Google Labs has open-sourced DESIGN.md, a format introduced through its Stitch design tool to encode design systems in plain Markdown. The file is meant to capture a brand's colors, typography, spacing, and component rules, allowing tools such as Claude Code, Cursor, Gemini CLI, and Google's own Antigravity platform to read those instructions directly when generating UI.
Turning design systems into portable text files
The idea is straightforward: if software projects have README.md files that explain what a codebase is and how to use it, design systems could have a similar file that explains visual rules in a machine-readable way. A DESIGN.md document can include core color definitions, semantic color usage, font families, sizes, weights, line heights, spacing scales, grid conventions, border radius, and shadow rules. It can also record the reasoning behind those choices, not just the choices themselves.
Google says there are 3 ways to create the file. Stitch's AI can generate one from a prompt, it can extract a design language from an existing website URL or image, or teams can write the Markdown manually. Once created, the file can be reused across any tool that supports the format rather than staying locked inside a single design app.
Giving AI coding agents actual design rules to follow
This addresses a gap in the current design-to-development workflow. The common process has been familiar for years: designers build layouts in Figma, annotate dimensions, colors, and spacing, and developers implement the interface from those specs. The rise of AI coding agents has not erased that pattern. Even if an agent can access Figma, that does not mean it can reliably infer the design intent inside the file.
DESIGN.md is intended to make that intent explicit. With the file in place, an AI agent can generate UI that follows a brand system rather than producing something that merely looks acceptable. That distinction matters. One output is based on inference; the other is based on a declared set of rules.
A signal to the broader design tool market
The move also has strategic weight for the design software market. One of Figma's longstanding strengths has been that designers work there and developers retrieve specs there. If AI agents can read DESIGN.md and generate interfaces that already match brand constraints, the handoff layer around design specs becomes smaller.
The source notes that Figma shares fell about 10% in a single day when Stitch 2.0 was released, a market reaction presented as a direct read on that threat. Competition in design tools used to center on which product made designers more efficient. The question now is shifting toward which product helps AI agents understand design more accurately. DESIGN.md is Google's early move in that contest.