curl --request PATCH \
--url https://api.pandaprobe.com/evaluations/trace-scores/{score_id} \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"value": "<string>",
"reason": "<string>",
"metadata": {}
}
'{
"id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"trace_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"name": "<string>",
"value": "<string>",
"status": "SUCCESS",
"source": "AUTOMATED",
"created_at": "<string>",
"project_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"data_type": "NUMERIC",
"eval_run_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"author_user_id": "<string>",
"reason": "<string>",
"environment": "<string>",
"config_id": "<string>",
"metadata": {},
"updated_at": "<string>"
}Manually edit a trace score.
Only value, reason, and metadata can be changed by the
caller. status is automatically set to SUCCESS, source is
changed to ANNOTATION, and updated_at is set to now.
Auth: Bearer + X-Project-ID | X-API-Key + X-Project-Name
curl --request PATCH \
--url https://api.pandaprobe.com/evaluations/trace-scores/{score_id} \
--header 'Authorization: Bearer <token>' \
--header 'Content-Type: application/json' \
--data '
{
"value": "<string>",
"reason": "<string>",
"metadata": {}
}
'{
"id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"trace_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"name": "<string>",
"value": "<string>",
"status": "SUCCESS",
"source": "AUTOMATED",
"created_at": "<string>",
"project_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"data_type": "NUMERIC",
"eval_run_id": "3c90c3cc-0d44-4b50-8888-8dd25736052a",
"author_user_id": "<string>",
"reason": "<string>",
"environment": "<string>",
"config_id": "<string>",
"metadata": {},
"updated_at": "<string>"
}Bearer authentication header of the form Bearer <token>, where <token> is your auth token.
Editable fields for a trace score.
All fields are optional -- only provided fields are updated.
status is set to SUCCESS automatically on save, and
source is changed to ANNOTATION to indicate human edit.
updated_at is also set automatically.
Successful Response
Full trace score representation used by both list and detail endpoints.
Outcome of a trace score evaluation attempt.
SUCCESS, FAILED, PENDING Who produced the score judgment (not how it arrived).
AUTOMATED -- the eval system's LLM judge produced this score ANNOTATION -- a human assigned this score manually PROGRAMMATIC -- external code submitted this score via API/SDK
AUTOMATED, ANNOTATION, PROGRAMMATIC Data type for a trace score value.
NUMERIC, BOOLEAN, CATEGORICAL