A library and platform to easily convert and deploy your existing agent and multi-agent projects as MCP servers for use with MCP clients like Cursor and Claude Desktop.
Built something with CrewAI, LangGraph, LlamaIndex, OpenAI’s Agents SDK, Pydantic AI, or mcp-agent? Now you can turn it into an MCP server without modifying any project code.
To use an existing agent that you’ve built via MCP clients like Cursor, you shouldn’t need to be a backend engineer. That’s why we built AutoMCP.
Just add the automcp library to your project, run our CLI, and we’ll scaffold a run_mcp.py file. With a few small edits, you're up and running locally.
Even better, we built a platform to deploy your existing agent codebase:
🖥️ Paste your GitHub repo
⚡ Click deploy
🔗 Get a live MCP server URL
It’s like Vercel or Heroku — for AI agents.
We’re working on streamlining the user flow, as well as building a publishing layer so you can share, remix and compose agents across the ecosystem.
🎥 Try the CLI, use the deployment platform and watch the demo: https://auto-mcp.com
We’d love to hear what you're building, and what you want to build next.
MCPify is a free library and platform that helps you turn your existing agent and multi-agent projects into ready-to-use MCP servers. Connect seamlessly with MCP clients like Cursor and Claude Desktop. No need to rewrite anything—just convert, deploy, and start using your agents in powerful desktop environments. Clean, simple, and AI-free.
Google introduced A2A, the Agent-to-Agent protocol. What are your thoughts on it?
Wow — AutoMCP makes deploying AI agents so much easier. Love that you can go from GitHub repo to live MCP server in one click. Great job on removing friction for devs 👏
This is 🔥 — AutoMCP feels like the missing piece for making agent infra actually usable.
Being able to turn any project into an MCP server with almost zero refactoring is a huge unlock, especially for devs who just want to ship and not get bogged down in backend scaffolding. The Vercel-for-agents angle is super compelling too — deploy from GitHub and you’re live? Yes, please.
Really excited about the publishing/remix layer you mentioned — feels like it could power an entirely new ecosystem of reusable agent behaviors.
Massive congrats on the launch. Can’t wait to see how this evolves! 🚀
This is exactly what’s needed to bring more flexibility to the agent ecosystem. Making it easier to deploy existing projects into MCP-compatible environments is a big win.