惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

雷峰网
雷峰网
Y
Y Combinator Blog
酷 壳 – CoolShell
酷 壳 – CoolShell
The Cloudflare Blog
博客园_首页
J
Java Code Geeks
A
About on SuperTechFans
人人都是产品经理
人人都是产品经理
量子位
C
Check Point Blog
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
博客园 - 三生石上(FineUI控件)
L
LangChain Blog
N
Netflix TechBlog - Medium
Hugging Face - Blog
Hugging Face - Blog
B
Blog
美团技术团队
Microsoft Security Blog
Microsoft Security Blog
P
Proofpoint News Feed
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
宝玉的分享
宝玉的分享
罗磊的独立博客
MongoDB | Blog
MongoDB | Blog
Last Week in AI
Last Week in AI

DEV Community

Authentication Security Deep Dive: From Brute Force to Salted Hashing (With Java Examples) Why AI Systems Don’t Fail — They Drift Spilling beans for how i learn for exam😁"Reinforcement Learning Cheat Sheet" I Replaced Chrome with Safari for AI Browser Automation. Here's What Broke (and What Finally Worked) How Python Borrows Other People's Work The $40 Architecture: Processing 1 Billion API Requests with 99.99% Uptime Vibe Coding: A Workflow Guide (From Zero to SaaS) Most webhook security guides protect the wrong side. The scary part is delivery. Headless CMS for TanStack Start: Build a Blog with Cosmic EU Age Verification App "Hacked in 2 Minutes" — What Actually Happened Comfy Cloud’s delete function does not actually remove files Running AI Models on GPU Cloud Servers: A Beginner Guide Event-driven media intelligence with AWS Step Functions and Bedrock I scored 500 AI prompts across 8 quality dimensions — here's what broke How to Call Google Gemini API from Next.js (Free Tier, No Backend Needed) The Portal Protocol: Reclaiming Human Connection in the Age of AI How to Fix Your Team's Scattered Knowledge Problem With a Self-Hosted Forum Intro to tc Cloud Functors: A Graph-First Mental Model for the Modern Cloud Designing Multi-Tenant Backends With Both Ownership and Team Access I Built a Neumorphic CSS Library with 77+ Components — Here's What I Learned PostgreSQL Performance Optimization: Why Connection Pooling Is Critical at Scale Cómo construí un SaaS multi-rubro para gestionar expensas en Argentina con FastAPI + Vue 3 🚀 I Built an Ethical Hacking Scanner Tool – Open Source Project I Replaced /usage and /context in Claude Code With a Single Statusline A Pythonic Way to Handle Emails (IMAP/SMTP) with Auto-Discovery and AI-Ready Design I Collected 8.9 Million Polymarket Price Points — Here's What I Found About How Markets Really Move EcoTrack AI — Carbon Footprint Tracker & Dashboard Everyone's Using AI. No One Agrees How. 5 self-hosted ebook managers worth trying in 2026 Building Your First AI Agent with LangChain: From Chatbot to Autonomous Assistant
Skybridge V1.0: THE framework for building MCP apps
Julien Valli · 2026-05-19 · via DEV Community

Building an MCP App is a genuinely new problem. You're not building a website or a standard API; you're building something that runs inside an AI assistant, interacts with a non-deterministic model, and renders UI within a chat. A year ago, none of that existed. Today, MCP Apps run reliably in Claude and ChatGPT, the official SDK has matured, and +1,000 apps are present in the public app stores.

Earlier this year, we open-sourced Skybridge, a full-stack TypeScript framework for building MCP Apps. It handles the plumbing: the MCP server, view rendering, CSP configuration, dev tools, client compatibility, and the connection between your server logic and your UI components. The goal is to let you focus on what your app does, rather than how to wire it all together.

The response was bigger than we expected. Skybridge now has 100k monthly downloads and powers over 10% of apps on the Claude and ChatGPT stores, with teams ranging from Fortune 500 companies to early-stage startups. That adoption is what's driving this release.

Today we're thrilled to introduce Skybridge 1.0, a first-class framework with a stable API, a complete developer toolchain, and a clear commitment to the MCP Apps ecosystem. If you're building an MCP App, this is the foundation to build on.

Simpler, type-safe API

The goal of Skybridge has always been to abstract away underlying implementation details. In v1, that means one entry point: server.registerTool(config, handler). It replaces the old split between plain tools and tools with a view: include a view in the config if you need one, and Skybridge wires it up. The Vite plugin handles component discovery and bundling automatically. Less boilerplate, fewer wiring mistakes, and a cleaner mental model overall.

Think of Skybridge as the React Native of MCP App development: one codebase, compatible with every environment.

New skybridge API

Notably, we're the first framework to include strongly typed tool-to-component binding, which is particularly useful when building with agents, where small mistakes are difficult to track down. We've also standardized on the term "views" across the framework and docs.

A (really) complete dev loop

One of the hardest things about building MCP Apps has always been the feedback loop. You make a change, tunnel your local server, switch to ChatGPT or Claude, test, and repeat. For every small UI tweak or logic fix, that cycle compounds fast.

In V1, the dev tools have been redesigned from the ground up. They launch automatically when you run skybridge dev, with a layout reflecting how you actually work: views are now first-class in the interface, with a dedicated panel that makes it easy to see what your view looks like and how it responds to tool calls. There's a mobile preview mode for easy small-screen viewing.

The new and improved Skybridge dev tools, including a familiar helper we snuck into the template you get when you install Skybridge. He's been waiting 25 years for a comeback, and this time, he won't interrupt you.

Most importantly, the dev tools are now a unified control panel for your entire development workflow. From the dev tools header, you can:

  • Open a tunnel to test against a live AI client: Alpic Tunnel exposes your local server via a stable, authenticated URL that works with any MCP-compatible client.

  • Test your app in the Alpic playground, a production-like environment that avoids having to add your app to ChatGPT or Claude. It's the fastest way to iterate on tool behavior and UI together.

  • Audit your app with Beacon, a full compliance scan of your app against the Claude Connector Directory & ChatGPT App Store. Beacon catches the things that are easy to miss in local testing: CSP misconfigurations, missing tool descriptions, client-specific quirks.

  • Deploy directly to Alpic or wherever your stack lives once you're ready to ship.

We believe these devtools really combine everything a developer needs to build a production-grade MCP app with no platform or iteration friction.

Deploy everywhere

Lastly, we've long believed that a framework's value shouldn't come from locking you in.

1.0 makes good on that. Your app ships with a Dockerfile in the template, and the framework is now compatible with Cloudflare Workers. Alpic Cloud remains the fastest path to production if you want zero infrastructure work, but it's a choice, not a requirement.

In a similar vein, we've exposed the Express server giving you full control over the application and the ability to extend it with custom routes or middleware.

Getting started

For new projects:

npm create skybridge

Enter fullscreen mode Exit fullscreen mode

What's next?

V1 is a foundation, not a finish line, and it's a project we're investing more time into with every release. On the roadmap:

  • Dev tools: continued improvements, especially around CSP and auth debugging
  • Deployment: broader platform coverage
  • Protocol: simplified auth configuration & expanded client support
  • Agent-friendliness: programmatic devtool APIs, better plugins, and more

Is there something you'd like to see? Open an issue or find us on Discord!

Check out our latest Code with Fred to see the dev tools live in action.