Mcpadapt

🚀 MCPAdapt

MCPAdapt is an adapter designed to connect various models and tools to different conversational AI frameworks. It supports multiple popular AI frameworks and enables users to call tools on remote or local MCP servers through simple configuration.

🚀 Quick Start

Basic Usage

from mcpadapt import MCPAdapt

# Initialize the adapter
adapter = MCPAdapt()

# Register a tool
adapter.register_tool(tool_name, tool_function)

# Call the tool
result = adapter.call_tool(tool_name, parameters)

Supported Frameworks

• Smolagents
• LangChain
• CrewAI
• Google GenAI

📦 Installation

Install using pip:

pip install mcpadapt

✨ Features

1. Multi-framework Support: Easily connect to multiple popular AI frameworks.
2. Flexible Configuration: Call tools on the MCP server through simple parameter configuration.
3. Highly Scalable: Easily add support for new frameworks and tools.

💻 Usage Examples

Remote MCP Server

from mcpadapt import MCPAdapt

# Initialize the adapter and connect to the remote MCP server
adapter = MCPAdapt(remote_server="your-mcp-server-url")

# Define a simple tool function
def your_tool_function(parameters):
    # Process the parameters and return the result
    return {"result": "tool executed successfully"}

# Register the tool
adapter.register_tool("your_tool_name", your_tool_function)

# Call the tool
response = adapter.call_tool("your_tool_name", parameters={"key": "value"})

print(response)

Local MCP Server

from mcpadapt import MCPAdapt, serve

# Initialize the adapter and start the local MCP server
adapter = MCPAdapt()
serve(adapter, port=8000)

# Define a simple tool function
def your_tool_function(parameters):
    # Process the parameters and return the result
    return {"result": "tool executed successfully"}

# Register the tool
adapter.register_tool("your_tool_name", your_tool_function)

🤝 Contribution

If you find any issues or have suggestions for improvement, please feel free to submit an Issue or a Pull Request.

How to Add Support for a New Framework

1. Create an Adapter Class: Create a new adapter class in src/mcpadapt/{name_of_your_framework}_adapter.py:

from mcpadapt.base_adapter import ToolAdapter
from typing import Callable, Any

class YourFrameworkAdapter(ToolAdapter):
    def adapt(
        self,
        func: Callable[[dict | None], mcp.types.CallToolResult],
        mcp_tool: mcp.types.Tool,
    ) -> YourFramework.Tool:
        # Implement how to convert a simple function and an MCP tool (JSON Schema) into your framework tool
        pass
    
    def async_adapt(
        self,
        afunc: Callable[[dict | None], Coroutine[Any, Any, mcp.types.CallToolResult]],
        mcp_tool: mcp.types.Tool,
    ) -> YourFramework.Tool:
        # If your framework supports asynchronous functions, use async_adapt
        pass

2. Test and Submit: After implementation, make sure to add corresponding test cases and submit them to the repository to share with the community.

🗺️ Project Roadmap

• Completed:

  • Initial framework supporting the creation of multiple adapters.
  • Support for mainstream frameworks such as Smolagents, LangChain, CrewAI, and Google GenAI.
  • Support for connecting to remote and local MCP servers.

• Future Plans:

  • Add support for more AI frameworks (e.g., Pydantic-OpenAI).
  • Improve documentation by adding more examples and tutorials.
  • Enhance performance and stability.

👥 Contributors

We thank every developer and maintainer who has contributed to this project:

Core Maintainers:

• @grll

Contributors:

• @murawakimitsuhiro
• @joejoe2
• @tisDDM
• @sysradium

Thank you for your support of MCPAdapt!