The full upstream README, mirrored here for reference. Install config, tool schemas, adoption signals, and an original overview live on the PlateQuery MCP — US License Plate & Vehicle Information listing page.
License plate lookup for AI agents, built on PlateQuery — the license plate lookup site people use to notify vehicle owners, report bad driving, or flag illegal parking. This Model Context Protocol server lets an agent read those same messages for any U.S. license plate, plus public ALPR (automated license plate reader) sighting history from the Flock Safety camera network, by state or region.
https://api.platequery.com/mcp (Streamable HTTP)com.platequery/plate-api on the
official MCP registryThis repository holds public documentation for the server. The implementation lives in PlateQuery's main application repository, which is closed-source.
| Tool | Cost | Description |
|---|---|---|
check_credits | Free | Check your account's remaining credit balance. Doesn't count against the rate limit. |
lookup_plate | 1 credit | Quick summary for one plate: privacy status, the latest 3 messages (plus a total count), and a count of ALPR mentions, if any are public. |
list_plate_messages | 1 credit | Full, paginated list of messages posted about a plate. |
list_plate_alpr_mentions | 1 credit | Full, paginated ALPR search history for a plate, including the searching agency and source. |
post_plate_message | Free | Post a new message about a plate -- e.g. "reported for speeding" -- optionally with vehicle make/model/year/color and an approximate location. Requires the write scope. See Posting messages below. |
A plate marked private by its owner returns zeroed messages/mentions to
everyone except that owner. Free plan grants a starting credit balance on
signup; additional credits are purchased in bulk packages — see
api-settings for current pricing.
Paid tools are rate-limited per account; call check_credits as often as
you like to check balance without spending a credit.
post_plate_message is free but deliberately conservative, since it's
the only tool that writes anything:
vehicle_make/vehicle_model/vehicle_year/vehicle_color and
location are all optional free text -- location should be an
approximate description (e.g. "Hwy 37 near Los Altos, CA" or a
cross-street), never precise GPS coordinates.The server is OAuth 2.1 only — no manual API keys. It supports PKCE and dynamic client registration (DCR), so a compliant MCP client (Claude, or any client following the MCP authorization spec) can connect by:
https://api.platequery.com/mcp.WWW-Authenticate challenge to discover
https://platequery.com as the authorization server (RFC 9728
protected resource metadata).No client secret or pre-issued API key is needed — sign-in happens in the browser the first time an agent connects.
Two scopes exist: read (the default -- covers every tool except
post_plate_message) and write (required for post_plate_message,
opt-in only). A client must explicitly request write at connect time to
be able to post; an existing read-only connection can't post without
reconnecting.
Most MCP clients (including Claude) let you add a remote server by URL.
Add https://api.platequery.com/mcp as a Streamable HTTP connector and
follow the sign-in prompt on first use.
Questions or issues: https://platequery.com/contact/