The team behind shadcn/ui released @shadcn/lint, a linter for Tailwind CSS design systems that enforces component rules through code analysis rather than TypeScript types alone, according to the project's GitHub repository.
The tool catches violations like unauthorized styling changes, raw colors and arbitrary values in component usage, and lets teams configure what is allowed rather than hard coding restrictions into components, the repository says.
Its stated focus is AI agents writing code. "When an agent breaks a rule, the error explains what's wrong and suggests a fix based on your components, variants, and theme," the documentation says. The repository reports that agents using that feedback resolved violations in a single correction round, in some cases 10% to 48% cheaper than rule only approaches.
The linter works with React, Vue and Svelte, runs on ESLint 9.30 or newer or Oxlint 1.80 or newer, requires Node.js 20.19 or newer, and does not require shadcn/ui itself, according to the repository. It is released under the MIT license.
Most design system tooling was written for a human who reads an error once and fixes it. Writing the error message for an agent that will read it thousands of times a day, and pricing the fix in dollars saved per correction, is a small design choice that says a lot about who these tools now expect to be the primary reader.