API ReferenceScores

Create or update a log score

Create a score for a specific log/span. The backend keeps one score per (log, evaluator, scorer) and updates the existing score if the same combination is submitted again.

Path parameters

log_idstringRequired

Log/span unique ID to manage scores for.

Headers

AuthorizationstringRequired

Bearer token. Use Bearer YOUR_API_KEY for API key auth or Bearer <JWT> for dashboard auth.

Request

Use the value field matching the evaluator `score_value_type`: `numerical`/`percentage` -> `numerical_value`, `boolean` -> `boolean_value`, `single_select`/`multi_select`/`categorical` -> `categorical_value`, `text`/`comment` -> `string_value`, `json` -> `json_value`. The log ID comes from the path. The backend derives `scorer` from the authenticated user for this route.
evaluator_idstringOptional

Evaluator ID. Provide either evaluator_id or evaluator_slug.

evaluator_slugstringOptional

Custom evaluator slug. Provide either evaluator_id or evaluator_slug.

timestampdatetimeOptional
Log timestamp. Supplying it can avoid an additional log lookup.
environmentstringOptional
Score environment.
prompt_idstringOptional
prompt_version_numberintegerOptional
dataset_idstringOptional
automation_idstringOptional
scorerstringOptional

Optional score producer for general score creation. Log-scoped routes derive this from the authenticated user.

numerical_valuedouble or nullOptional

Numeric score value. Use for numerical and percentage evaluators.

string_valuestring or nullOptional

Text score value. Use for text and legacy comment evaluators.

boolean_valueboolean or nullOptional

Boolean score value. Use for boolean evaluators.

categorical_valuelist of strings or nullOptional

Categorical score values. Use for single_select, multi_select, and legacy categorical evaluators.

json_valuestring or nullOptional

JSON score value encoded as a string. Use for json evaluators.

Response

Updated existing score for this log/evaluator/scorer.

idstring
Score ID.
created_atdatetime
When the score was created.
typeenum
Evaluator execution type. Legacy rows may contain legacy type values.
environmentstring
Environment associated with the score.
numerical_valuedouble or null

Numeric score value. Use for numerical and percentage evaluators.

string_valuestring or null

Text score value. Use for text and legacy comment evaluators.

boolean_valueboolean or null

Boolean score value. Use for boolean evaluators.

categorical_valuelist of strings or null

Categorical score values. Use for single_select, multi_select, and legacy categorical evaluators.

json_valuestring or null

JSON score value encoded as a string. Use for json evaluators.

is_passedboolean or null
Whether the score passes the evaluator conditions.
costdouble or null
Evaluation cost in USD, when available.
evaluator_idstring or null
Evaluator ID associated with the score.
evaluator_slugstring or null
Evaluator slug associated with the score.
evaluator_namestring or null

Human-readable evaluator name, when available.

scorerstring or null

Who or what produced the score. For log-scoped score writes, the backend derives this from the authenticated user.

log_idstring or null

Associated log/span ID.

prompt_idstring or null
Associated prompt ID, when available.
prompt_version_numberinteger or null
Associated prompt version number, when available.
dataset_idstring or null
Associated dataset ID, when available.
automation_idstring or null
Associated automation ID, when available.
statusstring or null
Evaluation status, when available.
error_messagestring or null
Evaluation error message, when available.

Errors

400
Bad Request Error
401
Unauthorized Error
404
Not Found Error