Skip to main content
GET
List immutable evaluation verdicts

Authorizations

Authorization
string
header
required

A Recursion API key, created in the console under API keys.

Headers

x-tenant-id
string

Optional tenant consistency check. When present, it must name the tenant bound to the API key, including when x-organization-id is default.

Minimum string length: 1
x-organization-id
string

Organization in which to act. Required for a tenant-scoped API key. Optional for an organization-scoped key, where it must name that same organization. The value may be an organization id or default.

Minimum string length: 1

Query Parameters

targetSessionId
string<uuid>

Narrow to one target session. Unknown, deleted-without-history, and cross-organization ids return 404.

targetAgentId
string<uuid>

Narrow to verdicts for one target agent. Retained same-organization history remains readable after deletion.

evaluatorAgentId
string<uuid>

Narrow to verdicts produced by one evaluator agent. Retained same-organization history remains readable after deletion.

runSessionId
string<uuid>

Narrow to one evaluation run, by the runSessionId managedAgentsListEvaluationRunsV2 returns.

limit
integer
default:50

Maximum verdicts to return. Defaults to 50. The schema maximum is the summary view's: view=summary accepts up to 1000, while view=full (the default) accepts up to 100 and answers a larger limit with a 400.

Required range: 1 <= x <= 1000
cursor
string

Opaque signed continuation from nextCursor. It is bound to this organization, filters, and effective limit and expires after one hour.

targetAgentVersionId
string<uuid>

Narrow to verdicts for one immutable target-agent version. Requires evaluatorAgentVersionId. A version neither in the organization's catalog nor judged by that evaluator version returns 404.

evaluatorAgentVersionId
string<uuid>

Narrow to verdicts produced by one immutable evaluator version. A version neither in the organization's catalog nor in its retained verdicts returns 404.

createdAtGte
string<date-time>

Only verdicts recorded at or after this RFC 3339 instant.

createdAtLt
string<date-time>

Only verdicts recorded before this RFC 3339 instant. Must be after createdAtGte when both are supplied.

failedCriterionKey
string

Only verdicts whose rubric criterion with this key failed. Requires evaluatorAgentVersionId.

Maximum string length: 64
Pattern: ^[a-z][a-z0-9]*(?:[-_][a-z0-9]+)*$
view
enum<string>
default:full

full (default) returns every verdict field. summary omits summary and each criterion's rationale and evidenceEventIds — the evaluator's free text, most of a row's bytes — for callers that only aggregate verdicts. It shapes the representation only, so a cursor continues under either view when its limit is one both views accept (100 or less).

Available options:
full,
summary
trigger
enum<string>

Only verdicts of one trigger: outcome, a grading pass of a writer's outcome; manual, an evaluation a caller started; or simulation, the rubric pass of one simulated episode. With evaluations enabled, omission includes all visible origins for a targetSessionId or runSessionId list; an unscoped list includes manual and stamped outcome verdicts but excludes simulation. With evaluations disabled, omission selects outcome verdicts. A historical outcome round whose grader is no longer recorded is listed only in its outcome's evaluations.

Available options:
outcome,
manual,
simulation
outcomeId
string<uuid>

Only the verdicts of one outcome, newest round first. Requires targetSessionId, the outcome's writer session. A historical round whose grader is no longer recorded is listed only in the outcome's evaluations.

Response

OK

items
object[]
required

Newest-first immutable evaluation verdicts for this page.

nextCursor
null | string
required

Signed one-hour continuation, or null.

total
null | integer
required

Always null; the verdict list does not count matching evaluations.

Required range: 0 <= x <= 9007199254740991