7 Commits

Author SHA1 Message Date
sean 93cefb7bb3 Ask all user questions upfront in langsmith-code-eval skill 2026-02-18 10:31:47 -08:00
sean 84ada04ad6 Improve langsmith-code-eval skill for cross-agent compatibility
- Add environment discovery step (support uv, poetry, pdm, etc.)
- Add dataset inspection step with new inspect_dataset.py script
- Bundle evaluator signatures inline to avoid stale external doc links
- Use find_dotenv for better .env file discovery
- Add workflow checklist for progress tracking
- Remove agent-specific language for general compatibility
2026-02-16 18:00:54 -08:00
sean 5484818ab3 Add .env discovery step for flexible path handling 2026-02-16 16:43:01 -08:00
sean a042a8f1be Rename skill to langsmith-fetch and fix commands
- Rename from fetching-langsmith-traces to langsmith-fetch
- Show --project-uuid option and LANGSMITH_PROJECT auto-detection

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-02-16 16:25:03 -08:00
sean 66f6862e30 Add langsmith-fetch skill for trace retrieval
Skill for fetching LangSmith traces to troubleshoot agent behavior.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
2026-02-16 16:14:43 -08:00
sean 6c3c88d516 Rewrite SKILL.md for conciseness and flexibility
- Reduce from 176 to 47 lines (73% reduction)
- Remove prescriptive code patterns that may not match user's use case
- Add critical thinking checkpoint for trace validation
- Point to docs for API discovery instead of hardcoding patterns
- Add dotenv loading to inspect_trace.py
2026-02-13 10:54:13 -08:00
sean 5b46c12f62 Initial commit: LangSmith code evaluator skill 2026-02-12 19:12:27 -08:00