Doctor
Find the skills your agent actually needs.
Tell it which agent you use and what you want it better at. Instead of a list of results to sort through yourself, it recommends a small stack — why each skill was picked, which agent it’s actually confirmed for, what it depends on, and where two picks overlap enough that you don’t need both.
Method
How this works, plainly.
There’s no model call here. Your goal is matched against a small, curated vocabulary of capabilities — the same keyword-matching approach you can inspect yourself by hovering any capability tag in a result. The same input always produces the same recommendation.
Every skill in the set was already source-verified while researching this site’s Claude Code, Codex, and Cursor guides — nothing here was scraped fresh or invented for this tool. Each recommendation carries an evidence tier: confirmed means the skill’s own source or vendor states it works with that agent; portable means it only reaches that agent through the shared SKILL.md format and a directory convention — a real but weaker claim, and this is the one place on the site that distinguishes the two inline rather than in a table. The full model is on Agent Skills Compatibility.
Overlap between two recommended skills is a hand-checked relationship, not a similarity score — it only appears when the reasoning is something we could actually verify and state. Where a goal doesn’t match anything, or matches a capability with no verified skill for your agent yet, it says so rather than forcing an answer. For everything not covered here, see where to find agent skills and, for the format itself, SKILL.md.