deliveries
* fix(review): Reject stale home summaries visibly
* fix(review): Restore retained deliverable recording
* fix(review): Match landed artifacts and restore retention documentation
* fix(review): Disambiguate captain calls and restore landed artifact matching
* fix(review): Persist retained report and PR artifacts
* fix(review): Preserve staged artifacts before captain answers
* fix(review): Avoid wedging answers on unsupported report paths
* fix(review): Exclude unreleased captain-held pull requests
* fix(review): Exclude held local-only answers from landed
* fix(review): Preserve retained scout reports across snapshot rendering
* fix(document): Align landed lifecycle documentation with release semantics
* fix(review): Enforce landed artifact-kind ownership
* fix(review): Infer canonical task kinds in snapshots
* fix(review): Require captain-hold release before merges
* fix(review): Qualify merge lifecycle regression evidence
* fix(review): Serialize captain holds with merge operations
* fix(review): Document merge cleanup residuals honestly
* fix(test): Replace vacuous Bearings regression with behavioral cases
* fix(document): Align Bearings verification and merge lifecycle documentation
* fix(review): Serialize merges and exclude captain calls from landed
* fix(review): Harden merge identity and landed selection
* fix(document): Clarify landed selector compatibility filtering
* fix(bin): keep merge entrypoints usable on records without an incarnation
The merge identity guard refused any task record with no spawn_gen field.
That field identifies one exact incarnation, so comparing it across the wait
for the merge lock is what catches a task relaunched while the merge was
queued. Requiring it to be present is a different rule, and it refused every
record written before the field existed: a legacy task could no longer be
merged at all, and five behaviour suites refused before reaching the check
they were written to exercise.
The comparison only needs to notice a change. An absent field is now read as
an empty incarnation and compared like any other value, so a record that
gains, loses, or alters one is still refused, while a record that simply
predates the field merges. An ambiguous or unreadable field stays an error,
because a record that cannot name one incarnation cannot be compared. The
missing-record message each entrypoint had before the guard is restored, so
a genuinely absent record still says so in its own words.
The role partition now precedes reading the record. Refusing the supervision
branch is a statement about the actor, not about the task, so it cannot
depend on a record the wrong actor may not have.
A backlog file that does not exist meant "no longer an open captain call".
For a caller that asked to tell absence apart it now means absent, so a board
card whose home carries no backlog stays visible instead of being dropped as
resolved.
Fixture repositories pin their initial branch instead of inheriting
init.defaultBranch, which resolved to main on a developer machine and master
on a runner, so a fixture naming main failed only in CI.
* fix(review): read local-only note from body; surface pending-close failures
* fix(review): keep kindless local-only landings in Recently Landed
* fix(review): bind local-only note scan to the tasks-axi note line
* fix(review): Guard unavailable captain-hold authority records
* fix(document): Document unreadable authority predicate outcome
* fix(bin): read an absent backlog as absence, not an unreadable record
The merge gate refused every task whose home carries no backlog file. A
backlog that does not exist holds no captain call, so nothing can be held and
the merge is safe; only a backlog that exists and cannot be read may hide a
live hold. Those two states were collapsed into one refusal, which stopped
merges in any home that keeps no backlog.
The predicate now reports a missing backlog file as absence, alongside a row
the backlog does not carry. A record that exists but cannot be read still
leaves by the existing cannot-tell path, which both merge entrypoints already
refuse, so the restrictive direction is unchanged.
That leaves no way to reach the separate unavailable-record result, so the
result and the two branches that handled it are removed rather than left
describing an outcome that can no longer occur. The lifecycle documentation
loses the same claim.
Regressions cover both directions in each entrypoint: a home with a task
record and no backlog merges, and a backlog present but unreadable refuses
without reaching the forge.
* fix(review): Fail closed unreadable backend configuration
* fix(tests): pin the bare origin's initial branch in the remote seed fixture
The fixture created its bare origin with no initial branch, so that
repository's HEAD followed init.defaultBranch while the source repository
pushed the branch fm_git_init_commit pins. On a host that still defaults to
master the two disagreed: the bare origin's HEAD named a branch the push never
created, cloning it warned that the remote HEAD referred to a nonexistent ref
and checked out nothing, and the seed assertion for the cloned README failed.
A machine whose default is already main paired the two by accident and hid it,
which is why the fixture passed locally and failed on the runner.
Pinning the bare origin to the same branch removes the dependency on the
ambient default from both sides. Verified under both conditions: with
init.defaultBranch set to master, and set to main, the suite passes 26 of 26.
* fix(review): Fail closed unreadable user backend configuration
* fix(bin): republish the home summary as v1 and record two load-bearing rules
The published home-summary schema had moved to v3, which routed every
secondmate home still emitting the earlier version to the stale branch: their
landed rows, open decisions and holds all came back empty and their state read
as unknown until each home was updated. The payload never justified that. Its
field set, field order, truncations and the landed array construction are
byte-identical to v1, so only which rows the selector places in landed
differs, and a v1 consumer reads that the same way.
Republishing as v1 removes the rollout regression and, with it, the tolerance
machinery that existed only to soften the bump: the stale-schema predicate,
its two collection branches, the flag and its provenance branch, the omitted
surface that can no longer be reached, and the fixtures and assertions that
covered them.
Two rules that a scope review proposed removing are kept, each now carrying
the reason it exists, because both were measured to be load-bearing:
The artifact-kind ownership clause is what keeps an explicit scout that
recorded no report out of Recently Landed. Without it such a row has none of
the three artifacts, satisfies the compatibility fallback and renders as
shipped work with an empty artifact.
The kind fallback is needed because tasks-axi omits the kind metadata
entirely when a title begins with a canonical keyword. Without it a scout
titled "SCOUT ..." reports no kind, its recorded report stops counting as a
delivery, and it drops out of the section this selector exists to repair.
* fix(bin): move the scout guard note onto the rule and drop two dead pieces
The LOAD-BEARING note sat on an unreachable branch. Measured in both
directions: removing that branch together with the kind-is-not-scout guards
lets an explicit reportless scout into Recently Landed and fails
tests/fm-captain-hold-lifecycle.test.sh, while removing the branch alone
leaves that suite passing at 49 assertions. The guards carry the rule, so the
note now sits on them and the unreachable branch is gone. A note pointing a
later reader at the wrong line is the hazard this change corrects elsewhere.
summary_file_has_schema lost its only caller when the stale-schema machinery
was removed, so it goes with it.
* fix(review): Fix legacy report artifacts and canonical keyword boundaries
* fix(review): Update pinned Bearings test count to 56
* fix(document): Clarify landed summary compatibility documentation
* fix(review): Preserve unreadable backend configuration errors
* fix(review): Honor backend resolution errors at existing call sites
* fix(test): Stabilize remote collector tests under host load
* fix(document): Document backend resolution failure contracts
* fix(lint): Suppress intentional deferred probe expansion warnings
* fix(ci): Captain, quoted the two literal test IDs in tests/fm-backlog-atomicity.test.sh to fix SC2100 without changing behavior. Both warnings reproduced before the fix; the targeted fm-lint.sh run now passes with ShellCheck 0.11.0. Bash syntax and git diff --check also pass
* fix(ci): Fixed the resolver’s two configuration-parent checks to return 2 for inaccessible directories while preserving genuine absence. Added two behavioral tests; RED/GREEN and both requested mutation proofs confirmed. All 10 focused checks, targeted lint, syntax, and whitespace checks passed. Broader merge suite stopped after 10 passing cases under host load. Declined portable checks and merge-authority code remain unchanged
* fix(remote): keep fm-remote Herdr servers in the Aqua session (#4090)
* fix(remote): let the Aqua launch agent own the fm-remote Herdr session
A herdr server keeps the macOS audit session of whatever started it, and
only the Aqua login session (gui/<uid>) can read the login keychain
without a prompt. Herdr's SSH remote attach starts the fm-remote server
as its own child when it finds none, wins the socket at boot because sshd
accepts connections before the login session exists, and every claude
pane under that server then gets `security` exit 36, falls back to a stale
plaintext credentials file, and reports "Login expired". launchd's own job
lost the socket on every KeepAlive retry and the doctor still reported the
session ready because it only asked whether any server answered.
- Add bin/fm-remote-herdr-guard.sh, the launch agent's exec target: start
the server in the foreground when nothing owns the socket, exit 0 when an
Aqua-born server does, and otherwise stop the foreign server, wait for the
socket, and exec the server at once.
- Add bin/fm-remote-herdr-owner-lib.sh, the single owner of socket-owner
discovery (lsof; pgrep cannot see herdr's argv on macOS) and the birth
markers (SSH_*, XPC_SERVICE_NAME, FM_REMOTE_JOB_ACTIVE, sshd or
remote-client-bridge ancestry matched on argv[0] and whole arguments).
- Render the agent as the login shell exec'ing the guard with
KeepAlive={SuccessfulExit=false} and ThrottleInterval=10, check the loaded
job's successful-exit semaphore, and report a session served outside the
Aqua login session as fixable so --fix retakes it through launchd; the
reload waits for an Aqua-born owner rather than any running server.
- Correct the doctor and docs: the launch shell provides environment parity,
the launchd domain provides keychain access.
- Pin the guard's decision table and the doctor's verdicts against real
marker-carrying processes, and record the dated audit-session evidence.
* no-mistakes(review): Verify Aqua ownership through launchd domains
* no-mistakes(document): Document macOS lsof ownership requirement
* fix(bin): prefer a live no-mistakes run over a terminal one (#2881)
* fix(bin): prefer a live no-mistakes run over a terminal one
A worktree can bind to more than one recorded no-mistakes run at once.
The branch-and-code-identity rule in bin/fm-nm-run-lib.sh accepts both an
exact-equal commit and a worktree-is-an-ancestor match, but never stated
which wins when both bind, so the tie fell to whichever candidate the
caller reached first.
Observed on a live fleet: a crashed validation daemon left a FAILED run
at the worktree's own commit while the live run that replaced it
validated a descendant commit on the same branch. Bare `axi status`
answers with the most-recently-touched run - the corpse - and it bound by
the equal-commit rule, so every recomputation reported `failed` for a
task whose real run was healthy. The same label had also read `failed`
earlier while the work was genuinely stalled, so the signal was wrong in
both directions.
State the live-over-terminal policy in the matching rule's own contract,
where the equal-commit and ancestor rules already live, and add
fm_nm_run_status_class as the one classifier that decides liveness from a
recorded status word. fm-crew-state.sh applies it on both selection
paths: the runs listing now scans past a terminal row for a live one, and
a terminal `axi status` answer is provisional until the listing has been
asked whether this worktree also has a live run.
Same-liveness-class candidates keep the listing's newest-first
precedence, and a status word the classifier cannot place keeps the
caller's own ordering rather than displacing a known result, so a
single-run task and a task whose runs are all terminal are unchanged.
Regression coverage reproduces the proven case (terminal run at the
worktree's exact commit plus a live run descending from it) and its
runs-list twin; both fail under the old tie-break. Two companion cases
pin the no-widening half - two terminal rows still resolve newest-first,
and a terminal run with no live sibling keeps its full run-step detail -
and both pass before and after the change.
* no-mistakes(review): accept unfetched live sibling anchored at exact worktree head
* docs(bin): name both ledger reads behind the runs-limit setting
The FM_CREW_STATE_RUNS_LIMIT comment in bin/fm-crew-state.sh still described
the runs ledger as scanned only by the cross-branch fallback, but the
live-over-terminal fix also consults it as the live-sibling probe behind a
terminal axi status answer. Point the comment at docs/configuration.md as the
setting's owner instead of restating a second copy.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
* fix(bin): repair process-event shutdown and tighten guard timing (#4009)
* fix(procevent): make the ordinary stop signal actually stop a runner
The owner guard that shipped in #3904 half-reaps. Against a poll child that
handles the ordinary stop signal and keeps waiting, the guard signals the group,
loses the runner leader to its own signal, then reads that success as a
leaderless group and exits without escalating. It destroys the only proof of
ownership that would have authorised the forced signal, so the survivor becomes
unreachable by retire, reconcile, sweep-home and the guard alike. A guard that
turns a leaking-but-identifiable generation into a permanently unreachable one
is worse than no guard at all.
Two defects, and they hid each other:
- The escalation re-derived ownership from the leader. `runner_group_signal`
now takes a `proved` mode, passed only by the escalation inside the stop that
already proved and signalled that exact generation moments earlier. A leader
dying to our own signal is the ordinary outcome, not fresh ambiguity.
- Every stop held the per-source lock across its wait while the runner's own
exit cleanup waited unboundedly for that same lock. That circular wait was
broken only by the forced signal, so the forced signal silently became the
normal path - and, by keeping the leader alive through the whole window, it
masked the escalation defect above. The runner's exit cleanup now refuses that
lock instead of waiting for it, which is what its existing `return 0` already
said it did.
Fixing the lock alone would have turned every stop of a signal-proof child into
a refusal that leaves it running, so both land together and the tests pin that.
Measured on macOS with a stand-in poll child that traps TERM, INT and HUP:
the guard left it running past 70s and now clears the group within the lease
plus one check; retiring a healthy runner fell from ~2.8s with a forced group
signal every time to ~0.6s on the ordinary signal alone.
Unchanged and stated deliberately: a leader lost to anything other than the
stop's own signal still leaves a group that retire, reconcile, sweep-home and
the guard all refuse, permanently - and that source stops listening without
saying so. Whether such a group may ever be signalled is an open decision and
is not answered here.
* fix(review): Fix proved escalation race and stop regression assertions
* fix(review): Preserve proved escalation through transient identity failures
* fix(review): Simplify proved escalation and correct guard timing documentation
* fix(document): Clarify process-event stop ownership and cleanup limits
* fix(document): Clarify process-event stop ownership and fixture comments
* revert(skills): restore the leaderless-ambiguity limit to the loaded skill
An automatic documentation step in this branch's validation edited
.agents/skills/process-event-sources/SKILL.md, which no instruction in this
change asked it to touch. That file is not documentation about the code: it is
the agent-loaded instruction surface, what an agent reads to know what it is
permitted to do.
The step deleted this line:
- leaderless PID/PGID-reuse ambiguity preserves the claim without signalling
or replacement, as owned by the operating contract in
[`docs/configuration.md`](../../../docs/configuration.md#process-to-event-sources-stateprocevent);
and folded it, with its neighbour, into a generic "registration and ownership
transitions, stop authority, and claim reclamation follow the operating
contract".
That deleted line states a PROHIBITION - that such a group is preserved WITHOUT
SIGNALLING - and it is the exact limit an open captain decision currently rests
on. Folded into a pointer, an agent reading the skill to learn what it may do
would have to chase a second document to discover it may not signal. A
prohibition that requires a second lookup is not a prohibition. The effect was
to weaken, in the instructions themselves, the boundary that keeps one home from
signalling another's process group - while the question of whether that boundary
should move at all is still open.
This is a deliberate revert, not an oversight, and it restores the file exactly
to its pre-branch state. The full statement also survives in
docs/configuration.md; that does not rescue it, because the agent handling a
process-event wake loads the skill and not the documentation.
* revert(procevent): restore the open-question marking beside the escalation
The same automatic documentation step that edited the loaded skill also removed
this from the comment above runner_group_signal:
A leaderless group nobody in this call ever proved remains refused too, for
every caller. That untouched refusal is what makes a crashed leader's group
permanent, and relaxing it is a separate open question, not something this
path assumes.
and replaced it with a pointer to docs/configuration.md.
This one fails differently from the skill deletion, which is why it is restored
separately. There, a prohibition was moved out of the reader's path, and a
missing prohibition gets violated. Here the prohibition survives in code - the
unproved path still refuses - and what was removed is the fact that the limit is
UNDECIDED. A prohibition that has quietly lost its "this is still open" reads as
settled design, and settled design gets relied on, extended, and eventually
relaxed by someone confident they understand why it is there. That question is
open right now.
The rule this branch's four instances produce, stated once here because this is
the point of decision: an unresolved question must be marked unresolved AT THE
POINT OF DECISION, not only where the contract is documented. A reader who does
not know something is open will treat it as closed, and that default is stronger
than any pointer overcomes.
The pointer added by that step is kept alongside; this restores what it replaced
rather than reverting it.
* docs(verification): restore the measured guard bound and its reason
The document step's rewrite of this record dropped the concrete figure while
keeping the surrounding measurements. What went missing was the bound itself -
lease plus two consecutive failed checks plus the stop's grace, roughly 630
seconds at the shipped 600-second lease and 15-second check - together with the
reason there are two checks rather than one: a single unreadable read must not
be enough to kill a live runner.
The mechanism survived elsewhere and the reason survived in
docs/configuration.md, so nothing was lost from the repository. The concreteness
was, and that is what this restores. A number recorded without why it is that
number is the one a later reader shortens; the reason is the whole safety
argument for the debounce, and the debounce is what stops the reaper killing a
live runner on one bad read.
* fix(document): Replace stale stop-authority summaries with owner pointers
* test(procevent): make the guard-bound case able to fail for its own reason
An automated reviewer observed that this case allowed sixty seconds for a bound
of roughly eight, so it could not go red for the reason it names: it would have
passed a guard that took fifty-five seconds. That is correct, and it is the same
family as the defect the case exists to defend against - a check that is green
because it cannot fail, rather than because the thing it guards is working.
The deadline is now derived from the bound itself - the lease, plus the two
consecutive failed checks the guard debounces on, plus the stop's own ordinary
and forced signal windows - rather than from a flat wall-clock number, and the
shortened lease and check the fixtures run under have a single definition so a
derived deadline cannot silently diverge from the settings the guard is given.
The doubling that remains is a load allowance and is documented as one; widening
it to make a slow guard pass would convert the assertion back into decoration.
Proven by mutation rather than by argument. Against the repaired case:
correct code ok
guard debounces on 20 misses instead of 2 not ok - "still holding
the group after 16s,
against a documented
bound of 8s"
proved escalation removed (the original defect) not ok - same
code restored ok
The previous sixty-second version passes every one of those mutations.
The reviewer's other claim, that the guard can survive past the announced bound
when an owner disappears immediately after a check, was measured and does not
hold against what this branch announces. Sweeping the phase deliberately at
0.0, 0.2, 0.4, 0.6 and 0.8 of a check interval gave 7.21s, 7.31s, 6.75s, 6.49s
and 6.31s, worst 7.31s, against the announced lease plus two consecutive failed
checks plus stop grace, which is up to 8s at those settings. The mechanism the
reviewer describes is real and is the announced mechanism; the bound it was
measured against is a phrasing this branch no longer carries.
* revert(scope): return the instruction surfaces to their base state
This delivery is being split. It carries the two proven process fixes alone; the
instruction text travels separately, through a run that removes the
documentation step rather than refusing it at its gate.
Two surfaces are therefore returned to exactly what the base branch has, so this
delivery neither adds to them nor removes from them:
.agents/skills/process-event-sources/SKILL.md - identical to base again. Three
bullets an automatic documentation step had folded into a pointer, including
that leaderless PID/PGID-reuse ambiguity preserves the claim WITHOUT
SIGNALLING and that there is one identity-matched owner per canonical source
across homes sharing one store.
The header comment block of bin/fm-procevent.sh, which is what the script
prints as its own help. Seven lines were removed from it: that a live owner is
never displaced, that only a claim whose stale owner and independently absent
process group prove its whole generation gone is reclaimed, that a crashed
leader or reused pid whose process group still has members cannot relax
ownership cleanup, and that reconcile signals only a live identity-matched
runner group and otherwise keeps the claim without starting a replacement.
The help output is now byte-identical to base.
Neither removal was requested by any instruction in this change, and both were
made to text that predates it. Returning them is scoping, not a third
restoration: nothing is being added to those files here.
* fix(ci): Captain, live CI revealed a fixture deadlock: it suspended the runner before startup released its lock. Added a public-list synchronization barrier in tests/fm-procevent.test.sh. Forced-delay reproduction detected the deadlock before the fix; all four cases passed afterward. Targeted lint, Bash syntax, and whitespace checks passed. Greptile’s watchdog requirement conflicts with the recorded R2 decision; runtime behavior and documentation remain unchanged. Full CI rerun belongs to the outer executor
* test(procevent): make the post-TERM cases report what they saw when they fail
On the failure path only, these cases now print what they actually saw: the
identity recorded at claim time, the identity readable at that moment, the size
of the signals file, the leader's state and wchan, every live member of the
runner's process group with its own state and wchan, the elapsed time since the
stop began, and what retire said. None of it runs when a case passes.
WHY THIS IS KEPT, stated accurately rather than by its original reason. It was
written to make an unexplained CI failure verifiable. That failure is now
explained - it was a fixture deadlock, diagnosed and repaired in the preceding
commit - so that justification has expired and is not the reason given here.
The reason it stays is smaller and independent of that failure: it is already
written, it is small, it sits in the file whose assertion this change reworked,
and an assertion that could not say why it failed cost most of a morning to
diagnose from the outside. The next failure will not be this one.
WHAT A PASSING RUN WOULD NOT MEAN: a pass is a sample of behaviour already
observed many times, not proof that anything is fixed. Only a failure carrying
the evidence above establishes a cause.
* fix(document): Clarify process-event fixture diagnostic rationale
* fix(ci): Captain, fixed two cleanup races in tests/fm-procevent.test.sh: removed premature child completion and waited for runner exit before retiring the restart fixture. Controlled Linux reproductions demonstrated failure before and success after. The full Linux process-event suite, six focused macOS checks, targeted ShellCheck, Bash syntax, and whitespace checks passed. Runtime behavior, guard debounce, and documentation remain unchanged. CI rerun belongs to the outer executor
* fix(procevent): bound owner-guard cleanup at one check interval, not two
A THIRD WAY, not a capitulation to the reviewer and not a refusal of it.
The automated reviewer's grievance was the LOOSENESS OF THE BOUND, not the
number of observations the guard makes before it acts. It asked for a single
read because that was the only route it could see to an acceptable bound. There
was another route, and this change takes it: the bound is reached and both reads
are kept.
TIGHTENED - the SPACING of the guard's two reads, not their number. The owner
watchdog now sleeps half the configured check interval and still requires two
consecutive failing reads, so the pair completes inside one check interval
instead of costing two. Worst-case detection falls from the lease term plus TWO
check intervals to the lease term plus ONE. At the shipped 600s lease and 15s
interval the stated bound falls from ~635s to ~620s.
PRESERVED - the second read. bin/fm-procevent.sh's two-consecutive-miss rule is
untouched. WHY IT PROTECTS: the guard's inputs are a lease read and a state-root
identity read, and either can fail transiently on a live, healthy home. Acting
on the first failure would let one isolated unreadable read kill a live service.
Requiring a second, independent read is what makes that impossible, and it is a
protection rather than padding. Nothing was traded away to reach the bound.
Both properties are now guarded by their own case, and each was proven by
MUTATION rather than asserted:
- putting a full interval back between the two reads fails the bound case:
"still running 17.0s after the last owner activity, against a documented
bound of 15s";
- acting on one failed read fails the new debounce case: "one unreadable lease
read ended a runner whose home was still alive" - while the bound case then
passes FASTER, 9.9s against 13.1s. The unsafe variant being the quicker one
is exactly why these are two cases: one elapsed-time case would have
registered the removal of the protection as an improvement.
MEASURED, sampling the phase between the guard's check clock and the lease clock
across eight runs per variant, on macOS (Darwin 25.5.0). Reaping an orphaned
listener whose home stopped refreshing its lease:
lease 2s / interval 1s: 4.41-5.29s before, 3.48-4.65s after
lease 2s / interval 4s: 7.69-8.12s before, 5.94-6.13s after
The 4s configuration is the informative one: the gap is about one check
interval, which is precisely the term that was removed.
A previously unstated term of the bound surfaced while measuring: the lease age
is compared in whole seconds, so a configured lease of N is honoured until that
age reads N+1. It is now part of the documented bound and of the regression's
derivation instead of being absorbed into a fudge factor.
The bound regression derives its deadline from the documented bound instead of a
flat number, and PINS the phase between the guard's check clock and the lease
clock rather than sampling it, because with a sampled phase a guard spending two
intervals passes about half the time on a lucky alignment. Its load slack is
additive and stays under half a check interval, so an extra whole interval
cannot hide inside it. The two flat deadlines that were there before (40s and
20s) and the doubling allowance on the derived one are gone; that looseness was
the reviewer's third complaint.
The stop's own grace is untouched: 2s for the ordinary signal, then 2s for the
forced one. It is a ceiling paid only by a group that outlives the signal it was
sent, not a delay every stop pays - a healthy runner's whole retire measures
0.40-0.66s on this host. The reviewer's literal "lease plus one tick" is
unreachable by any implementation, since signalling a process and giving it any
chance to exit takes non-zero time; detection now meets it and the stop runs
inside its own ceiling, and the contract says so rather than glossing it.
NECESSARY BUT NOT SUFFICIENT, and written BEFORE this head's integration runs
start rather than after they report. On the previous head, "Behavior portable
serial 1" and "Behavior portable serial 4" were both CANCELLED at the job
ceiling, independently of this finding. A new head triggers fresh runs, so those
two lanes MAY complete this time. IF THEY DO, THAT IS NOT EVIDENCE THE CEILING
DEFECT IS FIXED. It is one more sample of a lane that has been cut repeatedly
and sometimes is not; the shard-packing repair for it is open separately. Do not
reread a lucky pass here as a resolution.
Relatedly, and deliberately: the per-script duration hint in bin/fm-test-run.sh
was NOT updated even though the two new cases add ~19s of wall clock.
docs/fm-test-portable-shards.md says those hints are replaced wholesale from CI
timing artifacts of green runs, and that repair is the open request doing it; a
hand-edited estimate here would collide with it and silently repack the shards.
This suite runs in portable serial shard 3, which was green in the last run.
Verification: tests/fm-procevent.test.sh green, plus
tests/fm-captain-hold-lifecycle.test.sh, the test-coverage guard, and
bin/fm-lint.sh. The unrelated "reconcile stops a runner whose registration was
removed" case flaked in 4 of 7 local full runs; an isolated 20-trial
reproduction measured it at 13/20 unclean before this change and 11/20 after, so
it is issue 4080 and is not aggravated here.
* fix(procevent): repair our decimal-interval regression and enforce the timing phase
REPAIRED BEFORE PUBLICATION, AND IT WAS OURS. The half-interval arithmetic added
by the previous commit read a zero-prefixed interval as octal: 010 halved to 4
instead of 5, and 08 was not a number at all, so the owner guard died before
reporting ready and the runner failed closed and never listened. The validator
accepts those values and `[` compares them as decimal, so this broke a
configuration that worked before. Introduced by this delivery, found in review,
repaired here. Forcing base ten before the arithmetic is the whole runtime fix.
Proven by driving it rather than by reading the source: a new case starts a real
listener at 08 and at 010 and observes the guard's actual sleep argument - 4s and
5s. Removing the normalisation turns that case red with "a zero-prefixed decimal
interval (08) prevented the listener from starting".
THE TIMING PHASE IS NOW OBSERVED AND ENFORCED, NOT ASSUMED. The bound case
pinned its phase by CONSTRUCTION, from an assumed startup time, and enforced
nothing. Review was right that this is not enough: once startup reaches about two
seconds the expiry lands in a different part of the interval and the case
silently stops rejecting a two-interval guard while still reporting success. A
bound that cannot fail for the reason it names is the defect this whole delivery
exists to correct, so it must not ship inside the fix for it.
Now the lease is synchronised to the guard's own FIRST observed lease read,
every later real read is recorded, and the case REFUSES unless one recorded read
proves the required phase: it read the synchronised reference, it was still
fresh, and it began late enough that two further full intervals could not finish
before the deadline. An unestablished precondition refuses; it does not proceed
on trust. The derived deadline, the two-read debounce and the additive slack are
unchanged, and the slack invariant is now asserted rather than left to a comment.
Review also found the deadline was only ever checked while the group was still
alive, so a sampler descheduled past it would see the group gone and certify
success. The observed completion time is now checked too.
PROVEN BY MUTATION, each one run against this code:
- remove the decimal normalisation -> the interval case fails on 08;
- a full interval between the two reads -> "the guard exceeded its bound:
group still running 17.1s ... against a documented bound of 15s";
- a full interval WITH startup forced to ~2.5s, which is exactly the condition
the old construction pin could not survive -> still red, same message;
- the same ~2.5s startup with the correct guard -> still passes, 13.0s against
the 15s bound, so the delay alone does not break the case;
- phase evidence made unavailable -> "could not establish the required
pre-expiry guard-read phase", a refusal rather than a pass, even though the
group stopped quickly;
- act on one failed read -> the debounce case fails and the bound case passes
FASTER, 9.5s against 12.7s, which is why these remain separate cases.
Verification: full tests/fm-procevent.test.sh green, and bin/fm-lint.sh clean.
* fix(document): Correct process-event timing and debounce comments
* fix(backlog): route lifecycle transitions through configured adapters (#3417)
* fix(backlog): honor configured task adapters
* no-mistakes(review): Harden backend purity lint against prefixed Beads calls
* no-mistakes(document): Document configured backend lifecycle transitions
* fix(backlog): preserve markdown exemptions
* no-mistakes(review): Enforce backend purity for explicit lint paths
* no-mistakes(document): Update lifecycle backend documentation
* no-mistakes(lint): Remove redundant backend lint pattern
* fix(backlog): close adapter routing gaps
* no-mistakes(review): Honor configured markdown paths and path-qualified Beads lint
* no-mistakes(document): Document environment-selected backlog adapters
* no-mistakes(lint): Fix empty local variable assignment
* fix(backlog): close quoted path gaps
* no-mistakes(review): Reject partially quoted direct Beads commands
* no-mistakes(document): Align lifecycle documentation with configured adapters
* test(backlog): keep structural cases markdown-only
* fix(backlog): honor configured task adapters
* no-mistakes(review): Harden backend purity lint against prefixed Beads calls
* no-mistakes(document): Document configured backend lifecycle transitions
* fix(backlog): preserve markdown exemptions
* no-mistakes(review): Enforce backend purity for explicit lint paths
* no-mistakes(document): Update lifecycle backend documentation
* no-mistakes(lint): Remove redundant backend lint pattern
* fix(backlog): close adapter routing gaps
* no-mistakes(review): Honor configured markdown paths and path-qualified Beads lint
* no-mistakes(document): Document environment-selected backlog adapters
* no-mistakes(lint): Fix empty local variable assignment
* fix(backlog): close quoted path gaps
* no-mistakes(review): Reject partially quoted direct Beads commands
* no-mistakes(document): Align lifecycle documentation with configured adapters
* test(backlog): keep structural cases markdown-only
* no-mistakes(review): Harden markdown lifecycle routing and close recovery
* fix(lint): catch dollar-quoted beads commands
* no-mistakes(review): Drop unused authorized-data-dir param; canonicalize fm-lint ROOT with pwd -P
* no-mistakes(document): Document fm-lint backend-purity check in header and CONTRIBUTING
* no-mistakes(ci): Two failing checks, one code-caused and one attestation-only. 1. Greptile Review (P1: ANSI-C quoting bypasses lint) — REAL DEFECT, FIXED. The backend-purity normalizer in bin/fm-lint.sh (invokes_bd awk function) stripped $'...' quote delimiters without decoding ANSI-C escapes, so a core script containing $'\x62\x64' close fm-example executes as `bd close` while the lint accepted it. Fix: the normalizer now tracks whether a single-quote context came from $' (ansi flag) and decodes ANSI-C escapes while building the command word: \xHH (1-2 hex), \uHHHH / \UHHHHHHHH (4/8 hex), \NNN (1-3 octal), \cX control characters, simple escapes (a/b/e/E/f/n/r/t/v decode to a placeholder that can never spell bd), NUL (value 0) truncates the word per bash C-string semantics, and unknown escapes drop the backslash per bash. Values outside printable ASCII decode to a placeholder so they can neither falsely match nor collide into `bd`. Regression coverage added to the existing lint-interface test test_rejects_direct_beads_cli_invocations in tests/fm-lint.test.sh: $'\x62\x64', $'\142\144' (octal), and b$'\x64' (split word), each written into a fixture and asserted rejected through the real fm-lint.sh executable. Verified regression property: with the fix stashed, the new hex case passes lint (reproduces the reported bypass); with the fix, it is rejected. Verification: all 29 tests in tests/fm-lint.test.sh pass, and the full CI-parity lint (CI=true bin/fm-lint.sh: ShellCheck 0.11.0 full analysis over the canonical set, backend-purity check, and actionlint workflow validation) exits 0. One iteration was needed because an awk comment containing a literal $'\x62\x64' sample broke shell-level quoting (bash -n / SC1001/SC2026); the comment was reworded without quotes. 2. PR must be raised via no-mistakes — NOT code-caused. The check failed with 'Pipeline attestation head_sha does not match the current PR head': the PR body attestation binds to 82b41c7 while the PR head is 1995cdb because a later pipeline push moved the head. This is exactly the stale-attestation condition the user intent describes; it clears when the outer pipeline re-runs 'git push no-mistakes' and re-binds the attestation to the new head (which now includes this Greptile fix). No code change can or should address it. Files changed: bin/fm-lint.sh (ANSI-C escape decoding in the backend-purity normalizer), tests/fm-lint.test.sh (three encoded-bd rejection cases)
* no-mistakes(review): fix tasks.toml hang, root authorization, lint quoting
* no-mistakes(review): validate tasks config before exemption; fix lint quote gap
* fix(backlog): address the markdown backlog as <data>/backlog.md
Resolving the markdown backlog through a configured `[markdown] path` was
scope this task never asked for. It is absent from main, which addresses
`<data>/backlog.md` everywhere, and it came from an earlier review round
rather than the task brief.
Making it effective on the transition path alone put that path at odds
with every other consumer of the same backlog - fm-captain-hold.sh,
fm-session-start.sh, fm-fleet-snapshot.sh, fm-inbox.sh,
fm-backlog-handoff.sh - which all still address `<data>/backlog.md`. In
fm-captain-hold.sh the split was live: its reads had already moved to the
shared gate while its writes had not, so the two could address different
files.
Address `<data>/backlog.md` from the shared gate, delete the unused
resolver, and drop the two tests that pinned the withdrawn behaviour.
What this task actually changes is unaffected: a configured non-markdown
adapter is still addressed by its own root, without `--file`.
* fix(backlog): honor configured task adapters
* no-mistakes(review): Harden backend purity lint against prefixed Beads calls
* no-mistakes(document): Document configured backend lifecycle transitions
* fix(backlog): preserve markdown exemptions
* no-mistakes(review): Enforce backend purity for explicit lint paths
* no-mistakes(document): Update lifecycle backend documentation
* no-mistakes(lint): Remove redundant backend lint pattern
* fix(backlog): close adapter routing gaps
* no-mistakes(review): Honor configured markdown paths and path-qualified Beads lint
* no-mistakes(document): Document environment-selected backlog adapters
* no-mistakes(lint): Fix empty local variable assignment
* fix(backlog): close quoted path gaps
* no-mistakes(review): Reject partially quoted direct Beads commands
* no-mistakes(document): Align lifecycle documentation with configured adapters
* test(backlog): keep structural cases markdown-only
* fix(lint): catch dollar-quoted beads commands
* no-mistakes(review): Drop unused authorized-data-dir param; canonicalize fm-lint ROOT with pwd -P
* no-mistakes(document): Document fm-lint backend-purity check in header and CONTRIBUTING
* no-mistakes(ci): Two failing checks, one code-caused and one attestation-only. 1. Greptile Review (P1: ANSI-C quoting bypasses lint) — REAL DEFECT, FIXED. The backend-purity normalizer in bin/fm-lint.sh (invokes_bd awk function) stripped $'...' quote delimiters without decoding ANSI-C escapes, so a core script containing $'\x62\x64' close fm-example executes as `bd close` while the lint accepted it. Fix: the normalizer now tracks whether a single-quote context came from $' (ansi flag) and decodes ANSI-C escapes while building the command word: \xHH (1-2 hex), \uHHHH / \UHHHHHHHH (4/8 hex), \NNN (1-3 octal), \cX control characters, simple escapes (a/b/e/E/f/n/r/t/v decode to a placeholder that can never spell bd), NUL (value 0) truncates the word per bash C-string semantics, and unknown escapes drop the backslash per bash. Values outside printable ASCII decode to a placeholder so they can neither falsely match nor collide into `bd`. Regression coverage added to the existing lint-interface test test_rejects_direct_beads_cli_invocations in tests/fm-lint.test.sh: $'\x62\x64', $'\142\144' (octal), and b$'\x64' (split word), each written into a fixture and asserted rejected through the real fm-lint.sh executable. Verified regression property: with the fix stashed, the new hex case passes lint (reproduces the reported bypass); with the fix, it is rejected. Verification: all 29 tests in tests/fm-lint.test.sh pass, and the full CI-parity lint (CI=true bin/fm-lint.sh: ShellCheck 0.11.0 full analysis over the canonical set, backend-purity check, and actionlint workflow validation) exits 0. One iteration was needed because an awk comment containing a literal $'\x62\x64' sample broke shell-level quoting (bash -n / SC1001/SC2026); the comment was reworded without quotes. 2. PR must be raised via no-mistakes — NOT code-caused. The check failed with 'Pipeline attestation head_sha does not match the current PR head': the PR body attestation binds to 82b41c7 while the PR head is 1995cdb because a later pipeline push moved the head. This is exactly the stale-attestation condition the user intent describes; it clears when the outer pipeline re-runs 'git push no-mistakes' and re-binds the attestation to the new head (which now includes this Greptile fix). No code change can or should address it. Files changed: bin/fm-lint.sh (ANSI-C escape decoding in the backend-purity normalizer), tests/fm-lint.test.sh (three encoded-bd rejection cases)
* fix(bin): preserve captain calls during teardown (#3595)
* fix(bin): never close a captain call during cleanup
A scout that held its own work item for the captain, which is what
captain-hold-lifecycle prefers ("hold the work item the question gates"),
was closed by bin/fm-teardown.sh's automatic backlog transition. The
completion gate passed, cleanup ran, and the captain's question moved to
Done with no recorded answer: the one thing the policy says must never
happen. `tasks-axi done` closes a held row silently, and nothing in
teardown asked whether the row was the captain's own call.
bin/fm-captain-hold.sh gains the read-only `open` predicate: exit 0 when
the task is still an open captain call, 1 when it is not, 2 when that
cannot be established. It reads the row through the transition library's
backend-aware probe, so it addresses the same backlog teardown does; the
script's other commands now address the configured data directory the
same way instead of FM_HOME, which also fixes captain holds in a home
with a relocated data directory.
Teardown asks `open` before any destructive step and refuses on 2. On 0
only the close changes: after cleanup and still under the task's own
lock, the row gets one "Deliverable of the finished work" line at the end
of its body and returns to Queued through `tasks-axi reopen`, keeping its
hold, so it lands in Captain's Call instead of reading as work under way.
--force does not lift this: it authorizes discarding unlanded work, never
the captain's question. The deliverable goes into the body because
`tasks-axi update --report` rewrites the title of a row that is not Done.
The crash window reuses the pending-close record teardown already stages:
a `mode=retain` line makes the existing replay record the deliverable and
reopen instead of closing, with the same validator, stale-generation
check, cleanup-incomplete marking, and non-blocking bootstrap lock as an
ordinary close. A retained row the captain answered first simply retires
the record. No parallel record type, recovery command, or second bootstrap
loop is introduced.
Regressions run the real executables: the captain-held scout survives
cleanup queued, held, with its deliverable and on the board, only
`answer` closes it, --force keeps it open, and an ordinary scout still
closes with its report; an interrupted cleanup leaves the row untouched
and the next session start retains it; a relocated backlog keeps the
retention in its one configured file; and a ship row whose hold cannot be
read refuses cleanup before anything destructive.
Claude-Session: https://claude.ai/code/session_01FqdTiHCwTqrAQrz8K2y4Np
* no-mistakes(review): Serialize captain holds and fix backend-aware listing
* no-mistakes(document): Update captain-call retention documentation
* no-mistakes(document): Fix relocated captain-hold backlog diagnostics
* fix(backlog): honor configured task adapters
* no-mistakes(document): Update lifecycle backend documentation
* fix(backlog): close adapter routing gaps
* no-mistakes(review): Honor configured markdown paths and path-qualified Beads lint
* no-mistakes(lint): Fix empty local variable assignment
* fix(backlog): close quoted path gaps
* test(backlog): keep structural cases markdown-only
* no-mistakes(review): Harden markdown lifecycle routing and close recovery
* no-mistakes(review): fix tasks.toml hang, root authorization, lint quoting
* no-mistakes(review): validate tasks config before exemption; fix lint quote gap
* fix(backlog): address the markdown backlog as <data>/backlog.md
Resolving the markdown backlog through a configured `[markdown] path` was
scope this task never asked for. It is absent from main, which addresses
`<data>/backlog.md` everywhere, and it came from an earlier review round
rather than the task brief.
Making it effective on the transition path alone put that path at odds
with every other consumer of the same backlog - fm-captain-hold.sh,
fm-session-start.sh, fm-fleet-snapshot.sh, fm-inbox.sh,
fm-backlog-handoff.sh - which all still address `<data>/backlog.md`. In
fm-captain-hold.sh the split was live: its reads had already moved to the
shared gate while its writes had not, so the two could address different
files.
Address `<data>/backlog.md` from the shared gate, delete the unused
resolver, and drop the two tests that pinned the withdrawn behaviour.
What this task actually changes is unaffected: a configured non-markdown
adapter is still addressed by its own root, without `--file`.
* no-mistakes(review): restore home boundary guard and tighten purity lint
* no-mistakes(review): authorize home boundary for every backlog adapter
* no-mistakes(test): complete tasks-axi stubs in fm-gotmp teardown fixtures
* no-mistakes(document): align backlog transition docs with adapter-neutral addressing
* no-mistakes(review): label adapter data-dir authorization, drop dead row_probe local
* no-mistakes(review): pin markdown backend at relocated-data addressing roots
* no-mistakes(document): point lint-definition mention at fm-lint.sh header
* no-mistakes(document): point mutate comment at adapter addressing owner
* no-mistakes(review): Fix leftover-symlink refusal on non-markdown homes; hoist config check and lint/dedup cleanups
* no-mistakes(document): Align fm-lint purity scope header with bin/backends
---------
Co-authored-by: Kun Chen <3233006+kunchenguid@users.noreply.github.com>
* fix(bin): suppress false Claude long-turn supervision alarms (#4119)
* fix(bin): stop false watcher-down alarms on long Claude turns
A healthy Stop auto-arm rewake or open claim already explains a mid-turn
beacon that has aged past grace, because turn-end will re-arm. Keep the
supervision-off banner for a missing, failed, or exhausted generation.
* no-mistakes(review): Bind Claude rewakes to active recovery generation
* no-mistakes(document): Document Claude long-turn supervision exception
* no-mistakes(lint): Fix empty ShellCheck assignment
* no-mistakes(ci): Fixed all reported CI failures: quoted the hyphenated recovery-delivery value to satisfy ShellCheck SC2100, and updated the session-lock auto-arm fixture to emit the recovery marker and watcher beacon now required for a valid rewake. Verified fm-session-lock-ancestry, fm-test-run, stale-banner, Claude auto-arm, targeted lint, ShellCheck, and workflow lint checks pass
* fix(herdr): recover gone and drifted worker endpoints (#4120)
* fix(herdr): classify a gone session's endpoint as recoverable
A task whose Herdr endpoint could not be read was classified `unreadable`,
which blocks recovery by design. The commonest reason that read fails is
that the recorded session's server is not running at all - a host reboot, a
server exit, a session never restored - and that is authoritative absence
for every pane in that session, not an ambiguous answer about one of them.
Tasks in that state had no sanctioned way back.
The recovery-grade read now settles an uninterpretable pane read with the
session server's own `.server.running` state: positively stopped reads
`missing`, while a running server, or a server state that cannot itself be
read, still reads `unreadable`. Resting the verdict on that field rather
than on the `server_not_running` error code is what keeps it working across
Herdr 0.8.x and 0.9.0, since the field is present on both and the code is
not.
Only that one boundary is widened. The husk classifier under it stays
strict, so duplicate prevention, rollback, and teardown - the paths that can
destroy something - keep refusing on exactly the reads they refused on
before.
Separately, a relaunch refused outright when the endpoint's shell had
drifted out of the recorded worktree. An agent's own exit routinely leaves
its shell somewhere else, so that refusal stranded tasks whose work was
sitting untouched on disk. The shell is now told once to return, and only a
shell that will not go refuses; the replacement still never starts outside
the copy holding the work.
Herdr 0.8.x is not installed on this host, so protocol-20 coverage is
structural plus the adapter fixture exercising both response shapes, and is
recorded as such rather than as a live result.
Fixes #4091.
* no-mistakes(review): Restrict drift recovery to Herdr endpoints
* no-mistakes(review): Correct Herdr recovery verification coverage
* no-mistakes(document): Document Herdr endpoint recovery boundaries
* fix(bin): prevent receiver wake failures from blocking remote handoffs (#4033)
* fix(bin): keep an escalated undelivered handoff wake retryable
A remote backlog handoff holds its outbox until the backlog receipt and
the receiver wake are both confirmed, and retries the wake under the same
pending-reply correlation on every resume. When that wake's remote
transport was lost, the correlation stayed undelivered in delivery_unknown
and the watcher's next pending-reply tick escalated it. Both the reuse
predicate and the known-undelivered reset refused an escalated record, so
the resume refused to resend the wake forever and every later handoff to
that mate jammed behind the outbox.
Treat an escalated record with no confirmed delivery as the undelivered
correlation it is: fm_pending_reply_corr_reusable accepts it for its own
task and fm_pending_reply_reset_known_undelivered returns it to
awaiting_report for the idempotent remote resend, while a delivered
record is still never reset and a missed-report escalation keeps its
meaning. The published delivery-unknown decision stays open until the
record resolves, so a repeat loss neither re-notifies nor strands it.
Reproduce the deadlock end to end in the remote handoff test (lost wake
transport, watcher escalation, resume) and pin the predicate contract in
the pending-reply suite; the fm-send fixture that pinned the refusal now
uses a genuinely stale delivered escalation.
* no-mistakes(review): Decouple durable outboxes from best-effort wake retries
* no-mistakes(review): Align handoff documentation with durable receipt release policy
* no-mistakes(review): Handle unrecordable wake state as dropped
* no-mistakes(review): Prevent stale wake markers blocking handoffs
* no-mistakes(review): Prevent stale delivered markers suppressing new wakes
* no-mistakes(document): Clarify retry escalation decision lifecycle
* no-mistakes(document): Document pending receiver wake retries
* fix: restrict captain address rule to user chat (#4075)
* docs: bound the mandatory captain address to the chat channel
AGENTS.md's opening address rule said "address the user as captain at
least once in every response" and never said what a response is. The
artefact exclusion two lines below governed only the optional nautical
seasoning, not the mandatory address. An agent that reads this file
without being the first mate - a pipeline corrector agent running inside
a copy of this repo - therefore read the obligation as applying
everywhere and the exclusion as applying only to flavour, and opened its
delivery message with "Captain,". That reading was correct.
Patch the existing owner rather than adding a rule elsewhere:
- bound the obligation to chat messages sent to the captain;
- state the artefact exclusion once, explicitly binding every agent that
reads this file whether or not it is the first mate, and naming commit
messages, PR and issue descriptions, briefs, code and comments;
- fold the seasoning under the same bound instead of carrying a second,
narrower copy of the exclusion.
The obligation itself is unchanged: the captain is still addressed in
every chat message.
AGENTS.md goes from 603 to 602 lines: the redundant "never send a
response with zero direct address" clause and the duplicated seasoning
exclusion pay for the new bound.
The two cross-references that paraphrased the unbounded wording
(bin/fm-parent-channel-lib.sh's header and
docs/secondmate-parent-channel.md's problem statement) now match the
owner; neither restates the rule.
* fix(review): Limit address exclusions to artifacts while preserving public replies
* fix(document): Consolidate captain address guidance
* fix(herdr): allow detached teardown of persisted-focused tabs (#4131)
* fix(herdr): close persisted-focused tabs when no live client is attached
The teardown active-tab guard treated Herdr's last-focused pointer as a live viewer, so detached sessions could not close panes on that tab.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(herdr): allow detached seeded-tab prune after live-client gate
Projection create still restored the persisted focused tab after a successful prune, so a detached last-focused seeded tab still quarantined the spawn.
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(herdr): probe live client after seeded prune only when that tab was focused
The extra title-clear read after every prune shifted canned CLI fixtures and failed projection create.
Co-authored-by: Cursor <cursoragent@cursor.com>
* no-mistakes(review): Tighten Herdr active-tab close guard
* no-mistakes(review): Guard Herdr mutations with fresh target focus
* no-mistakes(document): Document Herdr live-viewer teardown guard
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(bin): escalate decision-owned wakes once as the decision (#4169)
* fix(bin): escalate decision-owned wakes once as the decision
The away-mode daemon treated a needs-decision: queued payload as an
unknown wake, so suppression markers never committed and the same open
decision re-escalated on every poll.
Classify that payload through the existing signal path so it escalates
once, labelled as the decision, and an unchanged repeat is suppressed
on the same terms as any other signal.
Fixes #4096
* no-mistakes(review): Escalate captain-held decision-owned rows once as the decision
* no-mistakes(review): Self-handle captain-held decision-owned rows instead of escalating them
* no-mistakes(document): Name away daemon as needs-decision payload reader
* test(herdr): cover agent exit-to-shell liveness (#4172)
* test(herdr): pin leftover-shell vs live-idle via agent get
Herdr 0.9.0 already distinguishes a Pi that exits to a surviving pane shell
from a sibling live idle occupant. Pin that pair through agent get and the
recovery classifier so a lagged pane-get status cannot silently reclaim the
leftover shell as alive.
Co-authored-by: Cursor <cursoragent@cursor.com>
* no-mistakes(document): Document Herdr leftover-shell liveness regression
* no-mistakes(ci): Fixed Lint failure SC2034 by replacing the unused wait-loop variable with `_`. Verified with the pinned project lint command, Bash syntax check, and git diff check
---------
Co-authored-by: Cursor <cursoragent@cursor.com>
* fix(bin): rebalance portable parallel test lanes using CI timings (#4151)
* ci: rebalance the portable …
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Intent
The captain wants this deadlock fixed now (routed by the main firstmate 2026-09-08, who root-caused it; it jammed all backlog handoffs to this home for 3 days).
The defect: a remote-secondmate backlog handoff (bin/fm-backlog-handoff.sh) holds its outbox until BOTH the backlog receipt AND the receiver WAKE are confirmed. The wake is sent via fm-send with FM_PENDING_REPLY_EXISTING_CORR. When that wake's fm-send returns nonzero (an unknown/failed remote delivery - an ssh/transport blip, or the best-effort live doorbell not landing on a busy agent's pane), delivered_epoch stays empty with an
attempted=marker. If, BEFORE a successful wake retry, the pending-reply escalation lifecycle (bin/fm-pending-reply-lib.sh, fm-watch recovery to escalate) advances the record to phase=escalated, then: fm_pending_reply_reconcile_delivery only handles attempted+awaiting_report; fm_pending_reply_reset_known_undelivered only handles awaiting_report|delivery_unknown; and fm_pending_reply_corr_reusable returns true only for awaiting_report|recovery_sending|recovery_sent|delivery_unknown - never forescalated. So the resume sees delivered empty + corr_reusable false, refuses to resend the wake, and the outbox is never released - every subsequent handoff to that mate is blocked in the one-at-a-time queue. It hit this home because it is the busy/active mate: the handoff wake requires a confirmed live doorbell, which does not land on a busy agent pane, making the transport-blip-then-escalate race far likelier than for idle mates.Done means: an UNDELIVERED wake whose pending-reply reached
escalatedis retryable again so a handoff resume can re-send it - either extend corr_reusable and/or reset_known_undelivered (and reconcile) to cover the escalated+undelivered case, OR decouple the pending-reply escalation lifecycle from the handoff wake-retry lifecycle so escalation never strands an undelivered wake. A CONFIRMED (delivered) record must still never be reset, and genuine missed-report escalation semantics must not change. The outbox must never jam indefinitely on an unconfirmable wake once the backlog receipt already succeeded (the work is durably received; the wake is only a best-effort live nudge the mate does not strictly need - it reconciles its backlog regardless), so define the correct release/retry behavior for that case. Reproduce the deadlock in a regression test (drive a wake into attempted+escalated+undelivered and assert the resume path recovers instead of refusing forever). Secondary/optional: note (do not necessarily fix) why the best-effort live doorbell does not land on a busy agent pane over the remote transport; only the escalated-phase deadlock is required.What Changed
Risk Assessment
✅ Low: The change preserves confirmed-delivery safeguards, makes escalated undelivered correlations retryable, and decouples best-effort wake failures from durable remote backlog delivery without a substantiated remaining deadlock path.
Testing
The targeted remote handoff, local handoff, and pending-reply behavioral harnesses all passed, including escalated-undelivered retries, permanent wake failure, stale markers, confirmed-delivery preservation, and exactly-once delivery; localhost SSH was reachable, but a live remote run was not started because the real entrypoint requires remote worker and Herdr endpoint state outside the permitted worktree, and the worktree remained clean.
Evidence: Remote backlog handoff behavioral transcript
Source: Remote backlog handoff behavioral transcript
Evidence: Local backlog handoff behavioral transcript
Source: Local backlog handoff behavioral transcript
Evidence: Pending-reply lifecycle behavioral transcript
Source: Pending-reply lifecycle behavioral transcript
Pipeline
Updates from git push no-mistakes
✅ **intent** - passed
✅ No issues found.
🔧 **Rebase** - 1 issue found → auto-fixed ✅
docs/configuration.md- merge conflict rebasing onto origin/main🔧 Fix applied.
✅ Re-checked - no issues remain.
✅ **Review** - passed
✅ No issues found.
tests/fm-remote-backlog-handoff.test.shtests/fm-backlog-handoff.test.shtests/fm-pending-reply.test.shssh -F /dev/null -o BatchMode=yes -o StrictHostKeyChecking=no -o UserKnownHostsFile=/dev/null -o ConnectTimeout=3 localhost truegit status --short --untracked-files=all✅ **Document** - passed
✅ No issues found.
✅ **Lint** - passed
✅ No issues found.
✅ **Push** - passed
✅ No issues found.
Cancelled CI shard (documented, not a code failure)
Behavior portable parallel 1on CI run 34441064828 (attempts 2 and 3).timeout-minutescap, so under runner load it is cancelled by the provider before the list finishes. It passed on attempt 1 in 8m21s; on the reruns every test ran 15-50 percent slower with zero failing tests (not okcount 0, no nonzero test exit) before the cap cut it.dba3175athroughout), and this branch adds no test to that shard - its tests live in the serial lane. Every other check is green, includingBehavior portable serial 1, whose earlier flake (tests/fm-watch-triage.test.shstall) cleared on rerun.fm-ci-portable-parallel-shard1-timeout-flake, deliberately decoupled from this PR.