LANGGRAPH-MCP-AGENTS

What is LANGGRAPH-MCP-AGENTS?

LangGraph MCP Agents is an experimental project that transforms tools defined under the Modular Communication Protocol (MCP) into autonomous agents using the LangGraph framework. It enables effective task sharing and collaboration between these agentified MCP tools, such as planning, coding, and execution tasks.

How to use LANGGRAPH-MCP-AGENTS?

Clone the repository, install dependencies with pip install -r requirements.txt, then run python app_cli.py to start the CLI chatbot. The project uses a mcp_config.json to configure MCP servers (filesystem, SQLite, memory) and an mcp_client.py module to manage server connections and tool execution.

Key features of LANGGRAPH-MCP-AGENTS

• Transforms MCP tools into LangGraph‑powered autonomous agents.
• Manages multiple MCP server connections and tool initialization.
• Provides a command‑line chatbot interface using the OpenAI language model.
• Supports tool execution and resource cleanup via MCPServer class.
• Configurable via mcp_config.json for filesystem, SQLite, and memory servers.
• Licensed under the MIT License – free to use, modify, and distribute.

Use cases of LANGGRAPH-MCP-AGENTS

• Converting existing MCP tools (e.g., code execution, planning) into collaborative agents.
• Building a multi‑agent chatbot that delegates tasks to different MCP servers.
• Experimenting with agent orchestration on top of the MCP protocol using LangGraph.
• Prototyping autonomous workflows that combine filesystem, database, and memory tools.

FAQ from LANGGRAPH-MCP-AGENTS

What are the runtime requirements?

Python and pip are required. All dependencies are listed in requirements.txt and include mcp, langgraph, langchain, python-dotenv, langchain-mcp-adapters, and rich.

How do I install LANGGRAPH-MCP-AGENTS?

Clone the repository, navigate to the project directory, and run pip install -r requirements.txt. No additional package manager is needed.

Is LANGGRAPH-MCP-AGENTS production-ready?

No. The project is explicitly marked as experimental and intended for exploration and prototyping.

What servers are supported by default?

The mcp_config.json provides example configurations for filesystem, SQLite, and memory MCP servers. The MCPClient class can load arbitrary server configurations.

How are MCP servers managed?

The mcp_client.py module contains MCPClient and MCPServer classes. MCPClient manages connections to multiple servers, starts them, initializes tools, and handles cleanup.