Start TaskAgent Eval Run
Create a TaskAgent eval run and enqueue background execution. Callers are responsible for understanding candidate run side effects, external tool or integration mutations, and quota usage before starting an eval. An optional judge_config supplies one objective and an ordered rubric for the complete run. The reference remains a known-good comparator, and every rubric criterion must pass. Omitting judge_config uses the standard functional-equivalence rubric. The resolved configuration is returned by run detail. Objective and rubric text is sent to the selected external judge model provider and is visible to callers authorized to view TaskAgent Eval run detail. The candidate revision must differ from every reference run’s revision — evaluating a candidate against itself is rejected.
Authentication
API key or access token, sent as Authorization: Bearer <token>. The simplest option: create an API key in your dashboard under Settings, then paste it here.
API key or access token, sent as Authorization: Bearer <token>. The simplest option: create an API key in your dashboard under Settings, then paste it here.