💡 Summary
Skybridge is a fullstack TypeScript framework for building AI-embedded widgets with type safety and React support.
🎯 Target Audience
🤖 AI Roast: “Powerful, but the setup might scare off the impatient.”
Risk: Medium. Review: shell/CLI command execution; outbound network access (SSRF, data egress); filesystem read/write scope and path traversal; dependency pinning and supply-chain risk. Run with least privilege and audit before enabling in production.
Skybridge
Build ChatGPT & MCP Apps. The Modern TypeScript Way.
The fullstack TypeScript framework for AI-embedded widgets. Type-safe. React-powered. Platform-agnostic.
Documentation · Quick Start · Showcase
✨ Why Skybridge?
ChatGPT Apps and MCP Apps let you embed rich, interactive UIs directly in AI conversations. But the raw SDKs are low-level—no hooks, no type safety, no dev tools, and no HMR.
Skybridge fixes that.
| | |
|:--|:--|
| 👨💻 Full Dev Environment — HMR, debug traces, and local devtools. No more refresh loops. | ✅ End-to-End Type Safety — tRPC-style inference from server to widget. Autocomplete everywhere. |
| 🔄 Widget-to-Model Sync — Keep the model aware of UI state with data-llm. Dual surfaces, one source of truth. | ⚒️ React Query-style Hooks — isPending, isError, callbacks. State management you already know. |
| 🌐 Platform Agnostic — Write once, run anywhere. Works with ChatGPT (Apps SDK) and MCP-compatible clients. | 📦 Showcase Examples — Production-ready examples to learn from and build upon. |
🚀 Get Started
Create a new app:
npm create skybridge@latest
Or add to an existing project:
npm i skybridge yarn add skybridge pnpm add skybridge bun add skybridge deno add skybridge
👉 Read the Docs 👈
📦 Architecture
Skybridge is a fullstack framework with unified server and client modules:
skybridge/server— Define tools and widgets with full type inference. Extends the MCP SDK.skybridge/web— React hooks that consume your server types. Works with Apps SDK (ChatGPT) and MCP Apps.- Dev Environment — Vite plugin with HMR, DevTools emulator, and optimized builds.
Server
import { McpServer } from "skybridge/server"; server.registerWidget("flights", {}, { inputSchema: { destination: z.string() }, }, async ({ destination }) => { const flights = await searchFlights(destination); return { structuredContent: { flights } }; });
Widget
import { useToolInfo } from "skybridge/web"; function FlightsWidget() { const { output } = useToolInfo(); return output.structuredContent.flights.map(flight => <FlightCard key={flight.id} flight={flight} /> ); }
🎯 Features at a Glance
- Live Reload — Vite HMR. See changes instantly without reinstalling.
- Typed Hooks — Full autocomplete for tools, inputs, outputs.
- Widget → Tool Calls — Trigger server actions from UI.
- Dual Surface Sync — Keep model aware of what users see with
data-llm. - React Query-style API —
isPending,isError, callbacks. - Platform Agnostic — Works with ChatGPT (Apps SDK) and MCP Apps clients (Goose, VSCode, etc.).
- MCP Compatible — Extends the official SDK. Works with any MCP client.
📖 Showcase
Explore production-ready examples:
| Example | Description | Demo | Code | |------------------------|----------------------------------------------------------------------------------|-----------------------------------------------------|-------------------------------------------------------------------------------------| | Capitals Explorer | Interactive world map with geolocation and Wikipedia integration | Try Demo | View Code | | Ecommerce Carousel | Product carousel with cart, localization, and modals | Try Demo | View Code | | Everything | Comprehensive playground showcasing all hooks and features | Try Demo | View Code | | Productivity | Data visualization dashboard demonstrating Skybridge capabilities for MCP Apps | Try Demo | View Code |
See all examples in the Showcase or browse the examples/ directory.
See CONTRIBUTING.md for setup instructions
MIT License · Made with ❤️ by Alpic
Pros
- Type-safe development environment
- Rich interactive UI capabilities
- Cross-platform compatibility
- Comprehensive documentation and examples
Cons
- Steeper learning curve for beginners
- Potentially over-engineered for simple projects
- Dependency on TypeScript and React knowledge
- Limited community compared to more established frameworks
Related Skills
pytorch
S“It's the Swiss Army knife of deep learning, but good luck figuring out which of the 47 installation methods is the one that won't break your system.”
agno
S“It promises to be the Kubernetes for agents, but let's see if developers have the patience to learn yet another orchestration layer.”
nuxt-skills
S“It's essentially a well-organized cheat sheet that turns your AI assistant into a Nuxt framework parrot.”
Disclaimer: This content is sourced from GitHub open source projects for display and rating purposes only.
Copyright belongs to the original author alpic-ai.
