Defi Yields MCP

🚀 DeFi Yields MCP Server

An MCP server for AI agents to explore DeFi yield opportunities, powered by DefiLlama.

🚀 Quick Start

The DeFi Yields MCP Server is designed to assist AI agents in exploring DeFi yield opportunities. It offers a range of features to simplify data acquisition and analysis.

✨ Features

• Data Acquisition Tool: The get_yield_pools tool fetches DeFi yield pool data from DefiLlama. It supports filtering by chain (e.g., Ethereum, Solana) or project (e.g., Lido, Aave).
• Analysis Prompts: The analyze_yields prompt generates customized instructions for AI agents to analyze yield pool data, focusing on key metrics such as APY, 30 - day average APY, and predictions.
• Packaged for Easy Use: You can run the server directly via uvx defi-yields-mcp.

📦 Installation

To use this server with Claude Desktop, you can choose either automatic installation or manual configuration of the Claude Desktop configuration file.

Option 1: Automatic Installation

Install the server for Claude Desktop:

uvx mcp install -m defi_yields_mcp --name "DeFi Yields Server"

Option 2: Manual Configuration

Locate the configuration file:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Add the server configuration:

{
 "mcpServers": {
   "defi-yields-mcp": {
     "command": "uvx",
     "args": [ "defi-yields-mcp" ]
   }
 }
}

Restart Claude Desktop.

💻 Usage Examples

You can use the following commands:

• "Get the yield pools for the Lido project."
• "Analyze the yield pools on Ethereum."
• "What is the 30 - day average APY of Solana pools?"

The get_yield_pools tool retrieves and filters data, while the analyze_yields prompt guides the LLM for detailed analysis.

Basic Usage

When running the get_yield_pools tool with Ethereum as the filtering condition:

[
  {
    "chain": "Ethereum",
    "pool": "STETH",
    "project": "lido",
    "tvlUsd": 14804019222,
    "apy": 2.722,
    "apyMean30d": 3.00669,
    "predictions": {
        "predictedClass": "Stable/Up",
        "predictedProbability": 75,
        "binnedConfidence": 3      
    }
  },
  ...
]

📄 License

This project is open - source under the MIT License. For more details, please refer to the LICENSE file.