Dune Analytics MCP Server provides an MCP interface that exposes Dune Analytics data to AI agents. It offers two core tools: get_latest_result and run_query, both returning results as CSV-formatted strings for easy ingestion by downstream AI workflows. Prerequisites include Python 3.10+ and a valid Dune API key. The project can be installed automatically via Smithery for Claude Desktop or installed manually by cloning the repo and setting DUNE_API_KEY in a .env file or as a system environment variable. Development mode can be run with mcp dev main.py to enable hot reloading. To deploy for Claude Desktop, use mcp install main.py --name "Dune Analytics". Tool usage includes get_latest_result(query_id) and run_query(query_id) with corresponding inputs and outputs described. Example commands include retrieving the latest results for a given Dune query and executing a Dune query to obtain results. The project is MIT licensed.