Data Fulton

What the Data Fulton MCP server does

The Data Fulton MCP server exposes Fulton County Open Data through MCP. Its source is the Socrata catalog at sharefulton.fultoncountyga.gov, so an agent can locate datasets and work with their published records without manually browsing the catalog.

The service is useful when a task depends on a specific Fulton County dataset, such as identifying available data, checking a dataset’s structure, or retrieving rows that match conditions. Dataset discovery returns a resource ID that can then be supplied to the query and metadata tools.

How it works

The Data Fulton MCP server provides three pack-specific tools:

• datasets searches the catalog by keyword and returns each matching dataset’s resource ID, name, description, category, and update date.
• metadata accepts a resource ID and returns schema information, column names and types, row count, category, and last-updated details.
• query runs a Socrata SoQL request against a resource ID and returns matching rows as JSON.

For query, the where, select, group, and order arguments accept SoQL clauses without the leading $. limit and offset support result sizing and pagination. A typical workflow is to search with datasets, inspect the result with metadata, then use query to retrieve the needed records.

Setup and configuration

A hosted MCP connection can use this URL:

https://gateway.pipeworx.io/data-fulton/mcp

The README shows configuration examples for Claude Desktop, Cursor, and Windsurf. The hosted pack endpoint also exposes shared Pipeworx gateway meta-tools in addition to the three Fulton tools. Its exact tool list can change, and the connection’s initialize response is authoritative. The full gateway is available at https://gateway.pipeworx.io/mcp when direct access to all listed packs is preferred.

For a local connection, run the package with npx:

{
  "mcpServers": {
    "data-fulton": {
      "command": "npx",
      "args": ["-y", "@pipeworx/mcp-data-fulton"]
    }
  }
}

The local process communicates over stdin/stdout and exposes only this pack’s tools. The README states that no Pipeworx account is required for the standalone mode or for the first hosted calls.

Tools and capabilities

The Data Fulton MCP server supports catalog search, dataset inspection, and structured row retrieval. It can also be called without an MCP client through the hosted HTTP tool endpoint, including POST https://gateway.pipeworx.io/v1/tools/data_fulton_datasets with a JSON query. This option is useful for scripts or systems that can make HTTP requests but do not host an MCP client.

Limitations and notes

Queries require a Fulton County dataset resource ID, so discovery or prior knowledge of the ID is needed before querying records. The query language is Socrata SoQL, and clause values must omit the $ prefix. The hosted pack endpoint includes shared gateway tools that are not part of the local package; use the local stdio mode if the connection should expose only the three Fulton-specific tools.