Add thirdeye-review skill, rename use-thirdeye-evals → thirdeye-evals - #22
Merged
Conversation
2 tasks
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
thirdeye-review, for auditing other agents' traces and turning findings into concrete invocation, skill, or convention changes. Five-step workflow: scope cohort → identify tasks → audit tool/execution patterns → quantify token cost & errors → tie back to invocation surface and recommend.use-thirdeye-evals→thirdeye-evalsso the eval skill follows the samethirdeye-*naming asthirdeye-filterandthirdeye-review. (use-thirdeyestays as-is — it's the consumer-style "use the CLI" skill.)README.mdandpyproject.tomlpackage-data accordingly so the new files ship in the wheel.What's in
thirdeye-reviewSKILL.mdplus five focused references:population-scoping.md— agent-type / cwd / time-window cohort selection, sizing guidance.task-identification.md— labelling sessions from the first user turn, baselines per task shape.pattern-audit.md— tool-mix counts, waste signatures with thresholds (re-reads,grep_via_bash, unbounded Reads, retry loops, MCP dumps), execution-order patterns.token-and-error-analysis.md— spike isolation, tool-error clustering, recovery-shape reading.invocation-tieback-and-recommendations.md— pattern→surface map (prompt, skill description, conventions, allowlist, retry flags) with worked recommendation examples.Anti-patterns spelled out in the top-level skill: no reviewing a single session, no counting without comparing, no "be more careful" recommendations, no ignoring the invocation, no skipping the tagging step.
Test plan
python -m pytest tests/test_skill_install.py -q→ 14 passedpython -m thirdeye skill list→ showsthirdeye-evals,thirdeye-filter,thirdeye-review,use-thirdeyepip install -e .thenthirdeye skill install --only thirdeye-review /tmp/skills/lands a symlink to the new skill.thirdeye skill install --only thirdeye-evalsstill works under the renamed name.🤖 Generated with Claude Code