Create Evaluator Result
Path parameters
Request
Target span id. Not validated against existing spans (loose target policy).
Session id the target span belongs to. Denormalized so session-scoped reads stay fast.
Evaluator / metric identity (e.g. ‘faithfulness/v1’).
Discriminator for which of value / string_value carries the payload.
Numeric value. Required when data_type is NUMERIC or BOOLEAN (0|1).
String value. Required when data_type is CATEGORICAL or TEXT.
SCORED when the evaluator produced a value; FAILED when it ran but could not score this target (no value or string_value; put the reason in comment). Rollups count a FAILED result as a failed attempt instead of inferring failure from a missing row.
Free-text rationale or explanation.
Response
Whether the evaluator produced a value for its target.
A FAILED result carries no value: the evaluator ran and could not score (judge unreachable, unparseable output, errored trial). Rollups count it as a failed attempt rather than inferring failure from the absence of a row.