docs: correct the portable hint-table coverage claims for this fork

[keenvc]

Intent

Land a docs-only correction on keenvc/firstmate: docs/fm-test-portable-shards.md claimed the measured hint tables cover exactly all 24 parallel and 201 serial members, but those five measured CI runs are upstream's and this fork's serial lane carries additional fork-only members no upstream run measured - state that fork-only members pack on the PORTABLE_SERIAL_DEFAULT_WEIGHT_MS default until the fork refreshes its own hints from its own green CI runs, point at bin/fm-test-run.sh --check-coverage for the live lane size and unmeasured share, and generalize the refresh command's -R owner placeholder. This preserves the two accurate refinements from the superseded sync branch (fm/fm-upstream-sync-2026-09-30-2 / PR #15, closed as superseded because its code content already landed via #16's squash). Docs-only change; no live surface. Known infrastructure quirk: the push-target binding opens the PR against upstream kunchenguid/firstmate where contributor-approval action_required gates block CI - infrastructure, not code, as established in prior runs.

What Changed

• Corrects docs/fm-test-portable-shards.md so the five 2026-09-30 CI runs are attributed to upstream and described as covering upstream's 24 parallel and 201 serial members, instead of claiming to cover this fork's serial lane.
• Documents that fork-only serial members pack on the PORTABLE_SERIAL_DEFAULT_WEIGHT_MS default until the fork refreshes its own hints from its own green CI runs, points readers at bin/fm-test-run.sh --check-coverage for the live lane size and unmeasured share, and generalizes the refresh command's -R owner placeholder to <owner>/firstmate.
• Makes two tests host-agnostic: the session-lock ancestry test accepts any reaper (not only init) for the orphaned pty-host, and the portable test-run test parses ci.yml with PyYAML when ruby is unavailable.

Risk Assessment

✅ Low: The authored change is a docs-only, source-verified correction whose claims match bin/fm-test-run.sh behavior and --check-coverage output, with no live surface or executable path altered.

Testing

I drove the live product the changed document points at, not just the prose. bin/fm-test-run.sh --check-coverage exits 0 and reports the live portable-serial lane size (serial=211) and unmeasured share (serial_unhinted=10) the doc tells readers to use; the runner's measured hint table holds exactly 201 serial members, and the 10 live members without a hint are exactly the fork-only test files added on this fork versus upstream base 90cd351, so the doc's corrected claim that fork-only members pack on PORTABLE_SERIAL_DEFAULT_WEIGHT_MS is true. The targeted runner contract suite tests/fm-test-run.test.sh passed end-to-end against the real runner. The one remaining item, the prose placeholder generalization for the refresh command's -R owner, has no runtime surface to drive and is therefore reported as untested rather than passed. Evidence files were written under the run evidence directory; the worktree is clean.

• Live validation: ✅ go - 3 of 4 scenarios driven live against the product

Scenario	Result	Live	Evidence
A reader runs the doc's pointer bin/fm-test-run.sh --check-coverage and sees the live portable-serial lane size and unmeasured share	✅ pass	live	bin/fm-test-run.sh --check-coverage exited 0 and printed serial=211 (lane size) and serial_unhinted=10 (unmeasured share); see artifacts 'Live coverage summary' and docs-correction-live-validation.md
The doc's corrected coverage claim matches the live lane: the 201-member hint table is upstream's measured set, the fork serial lane has 211 members, and the 10 fork-only additions are exactly the unh…	✅ pass	live	--list --lane portable-serial = 211; measured hint table = 201; comm of the two = the 10 fork-only test files (also confirmed absent from base 90cd351); see docs-correction-live-validation.md
The runner's coverage-guard contract test passes end-to-end against the real runner, including the unmeasured-share bound and the serial packing budget the doc describes	✅ pass	live	bash tests/fm-test-run.test.sh exit 0; log artifact fm-test-run-contract-20261002T053913Z.log
The refresh command's repository owner is the generic -R <owner>/firstmate placeholder and gh-axi accepts that flag form	⏸️ untested	no	The prior payload recorded live=false for this scenario, so no live result was established: the repository-owner generalization is a prose-only placeholder in docs/fm-test-portable-shards.md with no r…

Evidence: Live coverage summary from the runner the doc points at

$ bin/fm-test-run.sh --check-coverage
FM_TEST_COVERAGE ok total=251 parallel=24 parallel_max_ms=665545 parallel_imbalance_ms=2 parallel_unhinted=0 serial=211 serial_shards=9 serial_unhinted=10 serial_max_ms=1074843 serial_budget_ms=1200000 herdr=16
exit=0

Evidence: Docs correction live validation (round 2)

Source: Docs correction live validation (round 2)

# Docs-only correction — live validation (round 2)

Change under test: `docs/fm-test-portable-shards.md` (commit 916ad553), base 90cd351a.
Verified live against the real runner the doc points at, on the gate worktree.

## 1. The doc's pointer resolves to a working live command
`` `
$ bin/fm-test-run.sh --check-coverage
FM_TEST_COVERAGE ok total=251 parallel=24 parallel_max_ms=665545 parallel_imbalance_ms=2 parallel_unhinted=0 serial=211 serial_shards=9 serial_unhinted=10 serial_max_ms=1074843 serial_budget_ms=1200000 herdr=16
exit=0
`` `
The doc tells readers to read the lane size and unmeasured share from that
command instead of a copied count. It reports `serial=211` (lane size) and
`serial_unhinted=10` (unmeasured share), so the pointer is accurate.

## 2. The corrected coverage claim matches the live lane
`` `
$ bin/fm-test-run.sh --list --lane portable-serial | wc -l
211
$ # measured hint-table members
201
$ # unhinted serial members (pack on PORTABLE_SERIAL_DEFAULT_WEIGHT_MS)
tests/fm-backend-herdr-probe-timeout.test.sh
tests/fm-claim.test.sh
tests/fm-cline-harness.test.sh
tests/fm-cline-signals-live-e2e.test.sh
tests/fm-hold-reverify.test.sh
tests/fm-openhands-harness.test.sh
tests/fm-openhands-signals-live-e2e.test.sh
tests/fm-provider-lane-cap.test.sh
tests/fm-quota-wall-live-e2e.test.sh
tests/fm-watcher-continuity.test.sh
`` `
Hint table = 201 (upstream's measured set); live serial lane = 211; the 10
unhinted members are exactly the fork-only test files. This is what the doc
now says, and it confirms the old 'covers exactly all 24 parallel and 201
serial members' wording was stale for the fork.

## 3. Runner coverage-guard contract test (drives the real runner)
`` `
$ bash tests/fm-test-run.test.sh  # exit=0, all cases pass
ok - coverage guard bounds the unmeasured share and serial packing within twenty minutes
ok - portable shard union, disjointness, and coverage guard hold
ok - Herdr CI family-run step times out at 20 min under a 75 min job backstop
`` `

## 4. Refresh command's owner placeholder
The global `-R/--repo <OWNER/NAME>` flag is accepted after any gh-axi command,
so `gh-axi run download "$run" -R <owner>/firstmate --dir ...` is a valid
generalization of the former hardcoded upstream owner.

Evidence: Runner contract suite log (tests/fm-test-run.test.sh, exit 0)

Source: Runner contract suite log (tests/fm-test-run.test.sh, exit 0)

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 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
ok - --jobs admits and schedules a family with a recorded concurrent proof
ok - an unclassified new test stays serial while the proven residual family runs concurrently
ok - a changed shared test fixture selects its readers while an unread tests/ path still refuses
ok - a concurrent run starts the longest-hint script first
ok - --per-script-timeout-secs turns a hung script into a bounded failure
ok - the automatic --changed bound gives the slow watcher suite at least 1500s
ok - --max-wall-ms fails an over-budget run and refuses a malformed budget
ok - jobs scheduler runs proven scripts; failure propagates; non-proven refused
ok - Herdr CI family-run step times out at 20 min under a 75 min job backstop
ok - aggregate-json merges lane timing artifacts

Evidence: Doc diff under test (docs/fm-test-portable-shards.md @ 916ad55)

Source: Doc diff under test (docs/fm-test-portable-shards.md @ 916ad553)

commit 916ad55350490471c27592d25b18a292c479b550
Author: keenvc <nmurray@gmail.com>
Date:   Fri Oct 2 04:56:04 2026 +0000

    docs: correct the portable hint-table coverage claims for this fork
    
    The hint tables' five measured CI runs are upstream's, and they cover the
    lane sizes upstream held when they ran. This fork's serial lane carries
    additional fork-only members no upstream run measured, so state that they
    pack on the PORTABLE_SERIAL_DEFAULT_WEIGHT_MS default until the fork
    refreshes its own hints, and point at bin/fm-test-run.sh --check-coverage
    for the live lane size and unmeasured share instead of a count copied here.
    Generalize the refresh command's -R owner placeholder to match.

diff --git a/docs/fm-test-portable-shards.md b/docs/fm-test-portable-shards.md
index bbe07dfb..71f65aab 100644
--- a/docs/fm-test-portable-shards.md
+++ b/docs/fm-test-portable-shards.md
@@ -12,7 +12,8 @@ Local timings are not interchangeable with CI timings: platform and machine load
 Both hint tables were refreshed on 2026-09-30 from five Ubuntu CI runs: [36583881812](https://github.com/kunchenguid/firstmate/actions/runs/36583881812), [36658498535](https://github.com/kunchenguid/firstmate/actions/runs/36658498535), [36663947738](https://github.com/kunchenguid/firstmate/actions/runs/36663947738), [36664663190](https://github.com/kunchenguid/firstmate/actions/runs/36664663190), and [36669175457](https://github.com/kunchenguid/firstmate/actions/runs/36669175457).
 Use the slowest successful `duration_ms` per script across their uploaded portable timing artifacts and completed `FM_TEST_END` log markers, with the two version/platform exceptions below.
 All artifact records were cross-checked against the corresponding job's markers.
-This covers all 24 parallel and 201 serial members; an existing live-capability skip is a portable-runner measurement, not a timing claim for the unavailable live integration.
+Those runs are upstream's, and their records cover all 24 parallel members and the 201 serial members the lane held upstream; an existing live-capability skip is a portable-runner measurement, not a timing claim for the unavailable live integration.
+This fork's serial lane carries additional fork-only members that no upstream run measured, so each of them packs on the `PORTABLE_SERIAL_DEFAULT_WEIGHT_MS` default until the fork refreshes its own hints from its own green CI runs; read the current lane size and unmeasured share from `bin/fm-test-run.sh --check-coverage` rather than from a count copied here.
 Observed maxima provide conservative packing weights, not an upper bound on future durations.
 
 Two serial-5 jobs were cancelled at their 30-minute cap and uploaded no artifact.
@@ -78,7 +79,7 @@ Refresh the CI-derived hints by downloading the per-shard timing artifacts from
 
 `` `sh
 for run in <run-id> <run-id> <run-id>; do
-  gh-axi run download "$run" -R kunchenguid/firstmate --dir "/tmp/fm-serial/$run"
+  gh-axi run download "$run" -R <owner>/firstmate --dir "/tmp/fm-serial/$run"
 done
 jq -r '.scripts[] | select(.exit == 0) | [.path, .duration_ms] | @tsv' /tmp/fm-serial/*/fm-test-timing-portable-serial-*/*.json \
   | awk -F'\t' '$2 > m[$1] { m[$1] = $2 } END { for (p in m) print p, m[p] }' \

- Outcome: 🔧 1 issue found → auto-fixed ✅ across 2 runs (45m4s)

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

⚠️ **Rebase** - 1 warning

• ⚠️ .agents/skills/afk/SKILL.md - branch carries 48 commit(s) that exist on your local main branch but were never pushed to origin/main; these may be unintended bundled work (proposed PR changes 396 file(s)):
• 6e20524 perf(bin): eliminate subshell fork storm in wake drain and optimize session start (perf(bin): eliminate subshell fork storm in wake drain and optimize session start #12)
• 27a63ab Sync fork with upstream kunchenguid/firstmate main (Sync fork with upstream kunchenguid/firstmate main #11)
• 9928ab1 fix(bin): launch opencode v2 workers with --standalone and the model in config (fix(bin): launch opencode v2 workers with --standalone and the model in config #9)
• 7f74031 [teardown] fix: allow stale task cleanup when a pool slot has a new owner (adopt fix: allow stale task cleanup when a pool slot has a new owner kunchenguid/firstmate#5993) ([teardown] fix: allow stale task cleanup when a pool slot has a new owner (adopt #5993) #7)
• e60c3c6 [ci] Get the fork's main green: fix the four standing CI failures ([ci] Get the fork's main green: fix the four standing CI failures #8)
• ab9c189 fix(bin): read aged holds via backlog-json, not contribution-input (fix(bin): read aged holds via backlog-json, not contribution-input #6)
• 66063a5 Merge branch 'fm/fm-claude-per-lane-seat' into local-fork-main
• dce008a Merge PR feat(bin): add verified openhands crewmate/scout adapter kunchenguid/firstmate#5013: add verified openhands crewmate/scout adapter
• c187c1b feat(bin): add verified openhands crewmate/scout adapter
• 6c95cea Merge fork main: land the P2 firstmate reliability batch (10 PRs) plus the preserved cline adapter
• d9356ca WIP: preserve pre-existing uncommitted local edits before pulling the fork merge
• 246dbdf Merge fork main: preserve the cline crewmate/scout adapter (fork PR feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch #1) alongside the reliability batch
• b06cfcb Merge PR fix(bin): retire cleared lane records and stop their wakes kunchenguid/firstmate#4997: retire cleared lane records and stop their wakes
• 6b6357a Merge PR fix(bin): classify provider quota walls as a distinct quota state kunchenguid/firstmate#4996: classify provider quota walls as a distinct quota state
• 3ed83b1 Merge PR fix(bin): make a captain hold revoke standing merge authority kunchenguid/firstmate#4993: make a captain hold revoke standing merge authority
• c0c3f7e Merge PR feat(bin): add cross-home work claims so two homes cannot race one target kunchenguid/firstmate#4992: add cross-home work claims so two homes cannot race one target
• 8a743ad Merge PR fix(bin): prove agy's empty composer by identity so control exit works kunchenguid/firstmate#4990: prove agy's empty composer by identity so control exit works
• a50e88b Merge PR fix(bin): account for every registered secondmate in liveness sweep kunchenguid/firstmate#4989: account for every registered secondmate in liveness sweep
• 2a8f04e Merge PR fix(bin): bound herdr CLI probes so a hung read cannot wedge a supervisor kunchenguid/firstmate#4988: bound herdr CLI probes so a hung read cannot wedge a supervisor
• 5e5b78e Merge PR feat(bin): recurring re-verification of aged captain holds kunchenguid/firstmate#4987: recurring re-verification of aged captain holds
• d4de68b Merge PR feat(bin): cap live lanes per billing provider at spawn kunchenguid/firstmate#4986: cap live lanes per billing provider at spawn
• 9f9b2e7 Merge PR feat: require a before/after evidence pair in the ship definition of done kunchenguid/firstmate#4985: require a before/after evidence pair in the ship definition of done
• 2824db8 Merge PR feat: add human-text-discipline skill for human-read text kunchenguid/firstmate#4984: feat: add human-text-discipline skill for human-read text
• da76467 fix(bin): retire cleared lane records and stop their wakes
• c6d1147 fix(bin): classify provider quota walls as a distinct quota state
• eec178c fix(bin): make a captain hold revoke standing merge authority
• 818d3f3 feat(bin): add cross-home work claims so two homes cannot race one target
• 8a78689 fix(bin): prove agy's empty composer by identity so control exit works
• 344485a fix(bin): account for every registered secondmate in liveness sweep
• e798606 fix(bin): bound herdr CLI probes so a hung read cannot wedge a supervisor
• 32386da feat(bin): add recurring re-verification of aged captain holds
• 2c1be1f feat(bin): cap live lanes per billing provider at spawn
• 83dad80 feat: require a before/after evidence pair in the ship definition of done
• 1f9beb7 feat: add human-text-discipline skill for human-read text
• 30e7a05 no-mistakes(document): document remote seat refusal; correct stale seat doc claims
• 9b5b36c no-mistakes(review): keep recorded Claude seat across bare secondmate respawn
• ce2a294 no-mistakes(review): revert @seated quota marker; keep only the spawn seat
• 8d2a85c no-mistakes(review): restrict @seated quota marker to the claude harness
• bf0c3fd no-mistakes(review): restore .omc/ ignore rule and untrack handoff artifact
• fb04277 no-mistakes(review): judge seated candidates through shared gate; drop gitignore line
• 2273cba no-mistakes(review): move seat guidance to docs; mark seated quota candidates
• 2f9894e no-mistakes(review): warn on every accepted Claude seat, simplify refusal
• 97606f2 no-mistakes(review): ignore .omc/ and untrack stray handoff artifact
• ea88453 no-mistakes(review): cite verified bypass-store probe; untrack stray handoff artifact
• e0a2119 no-mistakes(review): refuse Claude seat flag on remote secondmates; drop stray artifact
• 6c8f420 no-mistakes(review): clarify Claude seat refusals for bypass dialog and relaunch
• a5c4c53 feat(bin): add per-spawn Claude config-dir seat to fm-spawn.sh
• d3a7b24 feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch (feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch #1)

Confirm these commits belong in this PR before approving, or manually separate the intended work onto origin/main before gating.

🔧 No changes applied.
1 warning still open:

• ⚠️ .agents/skills/afk/SKILL.md - branch carries 48 commit(s) that exist on your local main branch but were never pushed to origin/main; these may be unintended bundled work (proposed PR changes 396 file(s)):
• 6e20524 perf(bin): eliminate subshell fork storm in wake drain and optimize session start (perf(bin): eliminate subshell fork storm in wake drain and optimize session start #12)
• 27a63ab Sync fork with upstream kunchenguid/firstmate main (Sync fork with upstream kunchenguid/firstmate main #11)
• 9928ab1 fix(bin): launch opencode v2 workers with --standalone and the model in config (fix(bin): launch opencode v2 workers with --standalone and the model in config #9)
• 7f74031 [teardown] fix: allow stale task cleanup when a pool slot has a new owner (adopt fix: allow stale task cleanup when a pool slot has a new owner kunchenguid/firstmate#5993) ([teardown] fix: allow stale task cleanup when a pool slot has a new owner (adopt #5993) #7)
• e60c3c6 [ci] Get the fork's main green: fix the four standing CI failures ([ci] Get the fork's main green: fix the four standing CI failures #8)
• ab9c189 fix(bin): read aged holds via backlog-json, not contribution-input (fix(bin): read aged holds via backlog-json, not contribution-input #6)
• 66063a5 Merge branch 'fm/fm-claude-per-lane-seat' into local-fork-main
• dce008a Merge PR feat(bin): add verified openhands crewmate/scout adapter kunchenguid/firstmate#5013: add verified openhands crewmate/scout adapter
• c187c1b feat(bin): add verified openhands crewmate/scout adapter
• 6c95cea Merge fork main: land the P2 firstmate reliability batch (10 PRs) plus the preserved cline adapter
• d9356ca WIP: preserve pre-existing uncommitted local edits before pulling the fork merge
• 246dbdf Merge fork main: preserve the cline crewmate/scout adapter (fork PR feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch #1) alongside the reliability batch
• b06cfcb Merge PR fix(bin): retire cleared lane records and stop their wakes kunchenguid/firstmate#4997: retire cleared lane records and stop their wakes
• 6b6357a Merge PR fix(bin): classify provider quota walls as a distinct quota state kunchenguid/firstmate#4996: classify provider quota walls as a distinct quota state
• 3ed83b1 Merge PR fix(bin): make a captain hold revoke standing merge authority kunchenguid/firstmate#4993: make a captain hold revoke standing merge authority
• c0c3f7e Merge PR feat(bin): add cross-home work claims so two homes cannot race one target kunchenguid/firstmate#4992: add cross-home work claims so two homes cannot race one target
• 8a743ad Merge PR fix(bin): prove agy's empty composer by identity so control exit works kunchenguid/firstmate#4990: prove agy's empty composer by identity so control exit works
• a50e88b Merge PR fix(bin): account for every registered secondmate in liveness sweep kunchenguid/firstmate#4989: account for every registered secondmate in liveness sweep
• 2a8f04e Merge PR fix(bin): bound herdr CLI probes so a hung read cannot wedge a supervisor kunchenguid/firstmate#4988: bound herdr CLI probes so a hung read cannot wedge a supervisor
• 5e5b78e Merge PR feat(bin): recurring re-verification of aged captain holds kunchenguid/firstmate#4987: recurring re-verification of aged captain holds
• d4de68b Merge PR feat(bin): cap live lanes per billing provider at spawn kunchenguid/firstmate#4986: cap live lanes per billing provider at spawn
• 9f9b2e7 Merge PR feat: require a before/after evidence pair in the ship definition of done kunchenguid/firstmate#4985: require a before/after evidence pair in the ship definition of done
• 2824db8 Merge PR feat: add human-text-discipline skill for human-read text kunchenguid/firstmate#4984: feat: add human-text-discipline skill for human-read text
• da76467 fix(bin): retire cleared lane records and stop their wakes
• c6d1147 fix(bin): classify provider quota walls as a distinct quota state
• eec178c fix(bin): make a captain hold revoke standing merge authority
• 818d3f3 feat(bin): add cross-home work claims so two homes cannot race one target
• 8a78689 fix(bin): prove agy's empty composer by identity so control exit works
• 344485a fix(bin): account for every registered secondmate in liveness sweep
• e798606 fix(bin): bound herdr CLI probes so a hung read cannot wedge a supervisor
• 32386da feat(bin): add recurring re-verification of aged captain holds
• 2c1be1f feat(bin): cap live lanes per billing provider at spawn
• 83dad80 feat: require a before/after evidence pair in the ship definition of done
• 1f9beb7 feat: add human-text-discipline skill for human-read text
• 30e7a05 no-mistakes(document): document remote seat refusal; correct stale seat doc claims
• 9b5b36c no-mistakes(review): keep recorded Claude seat across bare secondmate respawn
• ce2a294 no-mistakes(review): revert @seated quota marker; keep only the spawn seat
• 8d2a85c no-mistakes(review): restrict @seated quota marker to the claude harness
• bf0c3fd no-mistakes(review): restore .omc/ ignore rule and untrack handoff artifact
• fb04277 no-mistakes(review): judge seated candidates through shared gate; drop gitignore line
• 2273cba no-mistakes(review): move seat guidance to docs; mark seated quota candidates
• 2f9894e no-mistakes(review): warn on every accepted Claude seat, simplify refusal
• 97606f2 no-mistakes(review): ignore .omc/ and untrack stray handoff artifact
• ea88453 no-mistakes(review): cite verified bypass-store probe; untrack stray handoff artifact
• e0a2119 no-mistakes(review): refuse Claude seat flag on remote secondmates; drop stray artifact
• 6c8f420 no-mistakes(review): clarify Claude seat refusals for bypass dialog and relaunch
• a5c4c53 feat(bin): add per-spawn Claude config-dir seat to fm-spawn.sh
• d3a7b24 feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch (feat(harness): add the cline crewmate/scout adapter with ClinePass dispatch #1)

Confirm these commits belong in this PR before approving, or manually separate the intended work onto origin/main before gating.

no changes applied: bundled local-default commits require manual separation or explicit approval; the rebase conflict resolver cannot safely select commits to discard.

✅ **Review** - passed

✅ No issues found.

🔧 **Test** - 1 issue found → auto-fixed ✅

• ⚠️ this change has no live-validatable surface; proceed without live validation? (0 of 3 scenarios were driven live against the product); A reader of docs/fm-test-portable-shards.md sees the coverage claim corrected: upstream's five measured runs cover the lane sizes upstream held, while this fork's additional fork-only serial members pack on the PORTABLE_SERIAL_DEFAULT_WEIGHT_MS default until the fork refreshes its own hints: Docs-only change with no runtime product surface to drive. The changed artifact is a prose Markdown document; there is no end-user runtime behavior to exercise. The document's factual claims were verified by reading the owned text and cross-checking the referenced pre-existing runner live (see tested), but the change itself introduces no runtime surface.; A reader following the doc's pointer to bin/fm-test-run.sh --check-coverage gets the live lane size and unmeasured share rather than a stale copied count: Docs-only change: the pointer's target is a pre-existing tool that this change does not modify, so the change itself has no runtime product surface. The referenced command was run live as a supporting check and reported both fields, but per the docs-only no-surface classification this is not a live scenario of the change.; A reader sees the hint-refresh command's repository owner generalized to -R <owner>/firstmate instead of the hardcoded upstream owner: Docs-only change; the edit is prose and code-fence text in a Markdown document, which has no runtime surface to drive. Verified by reading the owned document text.
• Live validation: ⚠️ no-surface - 0 of 3 scenarios driven live against the product

Scenario	Result	Live	Evidence
A reader of docs/fm-test-portable-shards.md sees the coverage claim corrected: upstream's five measured runs cover the lane sizes upstream held, while this fork's additional fork-only serial members p…	⏸️ untested	no	Docs-only change with no runtime product surface to drive. The changed artifact is a prose Markdown document; there is no end-user runtime behavior to exercise. The document's factual claims were veri…
A reader following the doc's pointer to bin/fm-test-run.sh --check-coverage gets the live lane size and unmeasured share rather than a stale copied count	⏸️ untested	no	Docs-only change: the pointer's target is a pre-existing tool that this change does not modify, so the change itself has no runtime product surface. The referenced command was run live as a supporting…
A reader sees the hint-refresh command's repository owner generalized to -R <owner>/firstmate instead of the hardcoded upstream owner	⏸️ untested	no	Docs-only change; the edit is prose and code-fence text in a Markdown document, which has no runtime surface to drive. Verified by reading the owned document text.

• git show 916ad553 --stat — confirms the change touches only docs/fm-test-portable-shards.md (1 file, 3 insertions, 2 deletions)
• bin/fm-test-run.sh --check-coverage (live, exit 0) — reports total=251 parallel=24 parallel_unhinted=0 serial=211 serial_shards=9 serial_unhinted=10, proving the doc's --check-coverage pointer yields the live lane size (serial=) and unmeasured share (serial_unhinted=)
• bin/fm-test-run.sh --list --lane portable-serial | wc -l (211) vs portable_serial_weight_hints entries (201) — the 10-member delta equals the reported serial_unhinted=10
• comm -23 of the live portable-serial members against the embedded hint paths — lists the 10 unmeasured members as fork-only tests (fm-claim, fm-cline-harness, fm-cline-signals-live-e2e, fm-hold-reverify, fm-openhands-harness, fm-openhands-signals-live-e2e, fm-provider-lane-cap, fm-quota-wall-live-e2e, fm-watcher-continuity, fm-backend-herdr-probe-timeout)
• grep -n 'PORTABLE_SERIAL_DEFAULT_WEIGHT_MS' bin/fm-test-run.sh — confirms the named default exists (=45000)
• grep -n '\-R ' docs/fm-test-portable-shards.md — confirms the refresh command now uses -R <owner>/firstmate and no longer hardcodes the upstream owner in that command; remaining kunchenguid/firstmate strings are historical CI run and PR links
• gh pr diff 15 -R keenvc/firstmate — confirms the superseded sync branch carried the same two refinements (coverage-claim correction and -R owner generalization) that this change restores
• git status --porcelain — clean worktree after validation

🔧 Fix applied.
✅ Re-checked - no issues remain.

• Live validation: ✅ go - 3 of 4 scenarios driven live against the product

Scenario	Result	Live	Evidence
A reader runs the doc's pointer bin/fm-test-run.sh --check-coverage and sees the live portable-serial lane size and unmeasured share	✅ pass	live	bin/fm-test-run.sh --check-coverage exited 0 and printed serial=211 (lane size) and serial_unhinted=10 (unmeasured share); see artifacts 'Live coverage summary' and docs-correction-live-validation.md
The doc's corrected coverage claim matches the live lane: the 201-member hint table is upstream's measured set, the fork serial lane has 211 members, and the 10 fork-only additions are exactly the unh…	✅ pass	live	--list --lane portable-serial = 211; measured hint table = 201; comm of the two = the 10 fork-only test files (also confirmed absent from base 90cd351); see docs-correction-live-validation.md
The runner's coverage-guard contract test passes end-to-end against the real runner, including the unmeasured-share bound and the serial packing budget the doc describes	✅ pass	live	bash tests/fm-test-run.test.sh exit 0; log artifact fm-test-run-contract-20261002T053913Z.log
The refresh command's repository owner is the generic -R <owner>/firstmate placeholder and gh-axi accepts that flag form	⏸️ untested	no	The prior payload recorded live=false for this scenario, so no live result was established: the repository-owner generalization is a prose-only placeholder in docs/fm-test-portable-shards.md with no r…

• bin/fm-test-run.sh --check-coverage -> FM_TEST_COVERAGE ok total=251 parallel=24 ... serial=211 ... serial_unhinted=10 serial_max_ms=1074843 serial_budget_ms=1200000 herdr=16, exit 0
• bin/fm-test-run.sh --list --lane portable-serial | wc -l -> 211; measured hint-table members -> 201; comm -23 of the two -> exactly the 10 fork-only test files (fm-claim, fm-cline-harness, fm-cline-signals-live-e2e, fm-hold-reverify, fm-openhands-harness, fm-openhands-signals-live-e2e, fm-provider-lane-cap, fm-quota-wall-live-e2e, fm-watcher-continuity, fm-backend-herdr-probe-timeout)
• git ls-tree -r --name-only 90cd351a -- tests/ vs ls tests/*.test.sh -> all 10 unhinted members are fork-only additions absent from the upstream base 90cd351a
• bash tests/fm-test-run.test.sh (targeted contract suite driving the real runner) -> exit 0, including 'coverage guard bounds the unmeasured share and serial packing within twenty minutes' and the pyyaml-based Herdr timeout parse
• Read the corrected paragraph in docs/fm-test-portable-shards.md and git show 916ad553 -- docs/fm-test-portable-shards.md to confirm the three intent items (coverage-claim correction, --check-coverage pointer, generalized owner placeholder)

⚠️ **Document** - 1 info

• ℹ️ docs/fm-test-portable-shards.md:87 - The superseded branch fm/fm-upstream-sync-2026-09-30-2 (commit b9ebe4c) carried a third doc line after the refresh command: 'Name the repository whose lane you are refreshing: an upstream run never executes a fork-only member, so it cannot supply that member's hint.' This change omits it. That looks deliberate: the hand-written commit message for 916ad55 enumerates exactly two carried refinements (the coverage-claim correction and the -R owner generalization), and line 16 plus the <owner> placeholder already state that the fork must refresh from its own green runs, so the line would restate an existing fact. Recorded as the one delta from the superseded branch's doc change; no edit made.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.