langchain -client

The LangChain MCP Client is a Python-based client that connects to MCP servers and converts their available tools into LangChain-compatible tools. It enables you to seamlessly integrate multiple MCP servers and use them within LangChain workflows, leveraging a chosen LangChain-compatible LLM for decision making. The core feature is convert_mcp_to_langchain_tools(), which initializes multiple MCP servers in parallel and exposes their tools as a List[BaseTool] that LangChain can consume in your agents or tools pipelines. This client also enables a CLI workflow so you can interactively issue queries and test tool availability against connected MCP servers.

To use it, install the package, configure your MCP servers, and call convert_mcp_to_langchain_tools() in your code. You can then plug the resulting tools into a LangChain ReAct or other agent framework, experiment with tool invocation sequences, and iterate with example queries defined in your llm_mcp_config.json5 configuration. This makes it straightforward to swap MCP endpoints or add new servers without changing your core LangChain integration.

Prerequisites:

• Python 3.11 or higher
• pip (comes with Python)

Installation steps:

# Optional: create a virtual environment
python -m venv venv
source venv/bin/activate  # on Linux/macOS
venv\Scripts\activate.bat # on Windows

# Install the LangChain MCP Client
pip install langchain_mcp_client

Configuration steps:

1. Prepare environment variables and API keys as needed for your LLMs or MCPs. Create a .env file if desired and load them in your application.
2. Create or modify llm_mcp_config.json5 to specify LLM settings, MCP server connections, and example prompts that demonstrate how to invoke MCP tools. Refer to your MCP server configuration for compatibility.
3. In your Python script, import and use convert_mcp_to_langchain_tools() to obtain tools, then build your LangChain agent or workflow as you normally would.

# Example usage in code (conceptual)
from langchain_mcp_client import convert_mcp_to_langchain_tools

tools = convert_mcp_to_langchain_tools(
    mcp_server_names=["langchain-mcp-client"]
)
# Use `tools` with a LangChain agent as you would with any list of BaseTool instances

Tips and considerations:

• Ensure Python 3.11+ is active in your environment when running the MCP client.
• The mcp_config entry names (e.g., server-name) can be customized; you’ll reference them in your code when selecting which MCPs to connect to.
• If your MCP servers require API keys or environment-specific credentials, provide them via the env field in the mcp_config or load from a .env file at runtime.
• The README mentions using llm_mcp_config.json5; this file should define LLM parameters, MCP server connections, and example prompts to guide tool invocation. Adjust the configuration to reflect your environment and desired tool sets.
• If you encounter network or authentication errors, verify that MCP server endpoints are reachable and that any required API keys are correctly provided.
• The npm_package field is null for this Python-based client; Node.js users should use the appropriate npm package if available.