Twinned from https://github.com/cfsh/ninefold/issues/424 by tools/beads/import_github.py's reconcile pass.
Suggestions from ninefold-reviewer's review of #422. Both are advisory-minor — #422 is CONVERGED and neither blocks it.
1. tools/pr/poll.sh:251 — the ⓘ rev:off line contradicts a verdict printed below it.
The line says "no NFR verdict is coming", unconditionally. But muting a PR does not retract verdicts it already has, and the ⓘ line is printed *above* the NFR line, so a PR muted after a review (or reviewed via the --force <n> override) renders as:
``` ⓘ rev:off — auto-reviewer MUTED for this PR; no NFR verdict is coming (…) NFR 2026-08-03T… [fresh]: REVIEWER-VERDICT sha=… result=PASS ```
"no further verdict is coming" is true in every case and still says the thing the line exists to say.
2. tools/reviewer/dispatch.sh:393 — --force <n> on a muted PR buys a verdict comment, not a label.
The forced review runs; spawn() sets rev:pending; the verdict sets rev:pass. The next *bare* tick hits the same opt-out gate and clears it as a "stale state label". So the deliberate "no, review this one anyway" leaves a comment and no queue state, within one tick and with nothing on the PR explaining why the green went away.
Defensible under the README’s own rule (a muted PR is not in the review flow), but currently undocumented. Either exempt a PR whose newest verdict post-dates the mute, or add one line to the README’s "Opting a PR out" section saying the override yields a verdict, not a label.
None.
No comments.