chore(0.34.0): release — eval scorecard + agent profile cells#84
Merged
Conversation
Headline: a feature PR's eval can finally answer the question a single run cannot — did this change regress persona P on profile F, even while the aggregate improved? AgentProfile is the unit of variation; the append-only JSONL scorecard is the per-cell timeline; diffScorecard issues per-cell verdicts with Cohen's d + Welch's t-test. Also consolidates the paired statistics, unifies the LLM retry classifier, commits the long-missing pr-review-benchmark source, and adds three production-pattern examples. See CHANGELOG.md for the full list.
tangletools
approved these changes
May 22, 2026
Contributor
tangletools
left a comment
tangletools
left a comment
There was a problem hiding this comment.
Verified — 0.34.0 release bump. package.json + clients/python/pyproject.toml in lockstep at 0.34.0; CHANGELOG entry covers the scorecard, agent-profile-cells, stats consolidation, llm-retry unify, pr-review-benchmark commit, and examples. typecheck 0, suite 1306 green, build emits OpenAPI cleanly. Tag v0.34.0 after merge to trigger publish.yml.
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Headline
A feature PR's eval can finally answer the question a single run cannot — did this change regress persona P on profile F, even while the aggregate improved?
AgentProfile+agentProfileHash— the harness's unit of variation (feat: eval scorecard — (persona × profile) score timeline #78).(scenarioId, profileHash)(feat: eval scorecard — (persona × profile) score timeline #78).diffScorecard— per-cellimproved/regressed/flat/newverdict with Cohen's d + Welch's t-test (feat: eval scorecard — (persona × profile) score timeline #78).RunRecord+runEvalCampaign(feat(eval): add agent profile cells #79).pr-review-benchmarksource committed — restored green CI (fix: commit pr-review-benchmark source — restores green CI on main #83).No breaking changes — additive across the board. See
CHANGELOG.mdfor the full list.Test plan
pnpm typecheck— 0 errorspnpm test— 1306 passed (135 files)pnpm build— green; OpenAPI spec emittedAfter merge: push tag
v0.34.0to triggerpublish.ymlfor the npm + Python tarballs.