clean-code · davila7 — Claude Code skill: coding standards, install

What clean-code does

The skill lays out the rules the agent follows while writing and editing code. It ships as tables: core principles (single responsibility, DRY, KISS, YAGNI), naming conventions, limits on function size and nesting depth. A separate section lists anti-patterns — a comment on every line, a helper for a one-liner, a factory for two objects, a utils.ts holding a single function.

Before touching any file, the skill makes the agent check what imports it and which tests cover it, then edit the dependent files in the same task. When the work is done, there is a self-check: goal met, all files updated, lint and TypeScript pass. Anything that fails gets fixed before the agent reports back.

The skill also documents a set of verification scripts, but those belong to other skills in the same collection. Each agent runs only its own: ux_audit.py for frontend work, api_validator.py for backend. Without those skills installed the scripts are simply not there, and clean-code runs fine on its own. Tool access is limited to Read, Write and Edit.