Skip to content

trap: log #1083 pointer-route gate-3 finding (write-handoff.sh ancestor-guard skip) - #1100

Open
fdaviddpt wants to merge 1 commit into
mainfrom
trap/1083-pointer-skips-ancestor-guard
Open

fdaviddpt wants to merge 1 commit into
mainfrom
trap/1083-pointer-skips-ancestor-guard

Conversation

@fdaviddpt

Copy link
Copy Markdown
Contributor

Gate-3 release-audit finding (two independent rounds, dispatch tokens 7cf3acd0c77ea2f3 and a69f445acec9e55b) against the v0.44.1..HEAD delta, logged as a trap.d/ fragment per the non-blocking routing rule in skills/manager/phases/findings.md (ranking-table row: misreports, can ship behind a trap.d fragment).

What

#1083's pointer-store route in scripts/write-handoff.sh skips the pre-existing #750/#799 tracked/symlinked-ancestor write refusal, in the ordinary #776 subdirectory-project shape. Round 2 reproduced this with a control (not just reasoned): a tracked .remember/remember.md in a subdirectory project got silently overwritten through the pointer route, where the un-pointered path correctly refuses with "REFUSED: ... tracked".

See the fragment body for the full repro, the doctor.sh secondary site, and the fix direction (derive _WH_MEM_PROJ from the pointer's own store rather than the cwd guess).

Why this is not blocking the v0.45.0 release

The ranking table in skills/manager/phases/findings.md puts misreports in the "can ship behind a trap.d fragment" row, not the blocking row. gate3_disposition.py confirmed carry-forward-and-proceed for round 2.

No code change in this PR -- documentation only, for a later curation pass to promote into a jit-context rule or a tracked issue.

[AI-generated]

… the #750/#799 tracked/symlinked-ancestor write refusal in write-handoff.sh (non-blocking, misreports)\n\nCo-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>

Co-Authored-By: Max <noreply>
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant