Integrating AI tools, especially agents and multi-agent workflows, into the Message Compute Protocol (MCP) architecture often presents challenges for developers. Until now, many MCP servers have been implemented as standalone projects, making the integration of existing agent projects difficult. A new library called automcp and an associated deployment platform now promise a remedy and a significantly simplified conversion and deployment process.
The automcp library enables easy integration into existing agent projects. Developers can add it as a dependency to projects based on frameworks like CrewAI, LangGraph, Llama Index, OpenAI Agents SDK, Pydantic AI, and mcp-agent. Further frameworks are planned to be supported in the future. A simple CLI command generates a run_mcp.py file. After a few adjustments, this file can be executed to start the server locally. The run_mcp.py file functions similarly to a Procfile in Heroku, configuration files for Codespaces, or Infrastructure-as-Code solutions like Pulumi or AWS CDK.
In addition to the library, a deployment platform has been developed that further simplifies the deployment process. Developers simply enter the GitHub URL of their project and can deploy the server with one click. The platform generates a URL for the hosted SSE server, which can then be used with MCP clients like Cursor. This concept is similar to that of Vercel for web applications.
Although the platform already significantly simplifies the deployment process, some manual steps are still required. However, the platform's developers are working on further automating the process. A promising approach is the automatic creation of MCP servers for each orchestrator, agent, and tool in a project, instead of using a monolithic MCP server. This would further increase the modularity and flexibility of the architecture.
The automcp library and the associated deployment platform offer a promising solution for integrating agent projects into the MCP architecture. By simplifying the conversion and deployment process, developers can focus on developing and optimizing their agents and leverage the benefits of the MCP architecture. Further automation of the process and support for additional frameworks will further enhance the usability and benefits of these tools in the future.