Dc Dmv MCP Server

@pipeworx/dc-dmv

District of Columbia DMV data: service centers, the record-level vehicle inspection file, and traffic convictions. Keyless.

Part of Pipeworx — an MCP gateway connecting AI agents to 1476+ live data sources.

Why DC gets its own pack

DC's grain is unlike any state's. The District publishes 766,879 individual vehicle inspections keyed by VIN and 330,728 individual traffic convictions keyed by citation, both as ArcGIS tables on the DCGIS server — not the county-level aggregates most states publish. So the tools take VINs, violation text and date ranges rather than a county argument.

Tools

Auth

None. Every upstream is a public DCGIS ArcGIS MapServer layer.

The State Plane coordinate trap

The office layer's XCOORD/YCOORD columns are Maryland State Plane metres, not lat/lng. Georgetown Service Center reads 394380.55, 137456.23 — plausible-looking numbers that are wrong by the width of a hemisphere if treated as coordinates.

Every query here passes geometry: true to arcgisQuery (which sets outSR=4326) and reads latitude and longitude off the returned geometry. XCOORD/YCOORD are deliberately left out of the response so they cannot be mistaken for coordinates downstream. Georgetown's real position is 38.9050, -77.0648.

Gotchas worth knowing

These are live-verified behaviours, not guesses:

• The inspection file's STATUS column is empty on all 766,879 rows, so a pass/fail outcome is not published. What you get is that an inspection happened, when, and when the resulting certificate expires — an expiration roughly two years out is the usual sign of a pass, a short one of a re-inspection window. The tool's note says so rather than letting a caller read a null as "unknown result". For per-test pass/fail with make and model, nj_mvc_vehicle_inspections publishes it.
• GIS_ID, CREATOR and CREATED are also empty on the inspection table.
• About half the conviction rows prefix the violation text with a four-digit internal code — 0201SPEED IN EXCESS OF LIMIT and SPEED IN EXCESS OF LIMIT are the same offence in two buckets. A raw group-by therefore splits and mis-ranks every violation. The tool strips the prefix and merges: "DISOBEYING OFFICIAL SIGN OR SIGNAL DEVICE" is 20,113 convictions, not the 10,369 a naive query reports.
• DCGIS caps any single response at 2,000 rows, grouped statistics included. Month and year rollups group by day upstream (1,874 distinct days across the whole file — inside the cap) and aggregate in code; if that ever hits the cap the response flags truncated and says to narrow with inspected_since.
• Coverage: inspections run 2020-11-01 → 2026-06-30 and are still being appended; convictions run 2002-07-01 → 2026-06-27. Both are returned in a coverage field rather than assumed.
• Convictions are convictions. A citation that was dismissed or never adjudicated is absent, so these counts are not ticket counts.
• The office layer carries no ZIP and no hours. MATCHADDR is the District's own geocoded street address; dmv.dc.gov posts hours per location.

Available but not wrapped

The same Transportation_DMV MapServer carries further DMV tables that are queryable but not exposed as tools:

Layers 149 and 150 are the historical inspection archives and share the current table's schema, including the same empty STATUS.

Data sources

• DCGIS Public_Service layer 4 — District agency locations, filtered to the 9 DMV rows
• DCGIS Transportation_DMV layer 1 — DMV Vehicle Inspections
• DCGIS Transportation_DMV layer 151 — DMV Convictions
• Dataset discovery: opendata.dc.gov

Quick Start

Add to your MCP client (Claude Desktop, Cursor, Windsurf, etc.):

{
  "mcpServers": {
    "dc-dmv": {
      "url": "https://gateway.pipeworx.io/dc-dmv/mcp"
    }
  }
}

What this endpoint actually serves

tools/list at https://gateway.pipeworx.io/dc-dmv/mcp returns the tools in the table above plus the shared Pipeworx meta-tools — ask_pipeworx, discover_tools, search_within, remember/recall and the rest of the gateway-wide set. So the tool count you see is larger than this table: a single-pack endpoint currently lists roughly 30 shared tools alongside the pack's own. The connection's initialize response states its exact scope, and is the authoritative answer for a given day.

This is deliberate, not multiplexing by accident. The meta-tools are what let a scoped connection answer a question this pack does not cover — via ask_pipeworx, which routes across the whole catalog — without you adding a second MCP server. There is currently no way to mount a pack endpoint without them; if the extra schemas cost you more context than the routing is worth, connect to the full gateway once rather than to several pack endpoints.

Or connect to the full Pipeworx gateway to get every pack's tools listed directly, instead of just this one's:

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

Both URLs reach the same gateway and the same 1476+ data sources. The only difference is which pack's tools are listed directly; ask_pipeworx reaches all of them from either one.

Using with ask_pipeworx

Instead of calling tools directly, you can ask questions in plain English — this works on the pack endpoint above as well as on the full gateway:

ask_pipeworx({ question: "your question about Dc Dmv data" })

The gateway picks the right tool and fills the arguments automatically.

More

• Docs and guides
• pipeworx.io

License

MIT