FastMCP with Claude Tool Integration Example

This repository demonstrates how to implement a web server that integrates Claude's tools capabilities with the FastMCP (Model Control Protocol) framework. It provides a complete example of defining custom tools that Claude can use to perform calculations and retrieve information.

Key Features

• Claude Tool Integration: Uses Claude's tool use capabilities with proper conversation flow and tool_use_id matching
• FastMCP Tool Definitions: Utilizes FastMCP to define and manage tool schemas and implementations
• Asynchronous Web Server: Built with Quart for high-performance async handling of requests
• Complete Conversation Flow: Demonstrates the full conversation loop with tool calls and tool results
• Error Handling: Robust error handling for tool execution and API communication
• Example Tools: Includes sample calculator and weather lookup tools

How It Works

1. Tool Definition: Tools are defined using the FastMCP decorators, which automatically generate the appropriate schemas
2. API Integration: The server converts FastMCP tool definitions to Claude's expected format
3. Conversation Flow: Manages the complete conversation flow, including preserving tool_use_id references
4. Tool Execution: Executes the appropriate tool functions when Claude calls them
5. Result Handling: Formats and returns tool results in the format Claude expects

Getting Started

Prerequisites

• Python 3.8+
• Anthropic API key
• Docker (optional, for running with PostgreSQL for the memory example)

Installation

1. Clone this repository:

git clone https://github.com/yourusername/fastmcp-claude-tools.git
cd fastmcp-claude-tools

1. Install dependencies:

pip install -r requirements.txt

1. Set your Anthropic API key as an environment variable:

Windows PowerShell:

$env:ANTHROPIC_API_KEY="your-api-key-here"

Windows CMD:

set ANTHROPIC_API_KEY=your-api-key-here

Linux/macOS:

export ANTHROPIC_API_KEY="your-api-key-here"

Running the Example

1. Start the server:

python app.py

1. In another terminal, run the client:

python client.py

Implementation Details

The repository demonstrates several key aspects of integrating Claude with FastMCP:

1. Tool Schema Generation: The application properly extracts tool schemas from FastMCP, avoiding hardcoded definitions
2. Proper Tool Use Handling: Implements the Claude API's expected format for tool use and tool results
3. Conversation Context Preservation: Maintains the conversation context with proper tool_use references
4. Error Handling: Gracefully handles errors in tool execution and API communication
5. Python Type Hints: Uses Python's type hints for automatic schema generation

Important Components

• app.py: Main server implementation with tool definitions and Claude API integration
• client.py: Simple client to demonstrate how to interact with the server
• memory_app.py and memory_system.py: Advanced example with a memory system (optional)

Claude Tool Use Control

The implementation demonstrates various methods to control Claude's tool use behavior:

1. Tool Choice: Configure whether Claude should use tools, which tools to use, etc.
2. Chain of Thought: Encourage Claude to explain its reasoning in thinking tags
3. Tool Use Structure: Properly structure the conversation to maintain tool context

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgements

• Anthropic for Claude API
• FastMCP for the Model Control Protocol framework
• Quart for the async web framework