How andrewwgordon/mcp-chinookdb-server works
Project type and purpose
• This is a small Python-based client/server tool set—specifically an MCP (Model Context Protocol) server exposing a local SQLite “Chinook” sample database plus a sample “Agno” agent client that drives it using an LLM.
• Purpose: Allow LLMs or interactive CLI clients to safely explore the Chinook database via standardized MCP resource endpoints and SQL-query tools.
File structure
• Top-level scripts:
• chinook_mcp_server.py – the MCP server implementation
• agno_test_client.py – a CLI agent client demonstrating LLM-driven queries
• Project metadata: pyproject.toml, uv.lock, .python-version – typical Python packaging and dependency files
• README.md – detailed overview, usage, architecture hints
• No separate “src” or “frontend/backend” directories; the architecture is captured by the two main processes.
Additional README insights
• Describes the MCP resource endpoints (schema://… and run_sql_query tool)
• Explains automatic database download/extraction, safe escaping, read-only SELECT enforcement
• Notes Agno agent’s use of OpenAI GPT-4 (or other LLMs) and REPL interface
• States STDIO transport by default