Skip to main content
POST
Run a run again

Authorizations

Authorization
string
header
required

API key authorization. Pass your API key in the Authorization header as a Bearer token. Both new (gsk_*) and legacy (gsk-) API keys are accepted, e.g. Authorization: Bearer gsk_... or Authorization: Bearer gsk-....

Path Parameters

id
string
required

Run ID

Example:

"run_123"

Body

application/json
versionId
string

Run the repetition against a different version. Only for a run the platform launched itself; a run that scored existing sessions refuses it, because those sessions already name their version.

Minimum string length: 1
Example:

"version_123"

useCurrentTestCases
boolean

Resolve the original request again instead of repeating the exact rows that ran, so test cases added since are included. On a run that scored existing sessions it also drops the metrics the source recorded, so the repetition scores with each specification's links as they stand today, which can change the metric set and the credit cost. Defaults to false, which keeps the repetition comparable with its source.

Example:

false

Response

Run launched again successfully

id
string
required
Example:

"run_123"

productId
string
required
Example:

"product_123"

versionId
string | null
required
Example:

"version_123"

userId
string | null
required
Example:

"user_123"

ordinal
integer
required

Per-product launch number, starting at 1 and never reused

Example:

42

customId
string | null
required

Caller-supplied label

Example:

"nightly-regression"

relaunchedFromRunId
string | null
required

The run this one repeats, when POST /runs/{id}/relaunch minted it

Example:

"run_122"

openedBy
enum<string>
required

LAUNCH when a launch endpoint opened the run, which then closes when that launch ends. CALLER when POST /runs opened it, which only POST /runs/{id}/close ends.

Available options:
LAUNCH,
CALLER
Example:

"LAUNCH"

status
enum<string>
required
Available options:
RUNNING,
COMPLETED,
FAILED,
CANCELLED
Example:

"COMPLETED"

evaluationCount
integer
required

Evaluations this run launched

Example:

120

sessionCount
integer
required

Distinct sessions covered by the evaluations of this run

Example:

30

kind
enum<string> | null
required

The shape of work the platform performed. Null on a run the caller opened, and on one several scoring calls filled.

Available options:
GENERATE_INFERENCES,
EVALUATE_SESSION,
EVALUATE_SESSION_BATCH,
REPLAY_METRIC,
null
Example:

"GENERATE_INFERENCES"

jobId
string | null
required

Queue job the run dispatched, when it queued one. Null on a run the caller opened, and on one several scoring calls filled.

error
string | null
required

Why the run failed, when it did

request
object | null
required

The launch request as the platform received it. Null on a run the caller opened, and on one several scoring calls filled.

resolvedInputs
object | null
required

What the request resolved to at dispatch: test cases, sessions and metrics. Keys depend on kind. Null on a run the caller opened, and on one several scoring calls filled.

finishedAt
string<date-time> | null
required
deletedAt
string<date-time> | null
required
report
object | null
required

What the scoring did, for a repetition that scores sessions inside the request. Null when the repetition drives the agent again, whose work is queued and reports on the new run itself instead.

createdAt
string<date-time>