Why does the server run locally rather than on a hosted endpoint?

agridata MCP follows the standard MCP architecture: the server runs as a local process called by the LLM client over stdio. This is how most MCP servers in the ecosystem operate. It removes operational dependencies, simplifies authentication (none is needed — the data is open), and gives users full control over the environment. The trade-off is a one-time local install.

Which MCP clients are supported?

All clients implementing the standard MCP protocol: Claude Desktop, Claude Code, VS Code with an MCP extension, Cursor, Windsurf, Gemini CLI, and others. The configuration format differs slightly between clients; the launch command is the same.

What is uv and why use it?

uv is a fast Python package manager from Astral. It resolves and installs dependencies in seconds and handles virtual environments transparently. The agridata server uses uv as its declared runtime; using pip is possible but no longer the recommended path.

Is the data freely reusable?

Yes. The data comes from the public agridata.tn portal and is published under the Licence Nationale de Données Publiques Ouvertes (LNDPO). Any reuse consistent with the LNDPO is permitted; attribution to the portal is required.

Where do I report a bug or request a dataset?

Open an issue on https://github.com/tanitdata/agridata-mcp or write to contact@tanitdata.org.

Guide

Get started with agridata MCP in 15 minutes

Clone the open-source agridata MCP server, install it locally, and connect Claude Desktop or another LLM client to run your first real query on Tunisia's open agricultural data.

Expected outcome

By the end of this guide, agridata MCP will be running on your machine, your usual LLM client will be configured to call it, and you will have run a first query that combines structured data and documentary references. Estimated total time: 15 minutes if Python and uv are already installed; allow ten extra minutes otherwise.

Prerequisites

An MCP-compatible LLM client: Claude Desktop, Claude Code, VS Code, Cursor, Windsurf, Gemini CLI, or similar.
Python 3.11 or later. Download from python.org if not already installed.
The uv package manager. Install with curl -LsSf https://astral.sh/uv/install.sh | sh on macOS / Linux, or powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex" on Windows. Full uv install docs.
An internet connection to clone the repository and reach catalog.agridata.tn at query time.

Step 1 — Clone and install

Pick a directory where you want the server to live and run:

bash

git clone https://github.com/tanitdata/agridata-mcp.git
cd agridata-mcp
uv sync

uv sync resolves and installs every dependency declared in pyproject.toml. On a fresh checkout this usually takes ten to twenty seconds.

Step 2 — Note the absolute path

The MCP client needs the absolute path to your clone directory. Run pwd from inside agridata-mcp (or cd + echo %CD% on Windows) and copy the full path. You will paste it in the configuration snippets below in place of /absolute/path/to/agridata-mcp.

Step 3 — Configure your LLM client

Pick the snippet for your client. The launch command is the same everywhere; only the file format changes.

Claude Desktop

Configuration file:

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

json

{
  "mcpServers": {
    "tanitdata": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/agridata-mcp", "tanitdata"]
    }
  }
}

Save the file, then restart Claude Desktop.

Cursor, Windsurf, other stdio-MCP clients

Most clients accept the same shape as Claude Desktop. Add the snippet to the client's MCP settings file (consult the client's own docs for the exact path):

json

{
  "mcpServers": {
    "tanitdata": {
      "command": "uv",
      "args": ["run", "--directory", "/absolute/path/to/agridata-mcp", "tanitdata"]
    }
  }
}

Claude Code (CLI)

A single command, no configuration file:

bash

claude mcp add tanitdata uv run --directory /absolute/path/to/agridata-mcp tanitdata

Step 4 — First test call

In your client, start a new conversation and ask:

"What tools are exposed by the tanitdata server?"

The client should display a list of tools provided by the server. If that list appears, the connection works.

Step 5 — Your first real query

Try a question that combines numbers and documentary context:

"What is the production of durum wheat by governorate over the past five years, and how does it compare to the objectives of the national plan?"

The server responds with the matching data and, for each result, points to the source resource on agridata.tn. You can follow the link to verify the source directly on the portal.

Troubleshooting

The client reports the server cannot be found. Verify that the path in the configuration is absolute (not relative) and points to the directory that contains pyproject.toml. A common mistake on Windows is using forward slashes only — both forward and backslashes work in JSON if escaped.
uv: command not found in the client logs. The client launches uv as a subprocess and inherits a limited environment that may not include uv's install path. Use the absolute path to the uv binary in the configuration (find it with which uv on macOS / Linux or where uv on Windows).
You want to verify the server runs at all before wiring it to a client. From the clone directory, run:
bash
```
uv run mcp dev src/tanitdata/server.py
```
The MCP Inspector opens in your browser and lets you call tools without an LLM client.

What's next

Once the server is running on the client side, browse the research index for more advanced use cases. To share feedback, file a bug, or propose a collaboration, the GitHub repository and the Get Involved page are the entry points.

Frequently asked questions

Why does the server run locally rather than on a hosted endpoint?: agridata MCP follows the standard MCP architecture: the server runs as a local process called by the LLM client over stdio. This is how most MCP servers in the ecosystem operate. It removes operational dependencies, simplifies authentication (none is needed — the data is open), and gives users full control over the environment. The trade-off is a one-time local install.
Which MCP clients are supported?: All clients implementing the standard MCP protocol: Claude Desktop, Claude Code, VS Code with an MCP extension, Cursor, Windsurf, Gemini CLI, and others. The configuration format differs slightly between clients; the launch command is the same.
What is uv and why use it?: uv is a fast Python package manager from Astral. It resolves and installs dependencies in seconds and handles virtual environments transparently. The agridata server uses uv as its declared runtime; using pip is possible but no longer the recommended path.
Is the data freely reusable?: Yes. The data comes from the public agridata.tn portal and is published under the Licence Nationale de Données Publiques Ouvertes (LNDPO). Any reuse consistent with the LNDPO is permitted; attribution to the portal is required.
Where do I report a bug or request a dataset?: Open an issue on https://github.com/tanitdata/agridata-mcp or write to contact@tanitdata.org.