Product designers
- Hand the scales, states and contrast checks to the engine, and keep the direction yourself.
- Every choice has a reason you can show in a design review.
- Export Figma variables so the design file and the code use the same names.
It asks the decisions a design system needs, one at a time, and suggests a sourced default for each. It checks contrast and writes design tokens, CSS, a Tailwind theme and a DESIGN.md that your team and your coding agents build from. Free and open source.
/plugin marketplace add ckryptickunal/OpenDesignerWorks withClaude Code, the Claude app, ChatGPT, Codex, Cursor and other tools that read Agent Skills.
How soft should corners feel: square, slightly rounded, rounded, or pill?
--ds-radius-control: var(--ds-radius-6); --ds-radius-container: var(--ds-radius-8); --ds-radius-overlay: var(--ds-radius-12);
Documentation
Process
Answer up to five questions in your AI tool, or point it at the app or brand you already have. You get a working draft of color, type, spacing, corners, depth and motion, each with its reason.
Import the tokens, the CSS variables or the Tailwind theme. Coding agents read DESIGN.md before they touch the interface, so new screens follow the same rules.
Changes go through a decision log that records who set what and why. A review command finds colors, sizes and radii in the code that drifted from the tokens.
--ds-space-16: 16px; --ds-radius-control: var(--ds-radius-6); --ds-color-brand-seed: #2338ff;
card.css:1 #f3f4f6 → use var(--ds-color-surface-sunken) card.css:1 padding: 13px → use var(--ds-space-12)
Who it is for
engine.py review names the token to use instead of a hard-coded value.Photos: Kaleidico, NEW DATA SERVICES, Declan Sun on Unsplash.
After the first run
Files in your project, readable by people and AI tools. Nothing lives on a server, and the same answers always give the same tokens and exports.
DESIGN.mdyour rules in plain wordsopendesigner/tokens/DTCG 2025.10build/css/tokens.cssCSS variablesbuild/tailwind/theme.cssTailwind v4build/figma/Figma variablesbuild/swift/, build/compose/iOS and Androidoverview.htmlthe system on one pagedecisions.mdevery choice and why
For teams
A written record your team and your AI tools build from, kept as files in your repository. It is early for teams: nobody has run a team pilot yet, and there is no hosted service and no SSO.
Every page
Status
Live from GitHub. Tell us where it breaks: that is the most useful thing right now.
FAQ
OpenDesigner is a free, open-source add-on for AI tools like Claude, ChatGPT, Codex and Cursor. It helps you make a design system: the rules for how your app looks, like colors, text sizes, spacing, corners and buttons. It reads your project, asks a few plain questions, suggests an answer for each with a reason, checks the result, and saves the rules as files in your project.
The code, skills and data are under the MIT licence. The written research and docs are under CC BY 4.0. You need your own AI tool, which may cost money. See the introduction.
Part of it is written instructions your AI follows, yes. The math is done by a Python engine. It works out color ramps, type and spacing scales, contrast checks and every export. The same answers give the same files. The questions follow a fixed order, so nothing is asked before the things it depends on. You can also run the engine with no AI at all. See How it works.
It is packaged for Claude Code, the Claude app, ChatGPT (as a Project), and for Codex, Cursor, VS Code with Copilot, Gemini CLI and other agents that read skills from .agents/. A recorded test run in each of those tools is still on the to-do list. See Install.
The engine runs on your computer, uses Python's standard library and makes no network calls. Your chat still goes to whichever AI provider you use, as with any prompt. The illustration skill searches open-licence art sites, but only when you use it. In the Claude app, the scripts run in Claude's code-execution sandbox.
The private log of your steps is off unless you say yes, and it stays in your project. Anonymous reports to the maintainers are a separate yes, and no server collects them yet, so nothing is sent today. See the privacy page.
The sketch is up to five questions. Its planned length is about 3 minutes, but nobody has timed real people yet. The engine part takes about two seconds. After that, you zoom in only where you want to. See Zoom levels.