Get a run's rubric scorecard
The deterministic rubric result for one run. No model, no spend.
Divide by sessionsGraded, never sessionsTotal, and keep failedGradingCount separate from failCount — a crashed judge is not a regression.
Authorizations
MCPJam API key (sk_…). Create one at Settings → API keys. Guest sessions cannot use the API, and API keys cannot manage other API keys.
Path Parameters
ID of the hosted project that contains the server.
Eval run ID, as returned by POST /eval-runs.
Response
The scorecard.
Deterministic rubric result for one run. NO MODEL IS INVOLVED and nothing is spent — reach for this before the model-generated wave insights, which usually is not the question you actually have.
EVERY criterion the run's rubric declared, in snapshot order — including ones nothing was graded against. An absent row would be indistinguishable from a criterion that was never configured.
The DENOMINATOR for any rate you compute. Never divide by sessionsTotal — 3 failures of 4 graded sessions out of 40 attempted is not 7.5%.

