Datascribe

Datascribe is a data documentation framework to strengthen accuracy in LLM-generated analytics queries. This mcp-server provides a set of tools to instruct an LLM application on how to use the Datascribe framework to produce Datascribe documentation.

Read the Datascribe famework overview

The MCP Servier helps any MCP-compatible LLM application generate rich data documentation that can support text-to-sql using the datascribe documentation framework. reason about real-world data by providing richly annotated metadata — including schema, provenance, business context, and known limitations.

Setup

Configure Claude desktop

Use Claude desktop to locate the claude_desktop_config.json file by navigating to Settings > Developer > Edit Config. Add the following configuration. (Note that datascribe requires the Filesystem MCP Server):

{
    "mcpServers": {
        "filesystem": {
            "command": "npx",
            "args": [
            "-y",
            "@modelcontextprotocol/server-filesystem",
            "/Users/username/Desktop",
            "/path/to/other/allowed/dir"
            ]
        },
        "datascribe": {
            "command": "uv",
            "args": [
                "--directory",
                "/your/path/to/datascribe/repo",
                "run",
                "datascribe/datascribe.py"
            ]
        }
    }
}

Restart Claude desktop. On the new chat screen, you should see a hammer (MCP) icon appear with the new MCP tools available.

Usage

Once configured, your AI tool will automatically make use of the MCP to interact with the datascribe tools. You may be prompted for permission before a tool can be used. Start by simply messaging "#datascribe start docs".

Tools

The datascribe MCP Server provides the following tools for AI assistants to use:

• datascribe_install: prompt your bot to begin working on datascribe docs
• datascribe_data_querying_instructions: querying guidelines for the bot
• run_bigquery_sql: execute any sql against the bigquery database designated in the docs

Limitations

• Only works currently on BigQuery databases, with service account-based authentication.

Contributing

I welcome your collaboration in improving the developer MCP experience. Please submit issues in the GitHub issue tracker.