fix(evidence): EVID-136 declared a review status where a direction belongs - #471
Open
explosivebit wants to merge 1 commit into
Open
explosivebit wants to merge 1 commit into
explosivebit wants to merge 1 commit into
Conversation
…longs `verdict: concerns` is not one of the three the scorer knows. Until PRD-086 FR-008 it fell through to `supports` and scored 1.0, so a code review whose own summary reads "three issues require coder attention before merge" counted as full confirmation of the thing it objected to. PRD-079 and RFC-011 carried that 1.0 for three weeks. The fail-closed change is what surfaced it — the first `score --all` after merging printed the warning. Set to `weakens`: the review does not refute the PRD, it lowers confidence in the state reviewed, and the three HIGH findings in the body are the substance of that. EVID-136 0.00 -> 0.30 (weakens at CL3) PRD-079 0.00 -> 0.30 RFC-011 0.00 -> 0.20 Those are the first honest numbers the three have had: 1.0 while the value was silently mapped, 0.0 once it was rejected, 0.30 now that it is stated. NOTE-052 records the trap, which is worth naming because both vocabularies are correct in their own place. `CONCERNS` is a real verdict here — it is what a Profile-B reviewer returns. The structured field asks a different question: which way does this evidence push the claim. A review can be CONCERNS and `weakens`, or CONCERNS and `supports` if the concerns are cosmetic. Review verdicts, test outcomes and CI statuses belong in the prose. One pack in 173. The convention holds, which is why the one that broke it went unseen for three weeks and why this needed a gate rather than a guideline. Refs: EVID-136, NOTE-052, PRD-086 FR-008
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The first
forgeplan score --allafter #470 found an artifact defect that threeweeks of use, and an adversarial review by 25 agents, had both missed.
What was wrong
EVID-136— a code review of the blog scaffold — declaredverdict: concerns.The scorer knows three values:
supports,weakens,refutes.Until PRD-086 FR-008 the unrecognised value fell through to
supportsandscored 1.0. So a review whose own summary reads "three issues require coder
attention before merge" — three HIGH findings, including an integration order
that violates its own RFC — counted as full confirmation of the thing it
objected to.
PRD-079andRFC-011carried that 1.0 for three weeks.FR-008 is what surfaced it. The fail-closed change turned a silent mapping into
a printed warning on the very first recompute.
Numbers
The third column is the first honest reading these three have had.
weakensatCL3 scores 0.30 — not zero, not full marks. The review does not refute the PRD;
it lowers confidence in the state that was reviewed.
Why NOTE-052 exists
CONCERNSis a real word in this project — it is exactly what a Profile-Breviewer returns (
PASS/CONCERNS/BLOCKER). The pack's## Verdictheading saying
CONCERNSwas correct: that is prose about the review.The structured field underneath asks a different question — which way does this
evidence push the claim. A review can be CONCERNS and
weakens, or CONCERNSand
supportsif the concerns are cosmetic. Two vocabularies, both correct intheir own place, one field.
The note records the distinction so the next author does not repeat it. Review
verdicts, test outcomes and CI statuses belong in the prose.
Scale
One pack in 173. The convention holds — which is precisely why the single pack
that broke it went unseen for three weeks, and why this needed a gate rather
than a guideline.
Artifacts only; no code changes.
Refs: EVID-136, NOTE-052, PRD-086 FR-008
🤖 Generated with Claude Code