Local-first, model-agnostic desktop AI research workbench for scientists, with agents, Python/R notebooks, scientific connectors and traceable artifacts.
nexu-io/open-design
Local-first, open-source AI design agent workspace that turns briefs into prototypes, decks, dashboards, images and videos using your own coding agent.
Open-source alternative to:Figma
- Stars
- 102,364
- Forks
- 11,743
- Open issues
- 1,208
- Last commit
- yesterday
What is it?
OpenDesign is a workspace for creating design work with AI agents. You start from a brief, pick a plugin or template, and an agent such as Claude Code, Codex, Cursor or DeepSeek Harness renders the output against a DESIGN.md brand contract. Supported outputs include web and mobile prototypes, live dashboards, slide decks, images and HyperFrames motion graphics, which can be exported to HTML, PDF, PPTX, MP4 or Markdown.
The project runs locally as a desktop app, a web UI, or via Docker, and it can also be used without a UI as an MCP server or CLI (the od command) inside a coding agent. It ships with 151 design-system packages, 100+ agent skills, and hundreds of plugins and templates. Teams can bring their own model keys through a BYOK proxy with SSRF protections.
The repository is licensed under Apache 2.0 and is written mainly in TypeScript. Linux users currently need to run it from source, since no prebuilt Linux release is published.
Key features
- 01Generates web and mobile prototypes, dashboards, decks, images and videos from a brief.
- 02Works with multiple coding agent CLIs and a BYOK API proxy for model providers.
- 03Applies a DESIGN.md brand contract and 151 bundled design-system packages.
- 04Exports decks to HTML, PDF, PPTX, ZIP or Markdown, and motion graphics to MP4.
- 05Exposes projects and files to other agents through a stdio MCP server and od CLI.
Use cases
- Turning a product brief into a clickable web or mobile prototype before engineering starts.
- Producing slide decks and marketing visuals that follow a team's existing brand system.
- Refactoring an existing git codebase's components to match a DESIGN.md spec.
- Letting a coding agent in another repo read live design tokens and components from local projects.
Quick start
- 1
Clone the repository
- 2
Read the quickstart guide for setup commands
Things to know
- Requires a coding agent CLI or a paid/own API key to generate designs.
- Some features like the desktop app and cloud offering depend on external services.
Topics
FAQ
- What is open-design?
- Local-first, open-source AI design agent workspace that turns briefs into prototypes, decks, dashboards, images and videos using your own coding agent. OpenDesign is a workspace for creating design work with AI agents. You start from a brief, pick a plugin or template, and an agent such as Claude Code, Codex, Cursor or DeepSeek Harness renders the output against a DESIGN.md brand contract. Supported outputs include web and mobile prototypes, live dashboards, slide decks, images and HyperFrames motion graphics, which can be exported to HTML, PDF, PPTX, MP4 or Markdown.
- Is open-design free to use? What license does it have?
- open-design is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is open-design still maintained?
- Actively maintained. The last commit to open-design was on Oct 19, 2026.
- What can I use open-design for?
- Turning a product brief into a clickable web or mobile prototype before engineering starts. Producing slide decks and marketing visuals that follow a team's existing brand system. Refactoring an existing git codebase's components to match a DESIGN.md spec. Letting a coding agent in another repo read live design tokens and components from local projects.
- How do I install open-design?
- Clone the repository Read the quickstart guide for setup commands
- Where can I get open-design?
- The source code and installation instructions are on GitHub: https://github.com/nexu-io/open-design
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
Open-source, local-first 3D building editor built on React Three Fiber and WebGPU, usable in the browser or from a CLI with MCP tools for AI agents.
A local-first AI office workbench that turns one request into editable PPTX, DOCX, XLSX and HTML files, using your own model keys.
Local-first open-source AI voice studio for cloning voices, generating speech, dictating into any app, and giving AI agents a voice.
A terminal coding agent with built-in LSP, debugger, browser and subagent tools, supporting 60+ model providers.
An agentic IDE built around the pi coding agent, running each task in isolated git-worktree workspaces with living specs and reusable skills.