Summary
Deferred doc-completeness follow-up from PR #460 (fixes #394).
plugins/repo-hygiene/skills/clean/context/git-tree-reset.md:39 currently reads:
Aborts the apply if reset --hard fails (exit 5) — clean and the restore guard never run, so a failed reset can never leave the tree cleaned but not reset.
This sentence is accurate — it describes the gating contract the script enforces (git-tree-reset.sh:190-195): on a failed reset --hard, clean and the restore guard are skipped, so the tree is never left cleaned-but-not-reset.
However, it is now slightly incomplete relative to the stderr failure message added in 745def6, which correctly surfaces that reset --hard is not atomic and may have partially modified tracked files before it failed. A caller reading only the context doc (not the stderr output) would not know to expect partial tracked-file modification after exit 5.
Suggested change
Add a parenthetical to line 39, e.g. (the reset itself may have partially modified tracked files), so the doc is consistent with the stderr message.
Why deferred rather than fixed in #460
PR #460 was at a converged, merge-ready head SHA (745def6, all checks green, reviewers converged). The reviewer (claude[bot]) explicitly marked this item "no action required / no code changes warranted / not a blocker / cosmetic." Pushing a one-line doc edit would move head off the ready SHA and re-trigger a full review cycle on an otherwise-done PR. Tracked here instead to preserve the merge-ready state.
Context
Summary
Deferred doc-completeness follow-up from PR #460 (fixes #394).
plugins/repo-hygiene/skills/clean/context/git-tree-reset.md:39currently reads:This sentence is accurate — it describes the gating contract the script enforces (
git-tree-reset.sh:190-195): on a failedreset --hard,cleanand the restore guard are skipped, so the tree is never left cleaned-but-not-reset.However, it is now slightly incomplete relative to the stderr failure message added in
745def6, which correctly surfaces thatreset --hardis not atomic and may have partially modified tracked files before it failed. A caller reading only the context doc (not the stderr output) would not know to expect partial tracked-file modification after exit 5.Suggested change
Add a parenthetical to line 39, e.g.
(the reset itself may have partially modified tracked files), so the doc is consistent with the stderr message.Why deferred rather than fixed in #460
PR #460 was at a converged, merge-ready head SHA (
745def6, all checks green, reviewers converged). The reviewer (claude[bot]) explicitly marked this item "no action required / no code changes warranted / not a blocker / cosmetic." Pushing a one-line doc edit would move head off the ready SHA and re-trigger a full review cycle on an otherwise-done PR. Tracked here instead to preserve the merge-ready state.Context
745def6)