[Profreshor]

Intent

I'll write the fork grant

Context: that was the captain's answer letting this firstmate open PRs on the captain's fork (Profreshor/firstmate) so the follow-ups from the Codex review on fork PR #3 could ship. One of them, fork PR #7 (Herdr crew scan scope), cannot go green: its "Behavior portable serial 2" job was cancelled twice at the workflow's 30-minute limit. On main the same shard took 29 minutes (run of 2026-10-16 20:41-21:10 UTC) while the other eight shards took 15-20 minutes, so shard 2 is at its limit regardless of PR 7's change.

What Changed

• Raise the portable serial shard packing target to roughly 25 minutes and update the coverage-budget boundary checks.
• Increase the supervision-host test timing hint so its slowest observed run packs alone in a shard.
• Document the revised timing baseline and remaining five-minute CI-tier headroom.

Risk Assessment

✅ Low: Captain: the bounded packing-hint update isolates the measured 25-minute host suite in one serial shard while preserving the existing complete/disjoint coverage checks and 30-minute CI timeout.

Testing

Live execution of the runner produced a passing coverage report with a 1,501,963 ms maximum under the 1,510,000 ms budget and showed the slow host suite alone in shard 1. The focused contract test passed its new exact-budget and one-millisecond-over-budget checks, but later failed an unrelated PTY reparenting fixture; no remote GitHub Actions run was driven in this phase.

• Live validation: ⚠️ inconclusive - 2 of 3 scenarios driven live against the product

Scenario	Result	Live	Evidence
Runner reports a serial packing model below the revised budget	✅ pass	live	portable-serial-coverage.txt
Slow supervision-host suite is isolated in its own portable serial shard	✅ pass	live	supervision-host-lane.txt
Actual GitHub 30-minute shard completes after rebalancing	⏸️ untested	no	No authority in this test phase to dispatch or control the remote CI workflow; run the branch’s GitHub Actions portable-serial jobs to measure the real runner duration.

Evidence: Portable serial coverage report

Source: Portable serial coverage report

FM_TEST_COVERAGE ok total=263 parallel=24 parallel_max_ms=665545 parallel_imbalance_ms=2 parallel_unhinted=0 serial=222 serial_shards=9 serial_unhinted=17 serial_max_ms=1501963 serial_budget_ms=1510000 herdr=17

Evidence: Supervision-host shard assignment

Source: Supervision-host shard assignment

portable-serial-1of9
tests/fm-supervision-host.test.sh

Evidence: Focused runner contract test transcript

Source: Focused runner contract test transcript

ok - exact suite coverage: --all lists every tests/*.test.sh once
ok - family selection returns a proper subset of the suite
ok - single-script selection lists exactly that path
ok - changed-file selection stays conservative (never silent full suite)
ok - a task marker refuses execution in the primary checkout and leaves worktrees and inspection alone
ok - runner and its documentation surfaces select their curated family, not just their contract owners
ok - shell line-ending policy selects runner coverage
fm-test-run: no tests selected for changes vs HEAD (map is conservative; use --all for the complete suite)
ok - changed selection covers dependents, fails closed for live unmapped source, and accepts retired unconsumed source
ok - a bin reference selects the referencing scripts, and consumers still select their curated families
ok - changed defaults to bounded automatic scheduling with serial override
ok - Windows emulation exempts only synthetic POSIX modes
ok - a plain script list defaults to bounded automatic concurrency without an automatic timeout
ok - family proofs run concurrently only within separate family phases
ok - empty changed selection emits deterministic text and JSON summaries
ok - timing markers and JSON artifact are valid
ok - aggregate exit reflects any script failure
ok - gate-skip accounting is honest and non-failing
ok - a gate skip records why it skipped
ok - a script that actually ran records no skip reason
ok - live guards are recorded as a capability class, not a bare env opt-in
ok - fail-on-gate-skip converts herdr-not-found into a hard failure
ok - exclude-family drops the named primary family after selection
ok - proven-isolated scheduling ignores parallel hints
ok - family, all, changed, and script selections ignore parallel hints
ok - portable shard union, disjointness, and coverage guard hold
ok - portable parallel lanes are fully hinted and packed within 5% of each other
ok - portable serial shards are a deterministic disjoint cover of the serial lane
ok - coverage guard bounds the unmeasured share and serial packing within twenty-five minutes
ok - serial packing accepts the exact budget and refuses one millisecond above it
ok - portable serial shard lanes refuse mismatched, out-of-range, and countless names
ok - --jobs refuses non-proven / stateful selections
not ok - --jobs on a proven family must be admitted, got 1:  FM_TEST_BEGIN 2026-10-16T22:48:29Z tests/fm-session-lock-ancestry.test.sh family=watcher-wake-lock expected_gate_skip=none
FM_TEST_BEGIN 2026-10-16T22:48:29Z tests/fm-supervision-events.test.sh family=watcher-wake-lock expected_gate_skip=none
ok - handle_push_transition: a blocked crew enqueues a stale wake naming its window and wakes the supervisor
ok - handle_push_transition: enqueue failure cannot commit the Herdr dedupe marker
ok - handle_push_transition: a declared-pause crew is absorbed (no fast wake), left to the poll loop's long cadence
ok - handle_push_transition: a captain-held crew is absorbed (no fast wake), left to the poll loop's long cadence
ok - event_wait_or_sleep: herdr windows go on the event pane list, but kind=secondmate endpoints are excluded
ok - event_wait_or_sleep: one cached capability probe owns validation across bounded waits
ok - event_wait_or_sleep: a home with no push-capable window is inert (sleeps POLL, never touches the event path)
ok - event_wait_or_sleep: consecutive event-path failures disable the fast-path and revert to pure polling (fail-closed)
# fm-supervision-events.test.sh: all assertions passed
FM_TEST_END 2026-10-16T22:48:29Z tests/fm-supervision-events.test.sh exit=0 duration_ms=495 gate_skip=false
ok - session-lock: a version-named Claude Code session is identified from its install path and argv[0]
ok - session-lock: a harness that is pid 1 of its own namespace is examined, not skipped
ok - session-lock: ordinary script paths under a harness directory are not harness processes
ok - session-lock: ownership stops at the first non-harness gap above the contiguous run
ok - session-lock: a live version-named session holding the lock is not mistaken for a stale owner
ok - session-lock: a trusted same-session id keeps owning a recycled background chain, and nothing weaker does
ok - session-lock: a trusted id anchors the lock on the model-loop process, anything else on the outermost pid
ok - session-lock e2e: a version-named session claims the home and arms supervision
ok - session-lock e2e: a session parented by a harness-named daemon claims the home and arms supervision
ok - session-lock e2e: a version-named session under a harness-named daemon keeps its own lock
not ok - the pty-host was not reparented to init after the daemon ended
FM_TEST_END 2026-10-16T22:49:37Z tests/fm-session-lock-ancestry.test.sh exit=1 duration_ms=67666 gate_skip=false
FM_TEST_SUMMARY total=2 failed=1 skipped_gate=0 duration_ms=67774
FM_TEST_SUMMARY_FAMILY family=watcher-wake-lock count=2 duration_ms=68161 failed=1
FM_TEST_SLOWEST rank=1 script=tests/fm-session-lock-ancestry.test.sh duration_ms=67666
FM_TEST_SLOWEST rank=2 script=tests/fm-supervision-events.test.sh duration_ms=495

- Outcome: ⚠️ 2 warnings across 1 run (7m38s)

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

⚠️ **Test** - 2 warnings

• ⚠️ tests/fm-test-run.test.sh - The focused runner contract test reached and passed the changed packing scenarios, then failed an unrelated session-lock fixture because its fake PTY host was not reparented to init after daemon exit. This reproduced on the captured rerun and requires a host/process-namespace environment where orphaned fixture children are adopted as the test expects.
• ⚠️ live validation verdict: inconclusive (2 of 3 scenarios were driven live against the product); untested: Actual GitHub 30-minute shard completes after rebalancing
• Live validation: ⚠️ inconclusive - 2 of 3 scenarios driven live against the product

Scenario	Result	Live	Evidence
Runner reports a serial packing model below the revised budget	✅ pass	live	portable-serial-coverage.txt
Slow supervision-host suite is isolated in its own portable serial shard	✅ pass	live	supervision-host-lane.txt
Actual GitHub 30-minute shard completes after rebalancing	⏸️ untested	no	No authority in this test phase to dispatch or control the remote CI workflow; run the branch’s GitHub Actions portable-serial jobs to measure the real runner duration.

• bin/fm-test-run.sh --check-coverage
• Enumerated bin/fm-test-run.sh --list --lane portable-serial-{1..9}of9 to locate fm-supervision-host.test.sh
• tests/fm-test-run.test.sh (focused runner contract; changed packing assertions passed, later unrelated PTY reparenting fixture failed)
• git diff ba356f1cd9002b3c8432ac8a409baf216f2697c5 | jev find "a change unrelated to raising the portable serial modeled budget and isolating the slow supervision host suite" (no unrelated hunk flagged)

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.

[chatgpt-codex-connector]

Codex Review Summary

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-10-16T23:02:52.725412Z	aebe736	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "@codex review" or "@codex security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.