OpenDesign

The open-source alternative to Claude Design — local-first, BYOK, your agent ships real designs.

🌐 open-design.ai  ·  ⭐ Star on GitHub  ·  𝕏 @nexudotio  ·  🇨🇳 Feishu Chinese Community / 飞书中文社区  ·  ✉ support@open-design.ai

---

Why OpenDesign exists

The strongest coding agents already live on your laptop. Claude Code, Codex, Cursor Agent, Gemini CLI, OpenCode, Qwen — all real, all yours, all BYOK. The missing piece isn't another agent runtime. It's the open-source surface area that turns those raw CLIs into a coworker that ships designs, joins your IM, runs skills, and remembers what happened yesterday.

That is what we build. Local-first, web-deployable, BYOK at every layer. Apache-2.0 / MIT, no telemetry, no lock-in.

What we ship

🎨 open-design · ⭐ 93.2k The open-source alternative to Claude Design. Local-first, web-deployable, BYOK. 259+ composable Skills + 142+ brand-grade Design Systems. Drives Claude Code / Codex / Cursor / Gemini / OpenCode / Qwen as the design engine. pnpm dev:all → magazine deck, mobile prototype, dashboard, docs page — sandboxed iframe preview, exported as HTML / PDF / PPTX / ZIP.	✨ html-anything · ⭐ 8.6k The agentic HTML editor — your local AI agent writes the HTML, you ship it. 75 Skills × 9 Surfaces (magazine · deck · poster · XHS / tweet · prototype · data report · Hyperframes). Sandboxed preview, 1-click export to WeChat / X / Zhihu / HTML / PNG. Zero API key — runs on Claude Code / Cursor / Codex / Gemini / Copilot / OpenCode / Qwen / Aider.
🦞 nexu · ⭐ 3.3k The open-source desktop client that connects your AI Agent to WeChat · Feishu · Slack · Discord in one click. Works with OpenClaw, Claude Code, Codex, GLM, MiniMax. Double-click install, OAuth or BYOK, local-first — your data never leaves your machine, your agent is on your phone 24/7.	🎬 html-video · ⭐ 4.5k Programmatic video for coding agents — HTML to video on your laptop. Turn HTML, CSS & data into real MP4s with pluggable render engines, 21 templates, AI soundtrack. Apache-2.0, no per-render fees. 🌐 open-design.ai/html-video
🌀 motion-anything · ⭐ 743 The agentic motion layer — describe the feeling, your AI ships the animation. Chat-native motion editing on running HTML pages (component by component) + a launch-video compositor with in-browser MP4 / GIF export, no watermark. 403 motion recipes, 8 agent engines + BYOK, exports JSON / CSS / React / Lottie / MP4 / GIF.

Principles

• Bring your own agent. We don't ship a model. The strongest agents already live on your laptop.
• Open by default. Apache-2.0 / MIT across the stack. No locked surface, no telemetry.
• Local-first. Your projects, conversations, and IM bridges run on your machine first, cloud second.
• Skills are files, not plugins. Drop a folder, restart, it works. The SKILL.md convention, end of story.
• Design Systems are portable Markdown, not theme JSON. The 9-section DESIGN.md schema.

Get involved

• Star the projects that resonate — cheapest, loudest signal.
• Add a skill — drop a SKILL.md folder and open a PR on open-design or html-anything.
• Add a design system — drop a DESIGN.md folder and open a PR.
• Add a video template — drop a template into html-video and open a PR.
• Add a motion recipe — drop a recipe folder into motion-anything and open a PR.

Issues, PRs, and Discussions are open across every public repo.

---

_{OpenDesign — next to you, by your AI coworker.
Apache-2.0 / MIT  ·  support@open-design.ai}