skills/agent-evals/assets/trigger-queries.reviewed-learning-inbox.json
|
[ { "id": "pos-01", "query": "you just corrected me on that classification bug — save the lesson somewhere I can review later", "should_trigger": true, "split": "train" }, { "id": "pos-02", "query": "keep a queue of things we learned from failures in this repo, but don't change any rules yet", "should_trigger": true, "split": "train" }, { "id": "pos-03", "query": "show me the captured lessons waiting for review and let me throw out the useless ones", "should_trigger": true, "split": "train" }, { "id": "pos-04", "query": "I want to see exactly what would be added to that Instruction before you touch it, then I'll approve it", "should_trigger": true, "split": "validation" }, { "id": "pos-05", "query": "turn that resolved test failure into a proposed improvement for an existing skill, with the evidence attached", "should_trigger": true, "split": "validation" }, { "id": "pos-06", "query": "why was my candidate refused when I pasted the log text into it?", "should_trigger": true, "split": "validation" }, { "id": "neg-01", "query": "there's no .memory-bank folder here, set one up before we start the refactor", "should_trigger": false, "split": "train" }, { "id": "neg-02", "query": "this SKILL.md is 700 lines, split it into references and fix the description", "should_trigger": false, "split": "train" }, { "id": "neg-03", "query": "run the trigger-query sweep and tell me the pass@k for these skills", "should_trigger": false, "split": "train" }, { "id": "neg-04", "query": "review this pull request diff and tell me which findings are blockers", "should_trigger": false, "split": "validation" }, { "id": "neg-05", "query": "the build fails only in CI, help me reproduce it locally and find the root cause", "should_trigger": false, "split": "validation" }, { "id": "neg-06", "query": "curate progress.md, it's about to blow its line budget again", "should_trigger": false, "split": "validation" } ] |