Repository navigation
chore: audit the frozen EV-07 failures and preserve causal evidence - #984
Conversation
…er-measured-campaign: audit frozen EV-07 discrepancies without relabeling (#971)
|
To preview the documentation for this pull request, visit the following URL:
|
Copilot approval · Human review neededStatus: Evidence passed; a human reviewer must decide. Technical evidenceRevision: |
There was a problem hiding this comment.
Current status: All findings originating in this review are resolved.
Last reconciled on1f3fbdd. See aggregate Bugbot status.
🤖 Bugbot review snapshot
Bugbot reported 1 potential problem when commit 6417074 was analyzed. This snapshot is historical; use the status block above for current state. 1 finding is linked to changed code.
Findings
- low: The Kotlin probe's execution runtime is misidentified —
quality/bugbot/evidence/issue-971/README.md:28
Bugbot: review complete
Pull request · Verified commit · Current review checks · Workflow run |
…er-measured-campaign: distinguish Kotlin compiler and probe runtimes (#971)
Codecov Report✅ All modified and coverable lines are covered by tests. Additional details and impacted files@@ Coverage Diff @@
## develop #984 +/- ##
===========================================
+ Coverage 95.25% 95.66% +0.41%
===========================================
Files 1172 1563 +391
Lines 39236 49137 +9901
Branches 11685 15352 +3667
===========================================
+ Hits 37373 47006 +9633
- Misses 698 794 +96
- Partials 1165 1337 +172 🚀 New features to boost your workflow:
|
The EV-07 audit reproduces the current and prior failed scores without changing labels, thresholds or predictions. All 17 scorer FP and 11 FN now have independently reviewable causal evidence and a next action.
What changed
03f8e6e3plan, predictions and score unchanged, bind both campaigns to their recorded identities, and reproduce 80.23% current / 87.95% prior precision with the identical rubric.Validation
Review notes
No production prompt correction is supported solely by these taxonomy differences. The malformed workflow has a separately reviewed fixture/rubric follow-up in #985; unresolved cases name their missing contracts. No current corpus labels or historical results change. A recovered local artifact is not independent server timestamp attestation. Synthetic primitives do not execute private consumers or qualify a model profile.
The native-review runtime attribution is corrected at
1f3fbddeb4b84dea231ddeb7dc18af54fa33a565. Its full installed hook passed; await final-SHA native checks and review.This audit makes zero model calls and zero evaluated-PR writes. Production remains disabled. #974 still requires merged prerequisites and a new committed exact preregistration; all 42 #975 slots remain unstarted.
Closes #971.