Skip to main content
POST
Compare two documents

Authorizations

Authorization
string
header
required

Workspace credential issued from the automation console at /automation. Send it as Authorization: Bearer vq_ws_.... This is NOT a Data API key: a vq_key_ credential is refused here and names the other product in the error.

Headers

Idempotency-Key
string | null

A unique value of your choosing, so a retried launch returns the FIRST operation instead of starting a second billable job. Reusing one with a different body is refused. Retrying with the same key and the same body is free and is the intended way to recover from a timeout.

Path Parameters

matterId
string
required

mat_ identifier of the matter to work inside. Everything in this API hangs off a matter, and the matter in the path is what the authorization boundary is checked against. Take it from GET /v1/matters.

Body

application/json

Start a comparison between two documents already in the matter.

Both sides are matter documents. The internal API also accepts drafts, draft versions and loose uploads; publishing those would mean publishing three more identifier spaces, and a customer that wants to compare arbitrary bytes uploads them to the matter first, which is one call it was going to make anyway.

There is no matterId field. The matter is the path, which is what makes the authorization boundary visible in the URL.

title
string
required

Display title for the comparison. Trimmed; whitespace alone is refused.

Required string length: 1 - 300
Example:

"MSA v2 against v3"

originalDocumentId
string
required

doc_ identifier of the ORIGINAL document, the one being compared against. Must be in this matter.

Example:

"doc_9f2c8b1e4a7d43c9b6e0f1a2c3d4e5f6"

revisedDocumentId
string
required

doc_ identifier of the REVISED document. Must be in this matter, and must differ from originalDocumentId.

Example:

"doc_9f2c8b1e4a7d43c9b6e0f1a2c3d4e5f6"

settings
ComparisonSettings · object

How the diff engine should treat the two inputs. Omit for sensible defaults.

Response

Successful Response

One long-running operation, whatever kind of work it is.

Readable for OPERATION_RETENTION_DAYS after creation, per AIP-151.

id
string
required

Public identifier, op_ followed by 32 hex characters. Poll GET /v1/operations/{operationId} with it.

Example:

"op_9f2c8b1e4a7d43c9b6e0f1a2c3d4e5f6"

type
string
required

What kind of work this is, for example matrix.run or draft.generate.

Example:

"matrix.run"

status
enum<string>
required

One of five values: queued, running, succeeded, failed, cancelled. There is no sixth and there are no synonyms. Stop polling once it is succeeded, failed or cancelled.

Available options:
queued,
running,
succeeded,
failed,
cancelled
Example:

"succeeded"

createdAt
string<date-time>
required

When the operation was accepted (RFC 3339).

Example:

"2026-08-19T14:32:10Z"

completedAt
string<date-time> | null

When the operation reached a terminal status (RFC 3339). Present if and only if the status is terminal.

Example:

"2026-08-19T14:32:10Z"

matterId
string | null

mat_ identifier of the matter this work belongs to, when it belongs to one.

Example:

"mat_9f2c8b1e4a7d43c9b6e0f1a2c3d4e5f6"

resource
OperationResource · object | null

What the operation produced. Absent until the underlying job row exists, which an idempotent replay can briefly observe, so treat absence as 'not yet' rather than 'never'.

error
OperationError · object | null

Why the work failed. Present only when status is failed. Partial success is succeeded with progress.done < progress.total, never an error.

progress
OperationProgress · object | null

How far along the work is, when the underlying job reports it. Absent does not mean no progress.

requestId
string | null

The X-Request-ID of the request that created this operation. Quote it in a support ticket.

Example:

"req_5f2c8b1e4a7d43c9b6e0f1a2c3d4e5f6"

Last modified on August 23, 2026