chore: snapshot upstream main for 2026-08-29 - #3277
Closed
withally wants to merge 13 commits into
Closed
Conversation
A crew with a current `paused:` (or verified `captain-held`) last status line kept surfacing bare `stale:` wakes every few minutes instead of being absorbed on the long FM_PAUSE_RESURFACE_SECS cadence. pause_state_class read backend agent liveness on EVERY evaluation, and for an ordinary crew whose agent is still live - the normal state of a paused crew - it answered `none`. `none` is a caller's "no declared wait here" signal, so the changed-hash caller ran clear_pause_tracking and erased the cadence flag, its long-cadence throttle, and the per-hash stale suppressor. Any pane redraw (a ticking clock, a token counter, the crew's own status echo) therefore re-armed a first sighting, and the next stable hash surfaced through surface_nonterminal_stale all over again. The liveness gate was only ever meant to give a live agent ONE inspection per declared wait, in case it declared a wait while actually parked at a decision gate. `.paused-<key>` already records that the key is on the bounded cadence, so the gate now reads it and is spent once: a live agent surfaces once and then joins the cadence, and the declaration - not the pane hash - decides how long the cadence holds. It survives further declaring appends and watcher restarts, and is dropped only when the last status line stops declaring a wait or the crew is provably working again. A confidently dead agent still joins the cadence with no initial surface, and a secondmate's endpoint liveness is still never read. The safety boundary is unchanged and now pinned: needs-decision, blocked and done still wake immediately through a fully armed cadence, on the signal path when their append is unseen and on the stale path when it has been seen. Verified against a real fm-watch.sh over five restarts with a churning idle pane: five bare stale wakes before, one after. (cherry picked from commit c735ec1)
* feat: add durable resource telemetry recorder * no-mistakes(review): fix(telemetry): real fsync, atomic lock, UTC dating, temp cleanup * no-mistakes(review): fix(telemetry): durable dir flush, argv tokens, guarded locks, arm probe * no-mistakes(review): fix(telemetry): bounded arm, owner-verified guard, timestamped diagnostics * no-mistakes(document): docs: list fm-telemetry.sh in bin toolbelt inventory * no-mistakes: apply CI fixes * no-mistakes: apply CI fixes (cherry picked from commit 6545983)
…#71) * fix(lock): bound stale recovery mutex acquisition * test(tool-updates): compare line counts numerically * test(lock): tolerate loaded-host timing * no-mistakes(review): fix lock status typing, steal reclaim, and E2E cleanup * no-mistakes(review): serialize steal reclaim, retire nested-steal debris, fail closed * no-mistakes(review): bind reclaim takeover to inspected marker, reclaim legacy steal dirs * no-mistakes(review): fix coordinator argv ownership and marker takeover gap * no-mistakes(review): guard conditional-context lock waits, retire legacy steal debris * no-mistakes(review): harden legacy steal reclaim, marker self-recovery, owner residue * no-mistakes(review): preserve steal-dir owner record, prove descendant-free attempts * no-mistakes(review): restore self-abandoned steal-mutex reclaim in non-recursive helper * no-mistakes(document): document typed lock statuses and non-recursive steal reclaim * no-mistakes(lint): quote stale-beacon verdict literal to fix SC2100 (cherry picked from commit bac5c84)
…ed on missing backend adapters (#72) * fix: restore portable baseline paths * test(remote): compare symlink residue portably * no-mistakes(review): make tree_snapshot fail closed on missing directories * no-mistakes(document): document adapter precheck, portable seed loop, tree_snapshot (cherry picked from commit 228119e)
…#73) * feat: add fseventsd footprint early warning * no-mistakes(review): fix fseventsd pressure encoding, doubling floor, growth window * no-mistakes(review): retain fseventsd action evidence, drop dead pressure clause * no-mistakes(review): retain fseventsd emergency evidence, update scripts index * no-mistakes(review): label fseventsd pressure by level, guard reason separators * no-mistakes(review): report every fseventsd action reason that holds * no-mistakes(review): isolate watcher tests from live fseventsd, alert on write failure * no-mistakes(document): document fseventsd early-warning wake in architecture (cherry picked from commit c8a82e7)
…ns (#75) * fix(pi): clear regular TUI after transcript shrink * no-mistakes(review): settle Pi viewport and init awk blank-run counter * no-mistakes(review): gate live-e2e settle on Calm ship across ship tick * no-mistakes(review): re-anchor bare-executable guard after clear-on-shrink assignment * no-mistakes(review): drop inert hidden-thinking setting from live shrink test * no-mistakes(test): make Pi clear-on-shrink live e2e reproduce stale rows * no-mistakes(document): point README Pi launch at clear-on-shrink rationale (cherry picked from commit e2a3fe4)
* feat: add upstream urgent tripwire * no-mistakes(review): fix upstream tripwire base parsing, wake dedupe, and fetch hardening * no-mistakes(review): report tripwire inspection failures and roll back failed arms * no-mistakes(review): separate hit and failure report keys; harden arm rollback test * no-mistakes(review): wake only on hits and unusable-tripwire conditions * no-mistakes(review): Fixed full-check timeout and fail-closed disarm * no-mistakes(review): Fixed upstream identity, record, state, and payload guards * no-mistakes(review): Fixed test isolation, record integrity, framing, and state creation * no-mistakes(review): Hardened arm state-path validation before creation * no-mistakes(review): Bound checks, rejected unsafe paths, classified merge-base failures * no-mistakes(review): Hardened timeout parsing, record validation, and UTF-8 reports * no-mistakes(test): Route unusable diagnostics to stderr; focused tripwire tests pass * no-mistakes(document): Refresh upstream-sync scope pointers * no-mistakes: apply CI fixes * no-mistakes: apply CI fixes * no-mistakes: apply CI fixes (cherry picked from commit f9c46d4)
* fix(afk): distinguish Herdr Claude foreground busy state * test(afk): handle Claude workspace trust in live guard * fix(afk): harden Herdr Claude submit confirmation * no-mistakes(review): Hardened Claude submit confirmation and E2E assertions * no-mistakes(review): Fixed Claude harness propagation and causal submit confirmation * no-mistakes(review): Fixed Claude submit causality and harness propagation * no-mistakes(review): Hardened Claude submit proof across retries * no-mistakes(test): Captain: live launcher escaping fixed; focused E2E passes * no-mistakes(test): Fixed stale post-turn native-status assertion; live E2E passes * no-mistakes(document): Document Herdr Claude busy-guard behavior * no-mistakes(lint): Scoped ShellCheck suppressions for intentional test subshells * no-mistakes(review): Registered regression; verified fixture alignment; focused tests pass * no-mistakes(document): Updated Herdr Claude supervision documentation * no-mistakes: apply CI fixes * no-mistakes: apply CI fixes (cherry picked from commit 6e32ed7)
* fix(ci): wait for no-mistakes attestation * no-mistakes(document): Update attestation timing documentation * no-mistakes(document): Document attestation convergence and upstream follow-up * no-mistakes: apply CI fixes (cherry picked from commit ca6ca85)
Confidence Score: 5/5The PR appears safe to merge with no concrete changed-code defect identified. The changed operational paths preserve fail-closed delivery, locking, watcher recovery, and attestation behavior, and the investigated failure scenarios were either contradicted or pre-existing without material worsening. Reviews (1): Last reviewed commit: "docs: record the 2026-08-29 catch-up" | Re-trigger Greptile |
Author
|
Opened on the wrong repo. Catch-up PRs belong on withally/firstmate, not kunchenguid/firstmate. Closing; reopening on the fork. |
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.
Tier: weekly
Snapshot
1fd7ea289b7a(upstream/main, feat(bin): add concurrent bounded remote transport lanes #3210).b0638f6..ca6ca85238435974d00034f9f4a064aa0f5fa677.-x; upstream won every conflict.docs/upstream-sync.mdbefore shipping.Verdict table
real-herdr-gatedtests.Validation
FM_BOOTSTRAP_DETECT_ONLY=1 bin/fm-bootstrap.shbin/fm-lint.shbin/fm-test-run.sh --changed --base "$UPSTREAM_BASE" --exclude-family real-herdr-gatedbin/fm-test-run.sh --family real-herdr-gated --fail-on-gate-skip 'herdr not found'— 12/12 passed in the named Herdr lab; default-session tripwire intact.bin/fm-doc-audience-check.shgit diff --check 1fd7ea289b7a...HEADThe weekly changed run completed 158 tests with 10 unrelated baseline/host failures; all retained sync families passed.
Follow-ups
BASHPIDfixtures; composer, bootstrap/cmux, gotmp, and Orca baseline cases.