GitHub radar
shadcn/lint: linter that keeps AI agents on your design system
An agent-first linter for Tailwind design systems from the shadcn team: catches design rule violations as the agent works and tells it exactly what to fix, so it corrects itself in the same pass.
@shadcn/lint is an agent-first linter for Tailwind design systems, made by the shadcn team. When an AI agent writes UI code, it often produces colors, spacing, or utility classes that break the project's design rules. This linter catches those violations and provides error messages that explain exactly what is wrong and what the correct alternative is, so the agent corrects itself in the same pass without additional rounds. In benchmark tests, Claude and GPT models reached zero violations after a single round of lint feedback in almost every task. Sonnet 5, Haiku 4.5, and Opus 5 all completed eight out of eight tasks with all errors brought to zero after one correction. Fixing violations with lint feedback cost 10% to 48% less in tokens compared to providing rules alone. It works with React, Svelte, and Vue projects that use Tailwind v4 and supports both ESLint and Oxlint.
Why a vibe-coder should care
If you build websites with an AI agent and often need several rounds of correction to get the design right, this tool removes that back-and-forth. The agent sees a precise error message in the same pass, corrects the component, and you get the result you described the first time.
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install @shadcn/lint in this project: https://github.com/shadcn-ui/lint — follow the README, configure the rules for our design system, add linting to the code review step, and show me the final config. Ask me if you need settings for specific components.
Runs on any laptop — only Node.js needed
Open on GitHub▌ More finds