Evaluation Run Object

skillEvaluationRunSchema in prompt-orchestra-contracts/src/evaluations.ts describes an evaluation batch.

Fields

• id, skillId, draftVersionId, publishedVersionId
• mode
• status
• createdAt, completedAt
• summary
• results

Each result may include:

• draftRunId
• publishedRunId
• contractComparison
• diagnosticSummary

API

• start: POST /api/skills/:id/evaluations/run
• poll: GET /api/skills/:id/evaluations/:evaluationRunId

Boundary note

The presence of draftRunId or publishedRunId does not create a family-agnostic run-detail API. Evaluation flows still use the evaluation result object directly, and skill-run inspection stays on the skill-run detail route family.

See also

• Evaluations
• Linking Results To Runs