composiohq/composio — reverse-engineered prompt

Reverse engineered prompt

GitHub

Build me a Composio style SDK repo that lets developers give their AI agents real world tools without wiring every API by hand. It should have both TypeScript and Python clients, with a simple setup where someone adds an API key, chooses a user, asks for tools from a toolkit like Hacker News, and plugs those tools into an AI agent.

Make it work cleanly with popular agent providers like OpenAI, Anthropic, LangChain, LangGraph, LlamaIndex, Gemini, Vercel AI SDK, Mastra, Cloudflare Workers AI, CrewAI, and AutoGen where each language supports them. Include authentication handling, tool lookup, context management, and a safe workbench for trying tool actions.

Please add clear install instructions, quick start examples, provider examples, basic tests, and docs for the TypeScript and Python SDKs. Look up current docs online if you need to.

Want more depth? Deep Reverse