Langchain Box MCP Adapter

🚀 langchain-box-mcp-adapter

This sample project implements a Langchain MCP adapter for the Box MCP server. It demonstrates how to integrate Langchain with the Box MCP server, utilizing tools and agents.

🚀 Quick Start

This project showcases the integration of Langchain with the Box MCP server, enabling seamless interaction between AI models and the server through tools and agents.

✨ Features

• Langchain Integration: Leverage Langchain's ChatOpenAI model for AI interactions.
• MCP Server Communication: Connect to the Box MCP server via stdio transmission.
• Tool Loading: Dynamically load tools from the MCP server.
• Agent Creation: Create a React-style agent to handle user prompts and tool interactions.
• Rich Console Output: Provide a user-friendly console interface with Markdown rendering and typewriter effects.

📦 Installation

Prerequisites

• Python 3.13 or higher
• Dependencies listed in pyproject.toml:
  • langchain-mcp-adapters>=0.0.8
  • langchain-openai>=0.3.12
  • langgraph>=0.3.29
  • rich>=14.0.0

Steps

1. Clone the repository:

git clone <repository-url>
cd langchain-box-mcp-adapter

2. Install dependencies:

uv sync

3. Create a .env file in the project root directory and fill in the information:

LANGSMITH_TRACING = "true"
LANGSMITH_API_KEY =
OPENAI_API_KEY =

BOX_CLIENT_ID = ""
BOX_CLIENT_SECRET = ""
BOX_SUBJECT_TYPE = "user"
BOX_SUBJECT_ID = ""

4. Ensure the MCP server is set up and accessible at the specified path in the project.
5. Update StdioServerParameters in src/simple_client.py or src/graph.py with the correct path to your MCP server script:

server_params = StdioServerParameters(
    command="uv",
    args=[
        "--directory",
        "/your/absolute/path/to/the/mcp/server/mcp-server-box",
        "run",
        "src/mcp_server_box.py",
    ],
)

💻 Usage Examples

Basic Usage

To run the simple client:

uv run src/simple_client.py

This will launch a console-based application where you can interact with the AI agent. Enter prompts, and the agent will respond using tools and AI capabilities.

Advanced Usage

To use the graph-based agent (LangGraph), call the make_graph function in src/graph.py:

uv run langgraph dev --config src/langgraph.json

You should see something like this:

📚 Documentation

Project Structure

• src/simple_client.py: The main program entry point.
• src/graph.py: Contains the graph-based agent setup.
• src/console_utils/console_app.py: Utility functions for console interaction.
• src/langgraph.json: Configuration file for LangGraph integration.

📄 License

This project is licensed under the MIT License. For more details, see the LICENSE file.

🔗 Contribution

Contributions are welcome! Please open an issue or submit a Pull Request for any improvements or fixes.