Agentipy Mcp Server

What is Agentipy Mcp Server?

Agentipy Mcp Server is a Model Context Protocol (MCP) server that provides onchain tools for Claude Desktop, allowing AI agents to interact with the Solana blockchain through a standardized interface. It uses the Agentipy library to enable blockchain operations such as balance checks, token transfers, trading, staking, price predictions, and token information retrieval via CoinGecko and Pyth Network.

How to use Agentipy Mcp Server?

Install by cloning the repository, creating a Python virtual environment, and installing dependencies. Configure a .env file with your Solana private key, RPC URL, and optional API keys. Then add the server to Claude Desktop by editing claude_desktop_config.json and pointing to the run_mcp.sh (Unix) or run_mcp.bat (Windows) script. Restart Claude Desktop to activate the tools.

Key features of Agentipy Mcp Server

• Check Solana wallet balances
• Transfer tokens between wallets
• Deploy new tokens on Solana
• Get price predictions via Allora
• Trade and stake tokens using Jupiter
• Retrieve token info from CoinGecko

Use cases of Agentipy Mcp Server

• Ask Claude to check your Solana wallet balance
• Have Claude transfer SOL or SPL tokens to another address
• Deploy a new token from a natural language prompt
• Get real-time price predictions for crypto assets
• Execute token trades or staking through Jupiter

FAQ from Agentipy Mcp Server

What are the runtime requirements?

Python 3.8 or higher is required. The server runs in a local Python environment and communicates with Claude Desktop via the MCP protocol.

How do I configure Claude Desktop to use the server?

Edit claude_desktop_config.json (location varies by OS) and add the server entry with the correct path to run_mcp.sh or run_mcp.bat, along with environment variables for RPC URL, private key, and optional API keys.

Where does my private key and data live?

Your Solana private key and RPC URL are stored in a local .env file and passed as environment variables to the server. All blockchain data is retrieved from the Solana RPC endpoint; no data is stored on the MCP server.

Are there any known limitations?

The README does not mention specific limitations. Security notes advise using a dedicated wallet for AI operations and testing on devnet/testnet before mainnet.

What transports / authentication are used?

The server uses the standard MCP transport (stdio) and relies on the Solana private key for onchain authentication. No additional authentication layer is described.