skill-authoring

Eval and improve a skill against golden cases — run the target skill blind in a fresh, context-free subagent on each example input, grade the artifact against a per-case bar, and let the gaps drive the edits. Use when the user wants to test/eval/improve/harden a skill, says "this skill keeps producing X / keeps missing Y", or hands a skill plus cases with a bar for what good looks like. Pairs with write-skills (the authoring principles every fix obeys). Judgment-first and harness-free — prefer skill-creator for harness-based benchmarking, variance analysis, or description-optimization loops.

Create or revise agent skills. Use when adding a new skill file, renaming a skill, simplifying an existing skill, improving trigger descriptions, or deciding what belongs in a skill versus references, scripts, assets, or ordinary docs. This is the judgment layer — authoring principles and failure modes for manual, principle-level work; prefer skill-creator when the user wants end-to-end interactive creation with generated test harnesses or quantitative description optimization.