MCP Agent

MCP Agent is a development framework that seamlessly integrates MCP rmcp tools with LLM async-openai clients. As a framework, it provides a structured way to build MCP tools by enabling developers to easily leverage the capabilities of large language models. The framework handles the integration of dependent MCP tools and LLM clients, allowing developers to focus on implementing their specific business logic.

Features

• Seamless integration of MCP tools with LLM clients
• Flexible configuration system
• Support for multiple MCP servers
• Configurable LLM parameters
• Simple and intuitive interface

Installation

You can install MCP Agent in several ways:

1. Using cargo add:

   cargo add mcp-agent

2. Or add the following to your Cargo.toml:

   # From crates.io
   mcp-agent = "0.1.0"
   
   # Or from GitHub
   mcp-agent = { git = "https://github.com/CeerDecy/mcp-agent", branch = "main" }

Configuration

MCP Agent uses a TOML configuration file to manage settings. The configuration file is typically named mcp-agent.toml. Here's an example configuration:

[mcp_servers.searxng]
command = "uvx"
args = ["mcp-searxng"]
[mcp_servers.searxng.env]
SEARXNG_URL = "https://searxng.example.com"

[mcp_servers.fetch]
command = "uvx"
args = ["mcp-server-fetch", "--ignore-robots-txt"]

[llm]
api_key = "your_api_key"
base_url = "https://api.openai.com/v1/"
model = "gpt-4.1"

Configuration Details

1. Dependent MCP Tools Configuration

   • Configure the third-party MCP tools that your project depends on
   • Each tool can have its own command, arguments, and environment variables
   • Support for multiple dependent MCP tools
   • These tools will be automatically managed by the framework

2. LLM Configuration

   • api_key: API key for the LLM service
   • base_url: Base URL for the LLM API
   • model: Name of the model to use

Usage Example

For detailed usage examples, please refer to the code in the examples directory. Here's a quick overview of the basic setup:

1. Create configuration file:

   cp examples/simple/mcp-agent.toml.template mcp-agent.toml

2. Edit the configuration file with your actual settings

3. Implement your MCP tool using the framework:

   use mcp_agent::*;
   
   // Your implementation here

For more detailed examples and best practices, please check:

• examples/simple/ - Basic usage example
• examples/advanced/ - Advanced usage patterns

Project Structure

mcp-agent/
├── src/
│   ├── agent/      # Core agent implementation
│   ├── llm/        # LLM client implementation
│   ├── mcp_server/ # MCP server implementation
│   └── lib.rs      # Library entry point
├── examples/       # Example code
└── Cargo.toml      # Project dependencies

Development

1. Clone the repository:

   git clone https://github.com/CeerDecy/mcp-agent.git
   cd mcp-agent

2. Install dependencies:

   cargo build

3. Run tests:

   cargo test

Contributing

Pull requests and issues are welcome!

License

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

Copyright (c) 2025 CeerDecy