The hook's missing-tool notice says lint is "skipped for this session", but no skip latch exists — only the NOTICE latches (skip-notices/ marker); the probe re-runs on every edit and recovered mid-session in reproduction. The wording makes operators (and agents) believe enforcement is gone until restart, so nobody retries. Also, the notice omits what PATH was probed, which made the 2026-08-15 false-negative (nvm-prefix issue, filed separately) needlessly hard to diagnose.
Requested changes: reword the notice to match actual behavior (per-edit probe, notice-only latch), and append the probed PATH (or the resolved-tool path on success) as a one-line diagnostic.
Severity: Low. The hook otherwise behaved exactly per design in the triggering session (honest probe, advisory exit 0, correct scope rejection of out-of-tree writes). Evidence packet: plugin-quality data dir, markdown-format/20260815T173035Z.
The hook's missing-tool notice says lint is "skipped for this session", but no skip latch exists — only the NOTICE latches (
skip-notices/marker); the probe re-runs on every edit and recovered mid-session in reproduction. The wording makes operators (and agents) believe enforcement is gone until restart, so nobody retries. Also, the notice omits what PATH was probed, which made the 2026-08-15 false-negative (nvm-prefix issue, filed separately) needlessly hard to diagnose.Requested changes: reword the notice to match actual behavior (per-edit probe, notice-only latch), and append the probed PATH (or the resolved-tool path on success) as a one-line diagnostic.
Severity: Low. The hook otherwise behaved exactly per design in the triggering session (honest probe, advisory exit 0, correct scope rejection of out-of-tree writes). Evidence packet:
plugin-qualitydata dir,markdown-format/20260815T173035Z.