test(spawn): add regression test for reused pooled worktree hook identity

[trillium]

Intent

Fix bin/fm-spawn.sh so a reused pooled treehouse worktree never inherits the prior tenant's per-task Claude hook identity (task id, busy-gen). Investigation found the write path already unconditionally overwrites every harness's hook artifact on every spawn (no skip-if-exists), so no functional code change is needed. Add a permanent regression test (tests/fm-spawn-reused-worktree-hooks.test.sh) covering spawn -> teardown -> respawn into the same pooled worktree, asserting the new task's hooks reference only its own id/gen and that firing its Stop hook never touches the prior task's turn-ended marker, plus a one-line invariant comment in fm-spawn.sh documenting the always-overwrite contract for future harness wiring.

What Changed

• Added tests/fm-spawn-reused-worktree-hooks.test.sh, covering spawn → teardown → respawn into the same pooled treehouse worktree, asserting the new task's hooks reference only its own task id/busy-gen and that firing its Stop hook never touches the prior tenant's turn-ended marker.
• Added a one-line invariant comment in bin/fm-spawn.sh documenting that the hook-artifact write path unconditionally overwrites every harness's hook identity on each spawn (no skip-if-exists), for future harness wiring.
• No functional code change: the write path already overwrote hook artifacts unconditionally, so this closes the gap with a permanent regression test rather than a fix.

Risk Assessment

✅ Low: The change is a comment-only addition to fm-spawn.sh (no functional code change) plus a new regression test whose assertions I verified against the actual hook-write, teardown-scrub, and busy-gen/turn-ended paths in fm-spawn.sh, fm-teardown.sh, and fm-busy-event.sh.

Testing

The diff (a one-line invariant comment in bin/fm-spawn.sh plus the new tests/fm-spawn-reused-worktree-hooks.test.sh) was exercised directly: the new test drives a real spawn -> teardown -> respawn sequence into the same pooled worktree via fm-spawn.sh/fm-teardown.sh with faked tmux/harness binaries, asserts the reused worktree's .claude/settings.local.json hooks reference only the second task's id and a distinct busy-gen, and fires the installed Stop hook to confirm it touches only the new task's turn-ended marker and never resurrects the prior tenant's — all assertions pass. The test is properly registered in the coverage guard and a portable CI lane, and sibling fm-spawn regression tests remain green, confirming the comment-only source change caused no regressions.

Pipeline

Updates from git push no-mistakes

✅ **intent** - passed

✅ No issues found.

✅ **Rebase** - passed

✅ No issues found.

✅ **Review** - passed

✅ No issues found.

✅ **Test** - passed

✅ No issues found.

• bin/fm-test-run.sh tests/fm-spawn-reused-worktree-hooks.test.sh — full end-to-end scenario, passes
• bin/fm-test-run.sh --check-coverage — new test is accounted for in the complete regression partition (total=106)
• bin/fm-test-run.sh --list --lane portable-serial — confirmed new test is scheduled in a CI lane (not silently orphaned)
• bin/fm-test-run.sh tests/fm-spawn-batch.test.sh tests/fm-spawn-dispatch-profile.test.sh tests/fm-spawn-worktree-settle.test.sh — sibling fm-spawn.sh regression tests still pass after the comment-only change
• git status --porcelain — worktree left clean, no test artifacts committed

✅ **Document** - passed

✅ No issues found.

✅ **Lint** - passed

✅ No issues found.

✅ **Push** - passed

✅ No issues found.

Summary by CodeRabbit

• Bug Fixes

  • Fixed stale automation settings when reusing a previously occupied worktree.
  • Ensured refreshed task hooks are applied correctly during respawn.
  • Prevented completion signals from affecting tasks other than the active one.

• Tests

  • Added end-to-end coverage for worktree reuse, teardown, respawn, and task-specific completion handling.

[coderabbitai]

No actionable comments were generated in the recent review. 🎉

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 9b962f83-845c-40c9-91c7-363d32a566cc

📥 Commits

Reviewing files that changed from the base of the PR and between 1e11ba8 and 8f4b12d.

📒 Files selected for processing (2)

• bin/fm-spawn.sh
• tests/fm-spawn-reused-worktree-hooks.test.sh

---

📝 Walkthrough

Walkthrough

The change documents unconditional hook artifact rewrites and adds an end-to-end test for spawning, tearing down, and respawning in the same pooled worktree.

Changes

Reused worktree hook validation

Layer / File(s)	Summary
Hook rewrite and reuse regression bin/fm-spawn.sh, tests/fm-spawn-reused-worktree-hooks.test.sh	Comments document task-incarnation and busy-generation bindings. The regression test verifies stale hook cleanup, fresh hook creation, and current-task Stop hook markers.

Estimated code review effort: 2 (Simple) | ~10 minutes

Suggested reviewers: kunchenguid

🚥 Pre-merge checks | ✅ 5

✅ Passed checks (5 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title clearly describes the main change: adding a regression test for hook identity in reused pooled worktrees.
Docstring Coverage	✅ Passed	No functions found in the changed files to evaluate docstring coverage. Skipping docstring coverage check.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fm/spawn-hook-refresh-fm-spawn-reused-pool-worktree-inherits-p-b6

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

_{Comment @coderabbitai help to get the list of available commands.}