Skip to main content

Inference health

GET 

/api/health/inference

Readiness for the two LLM-backed subsystems (the NL policy compiler and Agent-Bairav's Plain-English authoring) โ€” each backend's own health() is reused unchanged, this endpoint only aggregates. Read-only, no persistence, safe to poll. Never touches decide() โ€” inference readiness is completely independent of deterministic-evaluator health, matching ยง13's boundary.

Responsesโ€‹

Successful Response