diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml
index f9465522f50..c3e072160ed 100644
--- a/.github/workflows/ci.yml
+++ b/.github/workflows/ci.yml
@@ -192,16 +192,14 @@ jobs:
   tests-portable-serial:
     name: Behavior portable serial ${{ matrix.shard }}
     runs-on: ubuntu-latest
-    # Refreshed timings plus conservative hints for grown tests estimate ~84 min
-    # of serial work, balanced over six runners at ~14 min each. Keep the cap
-    # as a hang tripwire with runner-load margin; refresh hints per
-    # docs/fm-test-portable-shards.md when a shard approaches it.
+    # Keep this cap as a hang tripwire; when a shard approaches it, follow the
+    # balance evidence and refresh procedure in docs/fm-test-portable-shards.md.
     timeout-minutes: 20
     strategy:
       # Every shard reports so one failure never hides another shard's result.
       fail-fast: false
       matrix:
-        shard: [1, 2, 3, 4, 5, 6]
+        shard: [1, 2, 3, 4, 5, 6, 7]
     steps:
       - uses: actions/checkout@v6
         with:
diff --git a/bin/fm-test-run.sh b/bin/fm-test-run.sh
index 7de253f58e6..afc6dd3034a 100755
--- a/bin/fm-test-run.sh
+++ b/bin/fm-test-run.sh
@@ -92,11 +92,9 @@ JOBS_MAX=8
 
 # How many separate-runner shards the portable serial remainder splits into.
 # One owner: CI lane names carry this count and are refused when they disagree.
-PORTABLE_SERIAL_SHARDS=6
+PORTABLE_SERIAL_SHARDS=7
 
-# Balance hint for a portable-serial script with no measured duration, close to
-# the measured per-script mean so a newly added test neither starves nor
-# overloads the shard it lands in.
+# Fallback balance hint for a portable-serial script with no measured duration.
 PORTABLE_SERIAL_DEFAULT_WEIGHT_MS=20000
 
 usage() {
@@ -375,165 +373,179 @@ list_portable_serial() {
   done < <(all_repo_tests)
 }
 
-# Portable-serial duration hints in milliseconds, from CI artifacts, completed
-# log records, and conservative estimates documented in docs/fm-test-portable-shards.md.
+# Portable-serial measured duration hints in milliseconds; source evidence is
+# documented in docs/fm-test-portable-shards.md.
 # These affect balance only: the shard partition stays complete and disjoint, so a
 # stale hint costs balance rather than coverage. That doc owns the refresh
 # procedure.
 portable_serial_weight_hints() {
   cat <<'EOF'
-tests/fm-afk-inject-e2e.test.sh 26008
-tests/fm-afk-pi-herdr-return-e2e.test.sh 47
-tests/fm-afk-return.test.sh 1517
-tests/fm-ask-user-authority.test.sh 138
-tests/fm-backend-cmux-smoke.test.sh 30
-tests/fm-backend-cmux.test.sh 2950
-tests/fm-backend-herdr-focus-flash-e2e.test.sh 21
-tests/fm-backend-herdr-server-detach.test.sh 90000
-tests/fm-backend-orca.test.sh 19416
-tests/fm-backend-tmux-smoke.test.sh 315
-tests/fm-backend-zellij-smoke.test.sh 14
-tests/fm-backend-zellij.test.sh 5188
-tests/fm-backend.test.sh 19862
-tests/fm-backlog-atomicity.test.sh 78496
-tests/fm-backlog-handoff.test.sh 3343
-tests/fm-bearings-snapshot.test.sh 66118
-tests/fm-boat-lifecycle.test.sh 52034
-tests/fm-boat-omp-auth.test.sh 58941
-tests/fm-boat-routing.test.sh 200000
-tests/fm-bootstrap.test.sh 62184
-tests/fm-busy-adapter-wiring.test.sh 17306
-tests/fm-busy-state.test.sh 2821
-tests/fm-calm-pi-extension.test.sh 260
-tests/fm-check-unregister.test.sh 356
-tests/fm-classify-decision-key.test.sh 703
-tests/fm-claude-stop-autoarm-live-e2e.test.sh 17
-tests/fm-claude-stop-autoarm.test.sh 60519
-tests/fm-cleanup-skill.test.sh 72
-tests/fm-codex-continuity-live-e2e.test.sh 19
-tests/fm-daemon.test.sh 61329
-tests/fm-devin-adapter.test.sh 280
-tests/fm-devin-herdr-live-e2e.test.sh 13
-tests/fm-dispatch-resolve.test.sh 3995
-tests/fm-documentation-audiences.test.sh 708
-tests/fm-ext-bridge.test.sh 35693
-tests/fm-fleet-snapshot-view.test.sh 13508
-tests/fm-fleet-sync.test.sh 32815
-tests/fm-gate-refuse.test.sh 9270
-tests/fm-gitignore-config.test.sh 25
-tests/fm-gotmp.test.sh 713
-tests/fm-grok-continuity-live-e2e.test.sh 12
-tests/fm-grok-stop-live-e2e.test.sh 21
-tests/fm-guard-stale-banner.test.sh 4698
-tests/fm-herdr-session-cleanup.test.sh 4832
-tests/fm-herdr-submit-confirm-live-e2e.test.sh 21
-tests/fm-hermes-harness.test.sh 146669
-tests/fm-hermes-live-e2e.test.sh 28
-tests/fm-idle-sleep-watch.test.sh 8000
-tests/fm-idle-sleep.test.sh 15000
-tests/fm-inbox-result.test.sh 7377
-tests/fm-inbox.test.sh 7022
-tests/fm-kimi-harness.test.sh 16139
-tests/fm-omp-branch-bash.test.sh 44
-tests/fm-omp-branch-live-e2e.test.sh 12
-tests/fm-omp-branch-supervision.test.sh 4373
-tests/fm-omp-branch-types.test.sh 24
+tests/fm-afk-inject-e2e.test.sh 26538
+tests/fm-afk-pi-herdr-return-e2e.test.sh 76
+tests/fm-afk-return.test.sh 1524
+tests/fm-ask-user-authority.test.sh 154
+tests/fm-backend-cmux-smoke.test.sh 34
+tests/fm-backend-cmux.test.sh 3751
+tests/fm-backend-herdr-focus-flash-e2e.test.sh 22
+tests/fm-backend-herdr-server-detach.test.sh 2649
+tests/fm-backend-orca.test.sh 19716
+tests/fm-backend-tmux-smoke.test.sh 250
+tests/fm-backend-zellij-smoke.test.sh 22
+tests/fm-backend-zellij.test.sh 7129
+tests/fm-backend.test.sh 23256
+tests/fm-backlog-atomicity.test.sh 63215
+tests/fm-backlog-handoff.test.sh 3410
+tests/fm-bearings-snapshot.test.sh 81131
+tests/fm-boat-lane-runner.test.sh 119102
+tests/fm-boat-lifecycle.test.sh 44937
+tests/fm-boat-omp-auth.test.sh 62445
+tests/fm-boat-routing.test.sh 26287
+tests/fm-bootstrap.test.sh 71450
+tests/fm-busy-adapter-wiring.test.sh 19619
+tests/fm-busy-state.test.sh 2879
+tests/fm-calm-pi-extension.test.sh 265
+tests/fm-check-unregister.test.sh 436
+tests/fm-classify-decision-key.test.sh 501
+tests/fm-claude-stop-autoarm-live-e2e.test.sh 20
+tests/fm-claude-stop-autoarm.test.sh 60780
+tests/fm-cleanup-skill.test.sh 74
+tests/fm-codex-continuity-live-e2e.test.sh 20
+tests/fm-daemon.test.sh 60677
+tests/fm-devin-adapter.test.sh 349
+tests/fm-devin-herdr-live-e2e.test.sh 21
+tests/fm-dispatch-resolve.test.sh 8747
+tests/fm-documentation-audiences.test.sh 878
+tests/fm-ext-ask.test.sh 48724
+tests/fm-ext-bridge.test.sh 82962
+tests/fm-ext-decision.test.sh 161682
+tests/fm-ext-plugin-dest.test.sh 7190
+tests/fm-ext-thread-context.test.sh 18046
+tests/fm-ext-topic-threads.test.sh 100660
+tests/fm-ext-waiting-pin.test.sh 103172
+tests/fm-ext-watch.test.sh 7522
+tests/fm-fleet-snapshot-view.test.sh 12998
+tests/fm-fleet-sync.test.sh 36580
+tests/fm-gate-refuse.test.sh 11343
+tests/fm-gitignore-config.test.sh 29
+tests/fm-gotmp.test.sh 1168
+tests/fm-grok-continuity-live-e2e.test.sh 19
+tests/fm-grok-stop-live-e2e.test.sh 20
+tests/fm-guard-stale-banner.test.sh 5898
+tests/fm-herdr-session-cleanup.test.sh 6849
+tests/fm-herdr-submit-confirm-live-e2e.test.sh 22
+tests/fm-hermes-harness.test.sh 151616
+tests/fm-hermes-live-e2e.test.sh 44
+tests/fm-idle-sleep-watch.test.sh 4560
+tests/fm-idle-sleep.test.sh 13959
+tests/fm-inbox-result.test.sh 9828
+tests/fm-inbox.test.sh 9840
+tests/fm-kimi-harness.test.sh 18045
+tests/fm-landing.test.sh 43476
+tests/fm-main-ci-watch.test.sh 46302
+tests/fm-merge-local.test.sh 2502
+tests/fm-omp-branch-bash.test.sh 41
+tests/fm-omp-branch-live-e2e.test.sh 20
+tests/fm-omp-branch-supervision.test.sh 37015
+tests/fm-omp-branch-types.test.sh 21
 tests/fm-omp-compact-adviser-live-e2e.test.sh 20
-tests/fm-omp-compact-adviser.test.sh 298
-tests/fm-omp-fleet-hooks.test.sh 2407
-tests/fm-omp-harness.test.sh 1764
-tests/fm-omp-herdr-exit-live-e2e.test.sh 16
+tests/fm-omp-compact-adviser.test.sh 1697
+tests/fm-omp-fleet-hooks.test.sh 2743
+tests/fm-omp-harness.test.sh 66500
+tests/fm-omp-herdr-exit-live-e2e.test.sh 19
 tests/fm-omp-herdr-live-e2e.test.sh 19
-tests/fm-omp-primary-live-e2e.test.sh 18
-tests/fm-omp-primary.test.sh 45746
-tests/fm-omp-relaunch-guard.test.sh 4276
-tests/fm-omp-relaunch-tmux-live-e2e.test.sh 21
-tests/fm-omp-secondmate-live-e2e.test.sh 21
-tests/fm-omp-secondmate.test.sh 600000
-tests/fm-omp-task-inbox-doorbell.test.sh 20577
-tests/fm-omp-update.test.sh 448
-tests/fm-omp-worker-tmux-live-e2e.test.sh 17
-tests/fm-on.test.sh 9627
-tests/fm-opencode-primary-live-e2e.test.sh 16
-tests/fm-operational-input.test.sh 176
-tests/fm-pending-reply-sleep-reconcile.test.sh 243
-tests/fm-pending-reply.test.sh 19949
-tests/fm-pi-compatible-family.test.sh 56
+tests/fm-omp-primary-live-e2e.test.sh 19
+tests/fm-omp-primary.test.sh 61283
+tests/fm-omp-relaunch-guard.test.sh 5534
+tests/fm-omp-relaunch-tmux-live-e2e.test.sh 14
+tests/fm-omp-secondmate-live-e2e.test.sh 20
+tests/fm-omp-secondmate.test.sh 113332
+tests/fm-omp-task-inbox-doorbell.test.sh 20456
+tests/fm-omp-update.test.sh 689
+tests/fm-omp-worker-tmux-live-e2e.test.sh 20
+tests/fm-on.test.sh 9816
+tests/fm-opencode-primary-live-e2e.test.sh 20
+tests/fm-operational-input.test.sh 154
+tests/fm-pending-reply-sleep-reconcile.test.sh 229
+tests/fm-pending-reply.test.sh 15024
+tests/fm-pi-compatible-family.test.sh 70
 tests/fm-pi-primary-live-e2e.test.sh 21
-tests/fm-pi-watch-extension.test.sh 22031
-tests/fm-pr-check-handoff.test.sh 42228
-tests/fm-pr-check-security.test.sh 158907
-tests/fm-prepush-guard.test.sh 19208
-tests/fm-procevent-when.test.sh 24358
-tests/fm-procevent.test.sh 61916
-tests/fm-public-followup.test.sh 35559
-tests/fm-quota-array-dispatch-live-e2e.test.sh 12
-tests/fm-receipt-check.test.sh 11636
-tests/fm-receipt.test.sh 1331
-tests/fm-reflect-skill.test.sh 372
-tests/fm-remote-backlog-handoff.test.sh 15173
-tests/fm-remote-doctor.test.sh 6203
-tests/fm-remote-entrypoint.test.sh 208
-tests/fm-remote-job.test.sh 48694
-tests/fm-remote-reply.test.sh 99366
-tests/fm-remote-secondmate-lifecycle-e2e.test.sh 700000
-tests/fm-remote-secondmate-parent-binding.test.sh 15892
-tests/fm-remote-secondmate-sleep-reconcile.test.sh 3000
-tests/fm-remote-secondmate-trace-context.test.sh 49391
-tests/fm-runpod-lifecycle.test.sh 56695
-tests/fm-runpod-omp-auth.test.sh 1012
-tests/fm-runpod-pod-boot.test.sh 45589
-tests/fm-runpod-routing.test.sh 28030
-tests/fm-secondmate-harness.test.sh 151987
-tests/fm-secondmate-lifecycle-e2e.test.sh 5372
-tests/fm-secondmate-liveness.test.sh 20126
-tests/fm-secondmate-restart.test.sh 31003
-tests/fm-secondmate-safety.test.sh 55903
-tests/fm-secondmate-sync.test.sh 17159
-tests/fm-send-inbox-doorbell-live-e2e.test.sh 18
-tests/fm-send-inbox.test.sh 37394
-tests/fm-send-resolve-key.test.sh 12632
-tests/fm-send-secondmate-marker-herdr-e2e.test.sh 48
-tests/fm-send-secondmate-marker.test.sh 4320
-tests/fm-send-turn-start.test.sh 11143
-tests/fm-session-lock-ancestry.test.sh 1284
-tests/fm-session-start.test.sh 77942
-tests/fm-sessionstart-nudge.test.sh 284
-tests/fm-shared-captain-inheritance.test.sh 4259
-tests/fm-spawn-dispatch-profile.test.sh 193113
-tests/fm-spawn-pool-base-freshen.test.sh 35139
-tests/fm-spawn-relaunch-dead-endpoint.test.sh 31467
-tests/fm-spawn-worktree-settle.test.sh 5788
+tests/fm-pi-watch-extension.test.sh 22125
+tests/fm-pr-check-handoff.test.sh 15330
+tests/fm-pr-check-security.test.sh 163835
+tests/fm-prepush-guard.test.sh 19861
+tests/fm-procevent-when.test.sh 21536
+tests/fm-procevent.test.sh 69535
+tests/fm-public-followup.test.sh 47380
+tests/fm-quota-array-dispatch-live-e2e.test.sh 19
+tests/fm-receipt-check.test.sh 6254
+tests/fm-receipt.test.sh 1854
+tests/fm-reflect-skill.test.sh 473
+tests/fm-remote-backlog-handoff.test.sh 20170
+tests/fm-remote-doctor.test.sh 5787
+tests/fm-remote-entrypoint.test.sh 263
+tests/fm-remote-job.test.sh 48956
+tests/fm-remote-omp-update-restart-live-e2e.test.sh 20
+tests/fm-remote-reply.test.sh 82089
+tests/fm-remote-secondmate-lifecycle-e2e.test.sh 657637
+tests/fm-remote-secondmate-parent-binding.test.sh 15757
+tests/fm-remote-secondmate-sleep-reconcile.test.sh 1287
+tests/fm-remote-secondmate-trace-context.test.sh 59252
+tests/fm-remote-secondmate-update-restart.test.sh 30187
+tests/fm-runpod-lifecycle.test.sh 58778
+tests/fm-runpod-omp-auth.test.sh 1192
+tests/fm-runpod-pod-boot.test.sh 43135
+tests/fm-runpod-routing.test.sh 37449
+tests/fm-secondmate-harness.test.sh 170105
+tests/fm-secondmate-lifecycle-e2e.test.sh 6278
+tests/fm-secondmate-liveness.test.sh 20034
+tests/fm-secondmate-restart.test.sh 36334
+tests/fm-secondmate-safety.test.sh 58742
+tests/fm-secondmate-sync.test.sh 40216
+tests/fm-send-inbox-doorbell-live-e2e.test.sh 22
+tests/fm-send-inbox.test.sh 40199
+tests/fm-send-resolve-key.test.sh 22866
+tests/fm-send-secondmate-marker-herdr-e2e.test.sh 49
+tests/fm-send-secondmate-marker.test.sh 4337
+tests/fm-send-turn-start.test.sh 11083
+tests/fm-session-lock-ancestry.test.sh 1513
+tests/fm-session-start.test.sh 65773
+tests/fm-sessionstart-nudge.test.sh 303
+tests/fm-shared-captain-inheritance.test.sh 13503
+tests/fm-spawn-dispatch-profile.test.sh 249954
+tests/fm-spawn-pool-base-freshen.test.sh 35286
+tests/fm-spawn-relaunch-dead-endpoint.test.sh 35039
+tests/fm-spawn-worktree-settle.test.sh 11488
 tests/fm-stall-recovery.test.sh 48
-tests/fm-startup-memory-budget.test.sh 7593
-tests/fm-subagent-pretool-check.test.sh 709
-tests/fm-supervision-events.test.sh 462
-tests/fm-tangle-guard.test.sh 8797
-tests/fm-task-delivery.test.sh 9829
-tests/fm-task-inbox.test.sh 37450
-tests/fm-teardown-endpoint-safety.test.sh 1797
-tests/fm-teardown.test.sh 107078
-tests/fm-test-fixture-cleanup.test.sh 542
-tests/fm-test-isolation-proof.test.sh 426
-tests/fm-todo-project.test.sh 6065
-tests/fm-trace-context-lib.test.sh 1424
-tests/fm-trace-context-spawn.test.sh 46250
-tests/fm-treehouse-orphan-recovery.test.sh 125
-tests/fm-treehouse-status-read-only.test.sh 225
-tests/fm-treehouse-sweep.test.sh 4886
-tests/fm-turnend-guard.test.sh 20258
-tests/fm-update.test.sh 10591
-tests/fm-vendor-auth-probe.test.sh 42632
-tests/fm-wake-daemon-lifecycle-e2e.test.sh 19948
-tests/fm-wake-drain-open-decisions-cursor.test.sh 22217
-tests/fm-wake-drain-open-decisions.test.sh 3443
-tests/fm-wake-drain-unread-status.test.sh 14508
-tests/fm-wake-queue.test.sh 72347
-tests/fm-watch-arm.test.sh 66052
-tests/fm-watch-checkpoint.test.sh 4768
-tests/fm-watch-recovery-loop.test.sh 67796
-tests/fm-watch-triage.test.sh 407955
-tests/fm-watcher-lock.test.sh 38168
+tests/fm-startup-memory-budget.test.sh 6514
+tests/fm-subagent-pretool-check.test.sh 968
+tests/fm-supervision-events.test.sh 628
+tests/fm-tangle-guard.test.sh 10406
+tests/fm-task-delivery.test.sh 19399
+tests/fm-task-inbox.test.sh 37373
+tests/fm-tasks-axi.test.sh 18227
+tests/fm-teardown-endpoint-safety.test.sh 30883
+tests/fm-teardown.test.sh 137584
+tests/fm-test-fixture-cleanup.test.sh 681
+tests/fm-test-isolation-proof.test.sh 588
+tests/fm-todo-project.test.sh 9125
+tests/fm-trace-context-lib.test.sh 1997
+tests/fm-trace-context-spawn.test.sh 45140
+tests/fm-treehouse-orphan-recovery.test.sh 111
+tests/fm-treehouse-status-read-only.test.sh 223
+tests/fm-treehouse-sweep.test.sh 4347
+tests/fm-turnend-guard.test.sh 21290
+tests/fm-update.test.sh 9518
+tests/fm-vendor-auth-probe.test.sh 42927
+tests/fm-wake-daemon-lifecycle-e2e.test.sh 69696
+tests/fm-wake-drain-open-decisions-cursor.test.sh 26536
+tests/fm-wake-drain-open-decisions.test.sh 12358
+tests/fm-wake-drain-unread-status.test.sh 30489
+tests/fm-wake-queue.test.sh 85314
+tests/fm-watch-arm.test.sh 85430
+tests/fm-watch-checkpoint.test.sh 5858
+tests/fm-watch-recovery-loop.test.sh 66814
+tests/fm-watch-triage.test.sh 428164
+tests/fm-watcher-lock.test.sh 46975
 EOF
 }
 
diff --git a/docs/fm-test-portable-shards.md b/docs/fm-test-portable-shards.md
index 3d45a084456..4b43435cc46 100644
--- a/docs/fm-test-portable-shards.md
+++ b/docs/fm-test-portable-shards.md
@@ -51,37 +51,322 @@ The two parallel lanes use longest-processing-time assignment from those measure
 
 `portable-serial` includes every `tests/*.test.sh` that is neither proven-isolated nor `real-herdr-gated`.
 It keeps watcher, lock, AFK, real tmux, daemon, secondmate lifecycle, bootstrap, live-harness opt-in, GUI-backend, and other unproven work serial.
-Membership is derived rather than enumerated, so a newly added test lands here by default.
+The portable-serial test set is derived rather than enumerated, so a newly added test lands here by default.
+The per-shard membership snapshot below records the generated assignment for the current timing hints.
 
 ## Portable serial CI shards
 
-On green CI run [30725985757](https://github.com/kunchenguid/firstmate/actions/runs/30725985757), that remainder accumulated 19m04s of script time against a 20-minute job timeout.
-On [PR 1495](https://github.com/kunchenguid/firstmate/pull/1495), its main step ran about 19m51s before the job was cancelled at that boundary.
-`portable-serial-<k>of<n>` splits it across `n` separate CI runners.
+The refreshed hints use all 166 completed serial-script measurements from PR run [38774522507](https://github.com/dnth/firstmate/actions/runs/38774522507), totaling 5,461,112 ms (91m01.112s).
+The measured durations replace the older estimates and default weights in `portable_serial_weight_hints`.
+`portable-serial-<k>of<n>` splits that remainder across `n` separate CI runners.
 Each shard is still strictly serial in itself, and separate runners mean no two of these stateful scripts ever share a machine, so the split needs no concurrency isolation proof.
 
 `bin/fm-test-run.sh` owns `n` and refuses any lane whose `of<n>` disagrees with it.
 `.github/workflows/ci.yml` derives the same `n` from `strategy.job-total` rather than a literal, so changing the shard count in either file without the other fails the lane loudly instead of leaving part of the required suite unrun.
 
 Assignment is longest-processing-time bin packing over per-script duration hints embedded in `bin/fm-test-run.sh`.
-The hints were refreshed on 2026-10-04 from 116 completed, passing script timings in attempt 3 of [run 37159651011](https://github.com/dnth/firstmate/actions/runs/37159651011/attempts/3).
-Shards 3 and 4 produced timing artifacts; shards 1 and 2 hit the 20-minute job cap, so their completed `FM_TEST_END` log records supply the measurements.
-Existing hints for uncompleted scripts remain from [run 33904204200](https://github.com/dnth/firstmate/actions/runs/33904204200).
-The four grown tests use conservative operator-supplied estimates: remote lifecycle 700000 ms, OMP secondmate 600000 ms, Boat routing 200000 ms, and Herdr server detach 90000 ms.
-These estimates deliberately exceed this attempt's observed times for those scripts to budget for the reported slower runs.
-At that refresh, the remainder estimated about 83.9 minutes; even a balanced four-shard partition would exceed the cap, so six runners now share the work.
-A script with no hint gets the conservative `PORTABLE_SERIAL_DEFAULT_WEIGHT_MS` default.
 Hints only affect balance: the coverage guard keeps the partition complete and disjoint whatever they say, so a stale hint costs a slower shard rather than lost coverage.
+A script with no hint gets the conservative `PORTABLE_SERIAL_DEFAULT_WEIGHT_MS` default.
+
+### Timeout diagnosis
+
+The recent 20-minute cancellations came from aggregate serial work, not a hanging `fm-watch-checkpoint.test.sh`.
+The cancelled PR run is [38753521432](https://github.com/dnth/firstmate/actions/runs/38753521432).
+Its first attempt ended during `fm-on.test.sh` after the job reached its timeout, before the checkpoint test began.
+Its second attempt reached the checkpoint test about five seconds before cancellation, after the preceding serial scripts consumed the available job time.
+Its third attempt ended during `fm-wake-drain-unread-status.test.sh` after its wake acknowledgement, about ten seconds after that test began.
+On main run [38746979839 attempt 1](https://github.com/dnth/firstmate/actions/runs/38746979839/attempts/1), all 28 scripts emitted a 19m55.325s summary before the job was cancelled near its 20-minute limit.
+The passing main attempt completed the same shard's 28 scripts in 17m20.456s.
+
+| Run / attempt | Measured shard-3 script spans (UTC) | Result |
+|---|---|---|
+| [Main 38746979839 attempt 1](https://github.com/dnth/firstmate/actions/runs/38746979839/attempts/1), cancelled | `fm-watch-triage`: 13:56:29–14:03:33 (423,952 ms); `fm-ext-decision`: 14:10:55–14:13:28 (153,704 ms); `fm-watch-checkpoint`: 14:16:03–14:16:10 (5,993 ms). | All scripts completed; the 19m55.325s suite summary preceded cancellation. |
+| [PR 38753521432 attempt 1](https://github.com/dnth/firstmate/actions/runs/38753521432/attempts/1), cancelled | `fm-watch-triage`: 14:40:54–14:48:14 (440,356 ms); `fm-ext-decision`: 14:55:56–14:58:39 (163,088 ms); `fm-on` began 15:00:52 and the job cancelled at 15:01:00. | Cancelled in the final `fm-on` test; checkpoint did not start. |
+| [PR 38753521432 attempt 2](https://github.com/dnth/firstmate/actions/runs/38753521432/attempts/2), cancelled | `fm-watch-triage`: 15:05:34–15:12:49 (435,200 ms); `fm-ext-decision`: 15:20:20–15:22:55 (155,491 ms); `fm-watch-checkpoint` began at 15:25:32 and the job cancelled at 15:25:37. | The checkpoint log reached its wake acknowledgement before cancellation; no test-end marker was emitted. |
+| [PR 38753521432 attempt 3](https://github.com/dnth/firstmate/actions/runs/38753521432/attempts/3), cancelled | `fm-watch-triage`: 15:30:20–15:37:47 (446,890 ms); `fm-ext-decision`: 15:45:36–15:48:25 (169,627 ms); `fm-wake-drain-unread-status` began at 15:50:11 and the job cancelled at 15:50:21. | Cancelled after the wake acknowledgement in `fm-wake-drain-unread-status`. |
+| [Main 38746979839 attempt 2](https://github.com/dnth/firstmate/actions/runs/38746979839/attempts/2), passed | `fm-watch-triage`: 14:22:32–14:29:22 (410,055 ms); `fm-ext-decision`: 14:35:28–14:37:25 (116,759 ms); `fm-watch-checkpoint`: 14:39:36–14:39:41 (5,271 ms). | All 28 scripts completed in 17m20.456s. |
+| [PR 38774522507](https://github.com/dnth/firstmate/actions/runs/38774522507), passing six-shard run | `fm-spawn-dispatch-profile`: 17:06:27–17:10:37 (249,954 ms); `fm-ext-decision`: 17:10:37–17:13:18 (161,682 ms); `fm-ext-topic-threads`: 17:13:18–17:14:59 (100,660 ms); `fm-watch-arm`: 17:14:59–17:16:25 (85,430 ms). | The 30 scripts summed to 1,022,126 ms (17m02.126s); job wall time was 17m17s. |
+
+The checkpoint completed in 5.271s and 5.993s on the passing and cancelled main attempts, respectively.
+All three cancelled PR attempts reached the 20-minute limit in different tests: `fm-on`, `fm-watch-checkpoint`, and `fm-wake-drain-unread-status`, so no single test hung.
+The six-shard PR run 38774522507 completed every script, but its 91m01.112s total exceeds six times the ~14-minute budget (84 minutes); shard 3 summed to 17m02.126s and took 17m17s wall time.
+The slowest script in shard 3 was `fm-spawn-dispatch-profile.test.sh` at 249,954 ms (4m09.954s), so the shard overrun is aggregate growth, not a single hung test.
+Seven shards are required to keep the refreshed per-script assignment near the documented budget; test assertions and the 20-minute cap remain unchanged.
+
+Recent green `fm-watch-triage.test.sh` measurements provide the comparison for attempt 3's 446,890 ms:
+
+| Passing run | Duration (ms) |
+|---|---:|
+| [38540024417](https://github.com/dnth/firstmate/actions/runs/38540024417) | 406,842 |
+| [38546582306](https://github.com/dnth/firstmate/actions/runs/38546582306) | 414,725 |
+| [38551009549](https://github.com/dnth/firstmate/actions/runs/38551009549) | 419,614 |
+| [38535146206](https://github.com/dnth/firstmate/actions/runs/38535146206) | 421,505 |
+| [38746979839 attempt 2](https://github.com/dnth/firstmate/actions/runs/38746979839/attempts/2) | 410,055 |
+| [PR 38774522507](https://github.com/dnth/firstmate/actions/runs/38774522507) | 428,164 |
+
+The 446,890 ms result exceeds the maximum of the five preceding green runs, 421,505 ms, by 25,385 ms (about 6%), so it is not a large outlier.
+Against the newer passing measurement in the final row, the difference narrows to 18,726 ms (about 4.4%).
+Separately, [run 38759616524](https://github.com/dnth/firstmate/actions/runs/38759616524) failed the stale-suppressor assertion after 100,083 ms.
+That is a separate correctness signal, not evidence that a 447-second `fm-watch-triage` run caused the timeout; the test assertions remain unchanged.
+
+### Required test fix
+
+`resolve_ios_pending` in [fm-remote-secondmate-lifecycle-e2e.test.sh](../tests/fm-remote-secondmate-lifecycle-e2e.test.sh) now selects the captured acknowledgement by the pending request's correlation ID.
+The previous selection excluded the first three result files and picked the last sorted result filename, which did not reliably identify the reply for that request.
+Handling an unrelated result could leave an inheritance acknowledgement unresolved before the merged-PR reply assertion.
+[Run 38770719267](https://github.com/dnth/firstmate/actions/runs/38770719267) failed that assertion without the correlation-ID selection; [run 38785253557](https://github.com/dnth/firstmate/actions/runs/38785253557) passed with it.
+The fix preserves the assertions and changes only which captured reply is handled.
+
+### Rebalanced duration profile
+
+The six-shard PR run measured 5,461,112 ms (91m01.112s) across 166 serial scripts, averaging 15m10.185s per shard.
+Seven shards are needed because six shards cannot fit that measured total within the ~14-minute budget.
+
+PR run 38774522507 measured `fm-watch-triage.test.sh` at 428,164 ms (7m08.164s); the earlier PR attempt 3 measured 446,890 ms (7m26.890s).
+If `fm-watch-triage.test.sh` reaches 446,890 ms again, shard 2 rises by only 18,726 ms to 798,888 ms (13m18.888s), still within the ~14-minute budget.
+The seven-shard assignment covers the observed variation without raising the 20-minute hang tripwire.
+CI acceptance for this change requires all seven actual serial jobs to complete without cancellation and with each job's wall time at most 15 minutes; projected script sums alone do not satisfy that requirement.
+
+| Shard | Scripts | Measured six-shard script sum | Six-shard CI job wall |
+|---|---:|---:|---:|
+| 1 of 6 | 23 | 14:28.264 | 14:46 |
+| 2 of 6 | 27 | 15:27.115 | 15:44 |
+| 3 of 6 | 30 | 17:02.126 | 17:17 |
+| 4 of 6 | 30 | 14:58.449 | 15:13 |
+| 5 of 6 | 26 | 13:22.769 | 13:36 |
+| 6 of 6 | 30 | 15:42.389 | 15:59 |
+| Total | 166 | 91:01.112 | — |
+
+The seven-shard LPT assignment uses those same per-script measurements and balances within 15 ms.
+
+| Shard | Scripts | Projected script time |
+|---|---:|---:|
+| 1 of 7 | 17 | 13:00.156 |
+| 2 of 7 | 23 | 13:00.162 |
+| 3 of 7 | 23 | 13:00.151 |
+| 4 of 7 | 25 | 13:00.154 |
+| 5 of 7 | 26 | 13:00.161 |
+| 6 of 7 | 26 | 13:00.162 |
+| 7 of 7 | 26 | 13:00.166 |
+| Total | 166 | 91:01.112 |
+
+The longest measured script is `tests/fm-remote-secondmate-lifecycle-e2e.test.sh` at 657,637 ms (10m57.637s), which is the single-script lower bound for any shard count.
+
+The current shard memberships below are generated from the same hints.
 
-Current serial shard membership is derived by `bin/fm-test-run.sh`; its header owns the inspection commands, and its weight hints own the balancing inputs.
+### Current shard membership
 
-The single longest estimate, `tests/fm-remote-secondmate-lifecycle-e2e.test.sh` at 700000 ms, is the floor for any shard count.
+#### `portable-serial-1of7` — 17 scripts
+
+```text
+tests/fm-remote-secondmate-lifecycle-e2e.test.sh
+tests/fm-wake-drain-unread-status.test.sh
+tests/fm-pi-watch-extension.test.sh
+tests/fm-backend-orca.test.sh
+tests/fm-pr-check-handoff.test.sh
+tests/fm-gate-refuse.test.sh
+tests/fm-todo-project.test.sh
+tests/fm-secondmate-lifecycle-e2e.test.sh
+tests/fm-treehouse-sweep.test.sh
+tests/fm-omp-compact-adviser.test.sh
+tests/fm-runpod-omp-auth.test.sh
+tests/fm-test-isolation-proof.test.sh
+tests/fm-operational-input.test.sh
+tests/fm-cleanup-skill.test.sh
+tests/fm-send-inbox-doorbell-live-e2e.test.sh
+tests/fm-omp-branch-live-e2e.test.sh
+tests/fm-omp-herdr-exit-live-e2e.test.sh
+```
+
+#### `portable-serial-2of7` — 23 scripts
+
+```text
+tests/fm-watch-triage.test.sh
+tests/fm-omp-harness.test.sh
+tests/fm-runpod-lifecycle.test.sh
+tests/fm-main-ci-watch.test.sh
+tests/fm-task-inbox.test.sh
+tests/fm-secondmate-restart.test.sh
+tests/fm-backend.test.sh
+tests/fm-procevent-when.test.sh
+tests/fm-task-delivery.test.sh
+tests/fm-wake-drain-open-decisions.test.sh
+tests/fm-tangle-guard.test.sh
+tests/fm-ext-plugin-dest.test.sh
+tests/fm-guard-stale-banner.test.sh
+tests/fm-busy-state.test.sh
+tests/fm-receipt.test.sh
+tests/fm-gotmp.test.sh
+tests/fm-check-unregister.test.sh
+tests/fm-ask-user-authority.test.sh
+tests/fm-afk-pi-herdr-return-e2e.test.sh
+tests/fm-omp-branch-bash.test.sh
+tests/fm-omp-branch-types.test.sh
+tests/fm-omp-secondmate-live-e2e.test.sh
+tests/fm-omp-primary-live-e2e.test.sh
+```
+
+#### `portable-serial-3of7` — 23 scripts
+
+```text
+tests/fm-spawn-dispatch-profile.test.sh
+tests/fm-ext-topic-threads.test.sh
+tests/fm-bearings-snapshot.test.sh
+tests/fm-backlog-atomicity.test.sh
+tests/fm-secondmate-safety.test.sh
+tests/fm-trace-context-spawn.test.sh
+tests/fm-runpod-routing.test.sh
+tests/fm-fleet-sync.test.sh
+tests/fm-boat-routing.test.sh
+tests/fm-remote-backlog-handoff.test.sh
+tests/fm-tasks-axi.test.sh
+tests/fm-fleet-snapshot-view.test.sh
+tests/fm-inbox.test.sh
+tests/fm-ext-watch.test.sh
+tests/fm-omp-relaunch-guard.test.sh
+tests/fm-send-secondmate-marker.test.sh
+tests/fm-remote-secondmate-sleep-reconcile.test.sh
+tests/fm-supervision-events.test.sh
+tests/fm-calm-pi-extension.test.sh
+tests/fm-treehouse-orphan-recovery.test.sh
+tests/fm-backend-cmux-smoke.test.sh
+tests/fm-codex-continuity-live-e2e.test.sh
+tests/fm-remote-omp-update-restart-live-e2e.test.sh
+```
+
+#### `portable-serial-4of7` — 25 scripts
+
+```text
+tests/fm-secondmate-harness.test.sh
+tests/fm-ext-waiting-pin.test.sh
+tests/fm-watch-arm.test.sh
+tests/fm-bootstrap.test.sh
+tests/fm-session-start.test.sh
+tests/fm-remote-job.test.sh
+tests/fm-public-followup.test.sh
+tests/fm-runpod-pod-boot.test.sh
+tests/fm-omp-branch-supervision.test.sh
+tests/fm-afk-inject-e2e.test.sh
+tests/fm-omp-task-inbox-doorbell.test.sh
+tests/fm-kimi-harness.test.sh
+tests/fm-shared-captain-inheritance.test.sh
+tests/fm-on.test.sh
+tests/fm-herdr-session-cleanup.test.sh
+tests/fm-watch-checkpoint.test.sh
+tests/fm-backlog-handoff.test.sh
+tests/fm-afk-return.test.sh
+tests/fm-documentation-audiences.test.sh
+tests/fm-classify-decision-key.test.sh
+tests/fm-pending-reply-sleep-reconcile.test.sh
+tests/fm-pi-compatible-family.test.sh
+tests/fm-herdr-submit-confirm-live-e2e.test.sh
+tests/fm-grok-stop-live-e2e.test.sh
+tests/fm-grok-continuity-live-e2e.test.sh
+```
+
+#### `portable-serial-5of7` — 26 scripts
+
+```text
+tests/fm-pr-check-security.test.sh
+tests/fm-omp-secondmate.test.sh
+tests/fm-wake-queue.test.sh
+tests/fm-wake-daemon-lifecycle-e2e.test.sh
+tests/fm-boat-omp-auth.test.sh
+tests/fm-remote-secondmate-trace-context.test.sh
+tests/fm-boat-lifecycle.test.sh
+tests/fm-send-inbox.test.sh
+tests/fm-teardown-endpoint-safety.test.sh
+tests/fm-remote-secondmate-update-restart.test.sh
+tests/fm-prepush-guard.test.sh
+tests/fm-remote-secondmate-parent-binding.test.sh
+tests/fm-pending-reply.test.sh
+tests/fm-inbox-result.test.sh
+tests/fm-backend-zellij.test.sh
+tests/fm-remote-doctor.test.sh
+tests/fm-backend-cmux.test.sh
+tests/fm-session-lock-ancestry.test.sh
+tests/fm-omp-update.test.sh
+tests/fm-devin-adapter.test.sh
+tests/fm-remote-entrypoint.test.sh
+tests/fm-stall-recovery.test.sh
+tests/fm-backend-herdr-focus-flash-e2e.test.sh
+tests/fm-devin-herdr-live-e2e.test.sh
+tests/fm-omp-compact-adviser-live-e2e.test.sh
+tests/fm-omp-herdr-live-e2e.test.sh
+```
+
+#### `portable-serial-6of7` — 26 scripts
+
+```text
+tests/fm-ext-decision.test.sh
+tests/fm-boat-lane-runner.test.sh
+tests/fm-ext-bridge.test.sh
+tests/fm-procevent.test.sh
+tests/fm-omp-primary.test.sh
+tests/fm-daemon.test.sh
+tests/fm-landing.test.sh
+tests/fm-secondmate-sync.test.sh
+tests/fm-spawn-relaunch-dead-endpoint.test.sh
+tests/fm-send-resolve-key.test.sh
+tests/fm-turnend-guard.test.sh
+tests/fm-busy-adapter-wiring.test.sh
+tests/fm-spawn-worktree-settle.test.sh
+tests/fm-send-turn-start.test.sh
+tests/fm-dispatch-resolve.test.sh
+tests/fm-idle-sleep-watch.test.sh
+tests/fm-omp-fleet-hooks.test.sh
+tests/fm-trace-context-lib.test.sh
+tests/fm-subagent-pretool-check.test.sh
+tests/fm-reflect-skill.test.sh
+tests/fm-treehouse-status-read-only.test.sh
+tests/fm-send-secondmate-marker-herdr-e2e.test.sh
+tests/fm-gitignore-config.test.sh
+tests/fm-pi-primary-live-e2e.test.sh
+tests/fm-opencode-primary-live-e2e.test.sh
+tests/fm-omp-relaunch-tmux-live-e2e.test.sh
+```
+
+#### `portable-serial-7of7` — 26 scripts
+
+```text
+tests/fm-hermes-harness.test.sh
+tests/fm-teardown.test.sh
+tests/fm-remote-reply.test.sh
+tests/fm-watch-recovery-loop.test.sh
+tests/fm-claude-stop-autoarm.test.sh
+tests/fm-ext-ask.test.sh
+tests/fm-watcher-lock.test.sh
+tests/fm-vendor-auth-probe.test.sh
+tests/fm-spawn-pool-base-freshen.test.sh
+tests/fm-wake-drain-open-decisions-cursor.test.sh
+tests/fm-secondmate-liveness.test.sh
+tests/fm-ext-thread-context.test.sh
+tests/fm-idle-sleep.test.sh
+tests/fm-update.test.sh
+tests/fm-startup-memory-budget.test.sh
+tests/fm-receipt-check.test.sh
+tests/fm-backend-herdr-server-detach.test.sh
+tests/fm-merge-local.test.sh
+tests/fm-test-fixture-cleanup.test.sh
+tests/fm-sessionstart-nudge.test.sh
+tests/fm-backend-tmux-smoke.test.sh
+tests/fm-hermes-live-e2e.test.sh
+tests/fm-backend-zellij-smoke.test.sh
+tests/fm-claude-stop-autoarm-live-e2e.test.sh
+tests/fm-omp-worker-tmux-live-e2e.test.sh
+tests/fm-quota-array-dispatch-live-e2e.test.sh
+```
 
-Refresh the hints by downloading the per-shard timing artifacts from a green CI run and replacing the `portable_serial_weight_hints` table in `bin/fm-test-run.sh` with the measured `path`/`duration_ms` pairs:
+To refresh after a green run, replace the full hint table with measured per-script pairs and regenerate the membership snapshot above from the same hints.
+Copy the generated assignments into the snapshot and run the partition guard.
 
 ```sh
 gh run download <run-id> -R dnth/firstmate --pattern 'fm-test-timing-portable-serial-*' -D /tmp/fm-serial
 jq -r '.scripts[] | [.path, .duration_ms] | @tsv' /tmp/fm-serial/*.json | LC_ALL=C sort
+for i in 1 2 3 4 5 6 7; do
+  printf 'portable-serial-%sof7\n' "$i"
+  bin/fm-test-run.sh --list --lane "portable-serial-${i}of7"
+done
 bin/fm-test-run.sh --check-coverage
 ```
 
@@ -107,7 +392,7 @@ Portable shards, each portable serial shard, and the Herdr lane upload runner-ge
 | Job | timeout-minutes | Rationale |
 |---|---:|---|
 | portable parallel 1/2 | 10 | The measured shard sums are about three minutes and the timeout is a hang tripwire. |
-| portable serial 1-6 | 20 | Each rebalanced shard estimates about 14 minutes of script time, leaving about six minutes for setup and runner-load variance while still tripping on a wedged test. |
+| portable serial 1-7 | 20 | The seven-shard assignment projects 13m00.151s–13m00.166s from PR run 38774522507, leaving about 6m59.834s for setup and runner-load variance while keeping the timeout as a hang tripwire. |
 | Herdr | 40 | The real-Herdr lane keeps its dedicated timeout. |
 
 Timeouts are hang tripwires rather than expected healthy durations.
diff --git a/tests/fm-remote-secondmate-lifecycle-e2e.test.sh b/tests/fm-remote-secondmate-lifecycle-e2e.test.sh
index dd0062d7b78..68d8a2e8f46 100755
--- a/tests/fm-remote-secondmate-lifecycle-e2e.test.sh
+++ b/tests/fm-remote-secondmate-lifecycle-e2e.test.sh
@@ -970,7 +970,7 @@ resolve_ios_pending() {
     remote_env "$ROOT/bin/fm-procevent.sh" start "$SID" >/dev/null 2>&1 || true
     pending_result="" n=200
     while [ -z "$pending_result" ] && [ "$n" -gt 0 ]; do
-      pending_result=$(find "$PARENT/state/procevent-inbox" -name "$SID.*.result" ! -name "$SID.1.result" ! -name "$SID.2.result" ! -name "$SID.3.result" -print | sort | tail -1)
+      pending_result=$(grep -lF "done [corr=$pending_corr]:" "$PARENT/state/procevent-inbox/$SID".*.result 2>/dev/null | head -1 || true)
       [ -n "$pending_result" ] || { n=$((n - 1)); sleep 0.1; }
     done
     [ -n "$pending_result" ] \