Build a Custom AI Skill in 37 Lines (with working code)

A custom AI skill can be built in 37 lines of code, according to a developer guide. The skill reviews a git diff and returns a structured critique covering bugs, security, style, and performance, formatted for PR comments. The author argues that such skills encode team-specific rules, like banning console.log in production, making AI assistants more tailored to a codebase.

Member-only story Build a Custom AI Skill in 37 Lines with working code Most teams I work with have already adopted an AI coding assistant. Few of them have taught it anything specific to their codebase. That is the gap custom skills fill. A skill is a small, named bundle of instructions, usually 20 to 40 lines, that an AI agent loads on demand. It does not sit in the system prompt burning tokens. It lives in a folder, has a name, and gets invoked when the task matches. In this post we will build one. The whole thing: 30 lines, two files, one bash command to install. The skill reviews a git diff and returns a structured critique covering bugs, security, style, and performance. The output is formatted so you can paste it straight into a PR comment. By the end, you will have a working pattern. I will give you three more skill ideas at the bottom that you can ship in an afternoon each. By the way, if you are not Medium member, please click here to read. Why this matters: a vanilla AI assistant knows JavaScript. It does not know your team’s rule against console.log in production. A skill encodes that rule once, and every future session follows it. That is the difference between an assistant you use and an assistant that fits.