From 344b0a5fb8793255bcc3840e6371f7622a9cbf10 Mon Sep 17 00:00:00 2001 From: Sankalp Gilda <23521054+astrogilda@users.noreply.github.com> Date: Fri, 23 Oct 2026 17:06:47 -0400 Subject: [PATCH] docs(triage-review): hold replay decisions to the allowlisted checker Each summarized replay decision carries the checker release it came from. The reviewer step that allowlists a checker now says to check that field, and that a decision with no checker or a release outside the allowlist backs nothing. Signed-off-by: Sankalp Gilda <23521054+astrogilda@users.noreply.github.com> --- skills/triage-review/SKILL.md | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/skills/triage-review/SKILL.md b/skills/triage-review/SKILL.md index 7b2807b..fb68ada 100644 --- a/skills/triage-review/SKILL.md +++ b/skills/triage-review/SKILL.md @@ -99,6 +99,10 @@ the summaries as part of the report. manifest. It does not choose the checker. - The **reviewer agent** chooses or allowlists the checker, runs replay, attaches the report, and writes the review intent. +- Hold each decision to that allowlist: every entry in `evidence_replays[].decisions[]` + carries `checker` (`{package, version}`, or `null` when the checker named none). + A decision whose `checker` is `null` or names a release outside the allowlist + backs nothing; re-run replay with the allowlisted checker. - A `supported` decision can back a **met** review criterion; cite the report hash and the retained artifacts in the verification record. - `contradicted`, `not_established`, or `stale` must leave the corresponding