Pa Dmv MCP server

@pipeworx/pa-dmv

Pennsylvania DMV MCP — registered vehicles and electric-vehicle adoption by county and by ZIP code, from PennDOT Driver & Vehicle Services.

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

Tools

hybrid, fuel-cell and conventional hybrid registrations for each of Pennsylvania's 67 counties in one quarter, with plug-in share and a genuine statewide total. Answers "how many EVs are registered in Pennsylvania", "EV share in Allegheny County", "which Pennsylvania county has the most electric vehicles".

• padmvev_adoption(county?, year?, quarter?, limit?) — battery-electric, plug-in

fleet of every fuel type by county, with a statewide total and each county's share. Rides on the same layer, which carries TOTAL_REG alongside the EV columns.

• padmvvehicle_registrations(county?, year?, quarter?, limit?) — the whole registered

breakdown for ~1,830 Pennsylvania ZIP codes. Answers "how many EVs are registered in ZIP 19103", "which Pennsylvania ZIP code has the most electric vehicles".

• padmvevadoptionby_zip(zip?, year?, quarter?, limit?) — the same drivetrain

Auth

Keyless.

Data sources

— 67 counties × registration years 2023–2026. Attributes: COUNTYNAME, COUNTYNUMBER, FIPSCOUNTYCODE, DISTRICTNO, PLANNINGPARTNER, REGISTRATIONYEAR, then per-quarter HEVQ, PHEVQ, BEVQ, FUELCELLQ, TOTALEVQ, TOTALREGQ, PCTEVQ*. — 7,328 records total, ~1,832 per registration year. Same quarter columns keyed on ZIP.

Both layers have maxRecordCount 2000, so a full year fits in a single request and no paging is needed. Latest published quarter as of 2026-07-29 is 2026 Q2.

Gotcha: quarters are COLUMNS, not rows

Each feature is one county (or ZIP) for one REGISTRATION_YEAR, carrying all four quarters side by side as separate column groups. Quarters PennDOT has yet to file are present as null columns. There is no row to filter on and no date field to sort by, so "the latest data" means walking Q4 → Q1 looking for a quarter that actually carries values.

pickQuarter() does that walk once for the whole result set, not per row. Picking a quarter per row looks harmless on a single-county lookup and is a correctness bug on a ranking: you end up comparing one county's Q2 against another's Q1 and summing the mix into a "statewide total". A quarter is accepted only if at least half as many rows carry it as carry the best-covered quarter, so one early-reporting straggler cannot drag the whole answer forward.

Gotcha: hybrids sit outside TOTAL_EV

PennDOT's TOTALEV = BEV + PHEV + FUELCELL. Conventional hybrids (HEV) are reported separately and are deliberately excluded — a Prius appears in hybridgasoline and never in pluginvehicles. Verified across all 67 counties × both 2026 quarters: zero mismatches between TOTALEV and BEV + PHEV + FUEL_CELL.

Gotcha: the ZIP layer's 2026 Q1 columns are wrong upstream

In the ZIP layer only, and in the 2026 Q1 column group only, PennDOT's own derived columns are broken two ways:

against BEVQ1 314 + PHEVQ1 136 + HEV_Q1 841 = 1291.

• TOTALEVQ1 folds conventional hybrids in. ZIP 19103 publishes TOTALEVQ1 1291

ZIP 19103 publishes 0.14 where the plug-in share is 4.85%.

• PCTEVQ1 is written as a fraction of that inflated number rather than a percent.

Shortened. The full README is on GitHub.