List trace scores produced by an observer.

GET

/v1/observers/{observer_id}/scores

const url = 'https://app.everruns.com/api/v1/observers/example/scores';
const options = {method: 'GET'};

try {
  const response = await fetch(url, options);
  const data = await response.json();
  console.log(data);
} catch (error) {
  console.error(error);
}

curl --request GET \
  --url https://app.everruns.com/api/v1/observers/example/scores

Parameters

Path Parameters

observer_id

required

string

Prefixed public identifier

Query Parameters

session_id

string

Filter to one session.

limit

integer format: int64

Maximum number of scores returned (default 100).

offset

integer format: int64

Zero-based offset into the result set.

Responses

200

Success

application/json

Response wrapper for list endpoints. All list endpoints return responses wrapped in a data field.

object

data

required

Array of items returned by the list operation.

Array<object>

One score produced by an observer scorer for one trace slice. Linked back to the exact session/turn it graded; agent/harness identifiers are denormalized at scoring time for aggregation.

object

agent_id

Agent active in the session at scoring time.

string | null

created_at

required

string format: date-time

error_message

Error details if errored.

string | null

harness_id

Harness of the session.

string | null

required

External identifier (score_<32-hex>). Shown as “id” in API.

string

judge_cost_usd

Judge call cost in USD when the provider reports it (llm_judge only).

number | null format: double

judge_input_tokens

Judge LLM input tokens (llm_judge scores only).

integer | null format: int64

judge_output_tokens

Judge LLM output tokens (llm_judge scores only).

integer | null format: int64

label

Optional categorical label from an LLM judge (e.g. missing_source).

string | null

observer_id

required

Observer that produced this score.

string

pass

Whether the scorer passed (set when completed).

boolean | null

reason

Human-readable explanation (set when completed). For LLM judges this is the judge’s reasoning — retained as the raw material for the Phase 2 improvement loop.

string | null

scorer_key

required

Scorer key within the observer.

string

session_id

required

Session this score grades.

string

status

required

Lifecycle of one trace score. pending rows double as the scoring queue.

string

Allowed values: pending scoring completed errored skipped

turn_id

required

Turn this score grades (turn scope).

string

updated_at

required

string format: date-time

value

Score value 0.0–1.0 (set when completed).

number | null format: double

Example

{
  "data": [
    {
      "id": "score_01933b5a000070008000000000000001",
      "observer_id": "observer_01933b5a000070008000000000000001",
      "scorer_key": "grounded",
      "session_id": "session_01933b5a000070008000000000000001",
      "status": "pending",
      "value": 0.85
    }
  ]
}