ci(evals): add GitHub Check Run reporting for evaluation results #640
Conversation
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
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
.github/scripts/create-eval-check.js
Outdated
| `### Conclusion`, | ||
| ``, | ||
| conclusion === "success" | ||
| ? `✅ **Passed**: Average score (${avgScore.toFixed(2)}) is above the catastrophic failure threshold (0.50)` |
There was a problem hiding this comment.
Bug: Incorrect threshold message for passing score
The success message says the score "is above" the threshold, but the actual check uses >= which includes scores equal to 0.50. When the average score is exactly 0.50, the check passes but displays "Average score (0.50) is above the catastrophic failure threshold (0.50)", which is mathematically incorrect. The message needs to say "is at or above" to match the >= comparison.
![sentry[bot]](https://avatars.githubusercontent.com/u/1396951?s=60&v=4){kind=small}
| if: ${{ !cancelled() }} | ||
| uses: codecov/test-results-action@v1 | ||
| with: | ||
| token: ${{ secrets.CODECOV_TOKEN }} |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
Comment on lines
+94
to
+104
| evalResults.push({ | ||
| name: test.fullName || test.title, | ||
| file: testFile.name, | ||
| avgScore: test.meta.eval.avgScore ?? null, | ||
| scores: test.meta.eval.scores || [], | ||
| passed: test.status === 'passed', | ||
| duration: test.duration, | ||
| }); | ||
| } | ||
| } | ||
| } |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
Create automated check runs that report eval scores and statistics: - Add create-eval-check.js script to parse eval-results.json and create GitHub Check Runs - Update eval:ci to output JSON results via --reporter=json - Set check conclusion to 'success' if avg score >= 0.5, 'failure' otherwise - Display overall statistics and score distribution (green/yellow/red) - Run check creation even if evals fail (continue-on-error + !cancelled) - Add checks:write permission to eval workflow - Ignore generated eval-results.json in version control The check run provides: - Overall average score with pass/fail threshold (0.50) - Score distribution by category (>=0.75, 0.50-0.74, <0.50) - Individual eval scores sorted by performance - Rationale for failed or low-scoring tests Co-Authored-By: Claude Code <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
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.
Create automated check runs that report eval scores and statistics:
The check run provides: