tanitdata/agridata-mcp

MCP server for Tunisia's agricultural open data portal (agridata.tn) # agridata-mcp An AI-powered MCP server that turns Tunisia's agricultural open data portal (agridata.tn) into a conversational data source. MCP (Model Context Protocol) lets AI assistants call external tools — connect your AI client and start asking questions in natural language. The portal covers climate, crops, fisheries, water resources, livestock, prices, trade, and more. agridata-mcp handles querying structured DataStore tables and bibliographic catalogs so researchers can access both through conversation. Built for researchers, data scientists, agricultural policy analysts, and journalists covering food security in Tunisia and North Africa. > agridata-mcp is distributed as the `tanitdata` Python package (the installed command and module are named `tanitdata`); the project is published by the tanitdata studio. *Example: agridata-mcp returning a per-delegation rainfed-vs-irrigated cereal yield breakdown for the 2022/2023 campaign in Kairouan, sourced from CRDA Kairouan via catalog.agridata.tn.* --- ## Getting Started agridata-mcp runs locally as a stdio MCP server. Clone the repo, install dependencies, and point your MCP client at the local process. ### Prerequisites - Python 3.11+ - uv package manager ### Install and run ```bash git clone github.com cd agridata-mcp uv sync uv run tanitdata # starts stdio server ``` ### Connect to Claude Desktop Add to your config (`%APPDATA%\Claude\claude_desktop_config.json` on Windows, `~/Library/Application Support/Claude/claude_desktop_config.json` on macOS): ```json { "mcpServers": { "agridata-mcp": { "command": "uv", "args": ["run", "--directory", "/path/to/agridata-mcp", "tanitdata"] } } } ``` ### Test with MCP Inspector ```bash uv run mcp dev src/tanitdata/server.py ``` ### Run unit tests ```bash uv run pytest tests/ ``` --- ## Docker ```bash docker build -t agridata-mcp . docker run -p 8000:8000 agridata-mcp ``` The container runs in `streamable-h …