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

freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
V
Visual Studio Blog
IT之家
IT之家
博客园 - 聂微东
The Cloudflare Blog
月光博客
月光博客
阮一峰的网络日志
阮一峰的网络日志
S
SegmentFault 最新的问题
Apple Machine Learning Research
Apple Machine Learning Research
酷 壳 – CoolShell
酷 壳 – CoolShell
爱范儿
爱范儿
H
Help Net Security
博客园 - 叶小钗
V
V2EX
WordPress大学
WordPress大学
J
Java Code Geeks
Hugging Face - Blog
Hugging Face - Blog
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻
博客园_首页
C
Check Point Blog
B
Blog
D
DataBreaches.Net
美团技术团队
罗磊的独立博客

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
MCP Apps vs OpenAI Apps SDK: are they competing standards?
Muhammad Moeed · 2026-06-12 · via DEV Community

Muhammad Moeed

If you have started building tools for AI chat hosts, you have probably hit the same fork in the road. There are two ways to add a user interface to a tool right now. The first is MCP Apps, the open standard from the Model Context Protocol. The second is the OpenAI Apps SDK, the framework OpenAI shipped for the ChatGPT app store.

They both let your tool render a UI inside the chat. They both run inside a sandboxed iframe. They both talk to your server over JSON-RPC. So the question almost every developer is asking is the obvious one: are these competing standards?

The short answer

No. The OpenAI Apps SDK is built on top of MCP. It uses MCP for the wire format, the tool definitions, the UI rendering, and the security model. It then adds a thin layer with ChatGPT-specific extensions: a payment hook (window.openai.requestCheckout), a store directory, and a few helpers that only run inside ChatGPT.

So the choice is not "pick one or the other." The honest framing is:

  • Build to MCP Apps first. Your tool then works in Claude, Claude Desktop, VS Code, Goose, Postman, MCPJam, and any host that supports the open standard.
  • Light up OpenAI Apps SDK extensions only when you need ChatGPT-specific features. Mostly that means payments and the ChatGPT app store.

The four real differences

Strip away the marketing and there are only four things the OpenAI Apps SDK gives you that MCP Apps do not:

  1. A store and a directory — the ChatGPT app store will list your app and surface it to ChatGPT users. The open MCP spec has no centralised discovery.
  2. In-chat paymentswindow.openai.requestCheckout() runs a payment flow without leaving the chat. MCP Apps has no checkout primitive today.
  3. Conversation-aware helpers — tighter access to surrounding conversation state.
  4. Distribution to ChatGPT users specifically — ChatGPT has hundreds of millions of users; the Apps SDK is the only way to reach them through OpenAI's surface.

Everything else — the sandboxing, the iframe rendering, the JSON-RPC audit trail, the tool declarations — is the same on both sides because the Apps SDK uses MCP for them.

Which one to pick

If your audience is ChatGPT users specifically and you need in-chat payments, build with the OpenAI Apps SDK. For everyone else, build to MCP Apps and your tool runs across every host that supports the open standard.

I wrote the full comparison — code examples for both sides, the smart progressive-enhancement pattern, a decision tree, and the two pitfalls most teams hit — on my blog.

Read the full guide on moeed.app →