Data Howardcounty

What Data Howardcounty does

Data Howardcounty exposes Howard County Open Data through the Model Context Protocol. The underlying catalog is hosted at opendata.howardcountymd.gov and uses the Socrata platform. Agents can discover datasets by keyword, inspect a dataset before querying it, and retrieve matching records as JSON.

The integration is an independent Pipeworx connection and is not affiliated with, endorsed by, or published by Howard County or the upstream catalog provider.

How it works

The normal hosted connection uses this streamable HTTP endpoint:

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

An MCP client can add that URL as a server. Pipeworx also provides a full gateway endpoint at https://gateway.pipeworx.io/mcp; both endpoints reach the same gateway and data sources, while the scoped endpoint lists this pack's tools directly.

A local npm bridge is available as @pipeworx/mcp-data-howardcounty. The client starts it with npx, and the bridge forwards requests to the hosted endpoint. Network access is required because the package has no offline mode. Direct HTTP callers can invoke tools through the /v1/tools/ API, such as data_howardcounty_datasets.

Setup and configuration

For a hosted MCP connection, add the endpoint to the client's MCP configuration:

{
  "mcpServers": {
    "data-howardcounty": {
      "url": "https://gateway.pipeworx.io/data-howardcounty/mcp"
    }
  }
}

For the local bridge, use:

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

Anonymous calls are supported and are limited to 50 calls per day. A free Pipeworx account and PIPEWORX_API_KEY raise the limit to 200 calls per day. The key is passed as a Bearer token. Set PIPEWORX_META_TOOLS=1 for the local bridge if the client should also list Pipeworx-wide tools such as ask_pipeworx; otherwise, it lists the pack's own tools.

Tools and capabilities

• datasets searches the Howard County catalog by keyword. Results include each dataset's resource ID, name, description, category, and update date.
• query runs a Socrata SoQL query against a resource ID. It accepts where, select, group, and order clauses without the leading $, along with limit and offset, and returns rows as JSON.
• metadata returns a resource's schema and catalog details, including columns, data types, row count, category, and last-updated information.

A typical workflow is to search with datasets, pass a returned resource ID to metadata to understand its columns, then use query to select, filter, group, or paginate records.

Limitations and notes

The service depends on network access to Pipeworx and the Howard County catalog. The scoped endpoint also exposes shared Pipeworx meta-tools alongside the pack tools, so an MCP client's tool list may contain more entries than the three Howard County tools. ask_pipeworx can route natural-language questions across the wider Pipeworx catalog. Proxy users may need to enable Node's environment proxy support with NODE_USE_ENV_PROXY=1, depending on their Node version.