CarsXE

CarsXE is a vehicle data API covering VIN decoding, license plate lookups, market values, history reports, recalls, lien and theft records and vehicle imagery. Agents decode a VIN or plate, pull specs, value a vehicle and check recall history in one step.

Verified 21 actions API key available

CarsXE website Platform Documentation Capability data verified 2026-08-27

A trade-in appraisal arrives with a photograph of the VIN plate

Recognize VIN From Image reads the seventeen characters out of the photo

Get Vehicle Specifications decodes the VIN into make, model, year, trim and drivetrain

Get Lien And Theft Records looks for an undischarged lien or a reported theft

Get Market Value prices the vehicle for its state, mileage and condition

The decoded vehicle, the valuation and the lien finding post to the desk that owns the deal

The underwriter confirms the decoded vehicle is the car in the photograph before any figure is committed

What This Integration Enables

CarsXE is for teams whose decisions attach money to one specific car, not to a model line. A dealer taking a trade, a lender funding a title loan, an insurer opening a claim and a fleet manager checking recall exposure all need the same thing: a defensible file on a single vehicle, assembled fast enough to answer a customer standing at a desk. FlowRunner agents build that file. They decode a VIN or a license plate, pull factory specifications, price the vehicle against its state, mileage and condition, read its title and accident history, check NHTSA recalls, and look for liens and theft reports, all keyed to the same identifier.

Two things about this API shape the way a flow should be built, and the connectors layer handles both. CarsXE reports a refused or unresolvable lookup inside the response body rather than with an error status, so the connector reads the body and raises rather than handing a flow an object that looks like a decoded vehicle and is not one. And several options change what a call costs: Require VIN on a Spanish plate bills at twice the normal rate, and Deep Data on a VIN decode is a much slower premium lookup. Those are choices a flow makes deliberately, not defaults it drifts into.

Without FlowRunner

Seventeen characters by hand A VIN gets retyped from a windshield or a photo, and one wrong character quietly describes a different car

Four sites, four tabs Specifications, recall exposure, valuation and lien status each mean a separate lookup somewhere else

Adverse findings arrive late The lien or theft record surfaces after the offer has already gone out

With FlowRunner

The photograph is the input The agent reads the VIN or the plate off an image and decodes it before anyone types

One pass across every check Specifications, recalls, history, valuation and lien status resolve in a single flow against the same VIN

The bad news lands first The lien and theft check runs ahead of the number, so the adverse case reaches a person before the offer does

Use Case Scenarios

An appraisal that starts with a phone camera

A customer photographs the VIN plate on the door jamb and uploads it. Recognize VIN From Image turns the photo into seventeen characters. Get Vehicle Specifications decodes them into the trim, engine and drivetrain that actually drive the number, and Get Market Value prices the car for the state it sits in, the mileage on the odometer and the condition the appraiser recorded. Get Vehicle History adds title events, prior use and reported accidents. The agent writes the assembled file to a deal record in HubSpot and posts the summary to the desk in Slack. Nobody typed a VIN, which is where most VIN errors come from.
Recall exposure across a whole fleet

A fleet operator keeps thousands of VINs in a sheet. On a schedule, the agent hands the list to Submit Recalls Batch, which queues up to ten thousand VINs in one job and returns an identifier rather than results. The flow either polls Get Recalls Batch Status or waits for the webhook CarsXE posts on completion. When the job finishes, Get Recalls Batch Results reads the affected vehicles inline, or Download Recalls Batch Results hands back a file when the result set is large enough that carrying it through the flow would be wasteful. Affected units are appended to a maintenance sheet in Google Sheets with the campaign number, the defect and the remedy attached to each one.
A lien check before the money moves

A title loan application reaches the underwriting queue with a plate rather than a VIN. Decode US License Plate resolves the plate and, with Decode VIN turned on, returns the decoded vehicle in the same call. Get Lien And Theft Records then answers the question the loan actually turns on, because a vehicle carrying an undischarged lien is not the borrower's to pledge. Get Market Value sets the ceiling on what can be advanced against it. The agent assembles all of it, and then stops, because the next step is funding.

Human-in-Loop Highlight

The riskiest moment in this workflow is not the lien hit. It is the fact that the VIN under the lien hit was read off a photograph. Recognize VIN From Image is optical character recognition, and a misread character does not produce an error, it produces a different real vehicle that decodes cleanly and prices cleanly and belongs to somebody else. So before a lien or theft finding becomes a declined application or a reduced advance, the agent stops and posts to the underwriter: "VIN read from the uploaded photo as 1FTFW1E85MFA12345. That decodes to a 2021 Ford F-150 SuperCrew, 3.5L, 4WD. Get Lien And Theft Records reports one undischarged lien. The applicant described a 2021 F-150. Confirm the decoded vehicle matches the photo before I record the adverse finding." The underwriter looks at the image and the decode side by side and answers. This is FlowRunner's human-in-the-loop discipline aimed at the exact place it belongs: the agent did the reading, the decoding, the valuation and the search, and a person owns the moment where a machine-read string becomes an adverse decision about a customer.

Agent processes routinely

Detects was read off a photograph

Clear match Continues automatically

Ambiguous Routes to human via preferred channel

Human decides

Agent resumes with decision

Agent Capabilities

21 actions

VIN Decoding

Get Vehicle Specifications Decodes a VIN into make, model, year, trim, body style, engine, drivetrain, transmission, fuel type and factory equipment. This is the main decoder and the usual starting point. Deep Data returns considerably more attributes at the cost of a much slower response, so it stays off for anything interactive.
Decode International VIN Decodes a VIN registered outside North America. Use it when the standard decoder returns a partial result for a European, Asian or Australian vehicle, or call it directly when the market is already known.
Recognize VIN From Image Reads a VIN out of a photograph of a VIN plate, door jamb sticker or windshield. Takes either an image URL or the image as a base64 string. This is what lets a customer photograph a car instead of typing seventeen characters.

License Plates

Decode License Plate Resolves a license plate to its vehicle using the current v2 decoder, which covers the widest set of countries. Country is optional for the United States. Require VIN is supported only for Spanish plates and bills at twice the normal rate, so a flow turns it on deliberately.
Decode License Plate V1 Resolves a plate through CarsXE's earlier decoder, which requires a state and answers a different attribute shape. Kept for existing v1 based integrations.
Decode US License Plate Resolves a United States plate across the fifty states, the District of Columbia and Puerto Rico. Turn on Decode VIN to have the resolved VIN decoded in the same call and returned nested, which saves a second request.
Recognize Plate From Image Reads a plate number out of a photograph of a vehicle. Built for upload flows, and for parking and access control where a camera frame is the only input. Feed the result into Decode License Plate to reach the vehicle.

Recalls

Get Vehicle Recalls Returns the NHTSA safety recalls affecting one vehicle by VIN, with the campaign number, the defect, the remedy and the reporting date. The check to run before selling, insuring or lending against a car.
Get Recalls By Year Make Model Returns the recalls affecting a model line rather than an individual vehicle. Used to assess a whole model, or when no VIN is available.
Submit Recalls Batch Queues up to ten thousand VINs for recall checking in one job and returns a batch identifier. Nothing else comes back from this call, so the flow either polls or supplies a webhook URL. VINs can be sent directly, as inline CSV, or as a link to a CSV file.
Get Recalls Batch Status Returns the progress of a queued batch: still running, how many VINs are done, or failed. The operation to poll when no webhook was supplied.
Get Recalls Batch Results Returns a finished batch's recall results as JSON, one entry per VIN.
Download Recalls Batch Results Returns a download link for a finished batch instead of inline JSON. The right choice when carrying every result through the flow would be wasteful.

Valuation and History

Get Market Value Returns the estimated market value of a vehicle by VIN, adjusted for state, mileage and condition. Those adjustments are what make the number usable for a trade or for inventory pricing rather than a generic book figure.
Get Market Value V1 Returns a valuation through the earlier route, which accepts no mileage or condition adjustment. Kept for existing v1 based integrations.
Get Vehicle History Returns the recorded history of a vehicle by VIN: title events, reported accidents, odometer readings, registration changes and prior use. The due diligence step before a purchase, a loan or an insurance decision.
Get Lien And Theft Records Returns outstanding lien records and reported theft records against a vehicle. This is the check that decides whether the seller can actually convey the car, and it is the finding this page's human gate exists for.

Vehicle Catalog

Get Vehicle Images Returns photographs of a vehicle by make and model, optionally narrowed by year, trim, color, angle, interior or exterior, size and license type. Filter by license when the images will be published, because the license values separate what may be displayed from what may be modified and used commercially.
Get Year Make Model Returns the specifications of a vehicle identified by year, make, model and optionally trim, with no VIN required. All Trim Options returns every trim for that combination rather than a single match, which is what a configurator needs.
Get Year Make Model Options Returns the valid values for one level of the vehicle catalog, which is how cascading year, make, model and trim pickers get built. Supply a Dimension and the filters that level requires. Leaving Dimension empty lets CarsXE infer the level, which is convenient by hand and unpredictable in a flow.
Decode OBD Code Translates an on board diagnostics trouble code into its meaning, the systems it affects and the likely causes. Turns a raw code from a scanner or a telematics feed into something a service advisor can act on.

Frequently Asked Questions

What can FlowRunner do with CarsXE?

FlowRunner agents can run Get Vehicle Specifications, Decode International VIN, and Recognize VIN From Image in CarsXE, plus 18 more actions.

Does connecting CarsXE to FlowRunner require OAuth?

No. CarsXE connects to FlowRunner with an API key, no OAuth flow required.

Can CarsXE trigger a FlowRunner workflow automatically?

CarsXE doesn't currently expose triggers in FlowRunner. It connects as an action step inside workflows started by another trigger.

Start building with CarsXE

Free plan, no card required. Connect in minutes.

Start Building Free Talk to us