chatgpt-apps

Scaffold and validate ChatGPT Apps SDK applications with MCP servers and widget UIs.

Updated Jun 14, 2026
One-click install
npx skills add https://github.com/ironkid90-s/lucky5-v7 --skill chatgpt-apps-ironkid90-s
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: chatgpt-apps
Source: https://github.com/ironkid90-s/lucky5-v7/tree/main/.github/skills/chatgpt-apps
Command: npx skills add https://github.com/ironkid90-s/lucky5-v7 --skill chatgpt-apps-ironkid90-s

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires @modelcontextprotocol/ext-apps, @modelcontextprotocol/sdk, zod, and includes scripts (resource) and references (resource) components.

What problem does it solve? Building ChatGPT Apps SDK applications requires coordinating an MCP server, widget UI, tool metadata, CSP settings, and deployment steps, and this Skill keeps that work aligned with current OpenAI documentation instead of outdated patterns. ## Core Features & Use Cases - Docs-First Scaffolding: Fetches current Apps SDK docs via the openai-docs skill or OpenAI developer docs MCP server before generating any code. - Archetype Classification: Classifies requests into tool-only, vanilla-widget, react-widget, interactive-decoupled, or submission-ready shapes before choosing examples or scaffolds. - Server and Widget Generation: Produces MCP servers with tool annotations, resource metadata, and CSP, plus widgets using the MCP Apps bridge with optional window.openai extensions. - Use Case: Ask for a ChatGPT app that searches a company knowledge base, and the Skill classifies it as tool-only, implements standard search and fetch tools, scaffolds the MCP server, and validates the repo contract. ## Quick Start Use the chatgpt-apps skill with openai-docs to scaffold a ChatGPT app for my use case with a TypeScript MCP server and a vanilla widget.

Frequently Asked Questions about chatgpt-apps

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I build a ChatGPT app with the Apps SDK?

Start by classifying the app archetype, then fetch current Apps SDK docs before writing code. Adapt the closest official OpenAI example or ext-apps example, scaffold the MCP server and widget, and validate against the minimum working repo contract.

What is the difference between the MCP Apps bridge and window.openai?

The MCP Apps bridge (ui/* notifications, tools/call, ui/message) is the portable standard for new apps. window.openai is a ChatGPT-specific compatibility and extension layer for features like file upload, modals, and display mode requests.

When should a ChatGPT app use the standard search and fetch tools?

Use the standard search and fetch tools when the app is connector-like, data-only, sync-oriented, or meant for company knowledge or deep research. They must be read-only and return JSON-encoded text content matching the documented schema.

How do I test a ChatGPT app locally before deployment?

Run the MCP server on localhost, expose it with a public HTTPS tunnel such as ngrok, then add the tunneled URL plus /mcp in ChatGPT Developer Mode under Settings, Apps and Connectors. Refresh the app after tool or metadata changes.

What metadata does a ChatGPT app widget resource need for submission?

Submission-ready apps need _meta.ui.domain, accurate _meta.ui.csp connectDomains and resourceDomains, and an openai/widgetDescription. The MCP server must run on a public production endpoint, not localhost or tunnel URLs.