Both reviewers noted the marker write-failure test only overrides TMPDIR,
which Node ignores on Windows (os.tmpdir reads TEMP/TMP there), so the
test would fail in the Windows CI lane. Set TEMP and TMP alongside TMPDIR;
on POSIX the extra vars are ignored by os.tmpdir.
Test plan:
- tests/hooks/evaluate-session.test.js: 25 passed, 0 failed
- Windows behavior cannot be verified from this Linux host; the change
follows both reviewers' prescription exactly.
Address bot review on the Stop delivery path: the stop:evaluate-session
registration no longer carries async (an async Stop registration cannot
block), so delivery is synchronous, and the metadata fingerprint for the
changed entry is regenerated with the repo's own function; the delivery
marker is now claimed atomically with exclusive create and any write
failure fails closed instead of delivering without once-per-session
protection; the skill doc now marks the shell entry as counting-only,
requires the JavaScript evaluator through the plugin hooks for delivery,
and clarifies that Stop fires whenever Claude finishes responding; test
spawn helpers no longer coerce a failed spawn to exit 0.
Test plan:
- tests/hooks/evaluate-session.test.js: 25 passed, 0 failed (2 new:
marker write failure skips delivery, registration is synchronous);
pristine comparison: 22 passed, 3 failed (delivery, marker, registration)
- tests/hooks/hooks.test.js: 260 passed
- tests/hooks/stop-hooks-stdout.test.js: 48 passed
- tests/hooks/cursor-hook-dispatchers.test.js: 27 passed
- tests/integration/hooks.test.js: 27 passed
- scripts/ci/check-unicode-safety.js: passed
The v1 evaluator computed the learning signal but wrote it to stderr, which Claude Code never forwards to the model, so the learned-skills folder was never populated through this path.
Delivery now follows the maintainer-endorsed shape: an explicit stop_nudge_opt_in flag (default off, so existing installs are unchanged), the native Stop block contract (decision block plus reason) instead of plain stdout text, at most one delivery per session via a temp-dir marker, Stop-loop protection via stop_hook_active, and delivery only for sessions with a registered lease from SessionStart.
Tests: 7 new evaluator tests (opt-in off, no lease, one valid block then silence on repeat, stop_hook_active, short session, non-Stop event, missing session id). Evaluator suite 23/23 with the fix, failing on pristine; hooks suite 260/260; Stop stdout contract suite 48/48.
Limits stated plainly: end-to-end delivery in a live Claude Code session is not certified from this host (unit-level spawn plus parse only), and the fix ships inert by default until a user opts in.
- loadAliases resets to defaults when aliases field is a truthy non-object (string)
- detectFromPackageJson returns null for empty (0-byte) package.json
- evaluate-session uses learned_skills_path config override with ~ expansion
- evaluate-session.js: verify regex whitespace tolerance around colon
matches "type" : "user" (with spaces), not just compact JSON
- validate-rules.js: empty directory with no .md files yields Validated 0
- validate-skills.js: directory with only files, no subdirectories yields
Validated 0
Total: 800 tests, all passing
Round 53: Adds 3 hook tests — validates evaluate-session.js
falls back to CLAUDE_TRANSCRIPT_PATH env var when stdin JSON
is invalid, post-edit-console-warn.js truncates output to max
5 matches, and post-edit-format.js passes through data when
the target .tsx file doesn't exist.
The command cross-reference regex /^.*`\/(...)`.*$/gm only captured the
LAST command ref per line due to greedy .* consuming earlier refs.
Replaced with line-by-line processing using non-anchored regex to
capture ALL command references.
New tests:
- 4 validate-commands multi-ref-per-line tests (regression)
- 8 evaluate-session threshold boundary tests (new file)
- 6 session-aliases edge case tests (cleanup, rename, path matching)