# Operations, availability, and client resilience

## Current integration endpoints

- Identity: `https://identity.vinquery.com/connect/token`
- DecisioQ: `https://dde.vinquery.com/api/v1`
- Health probe: `GET https://dde.vinquery.com/health`

Use explicit connection and overall timeouts, bounded retry with jitter, and a circuit breaker for sustained failures. Generate a unique `X-Correlation-Id` (maximum 200 characters) for each business operation.

## Billing behavior

A successfully processed `POST /api/v1/decide` consumes one credit. Authentication, discovery, `/validate`, malformed requests, authorization failures, rate-limit responses, and server failures are not successful billable decisions.

## Service commitments not yet published

There is currently no public status page, numeric public rate-limit table, or client-facing SLA/SLO in this package. These must not be implied in contracts or user interfaces. Before offering them, VinQuery should establish:

1. An independently hosted public status page with incident history and subscriptions.
2. Measured availability and latency SLOs, maintenance windows, and escalation ownership.
3. Per-plan quotas and rate limits returned in machine-readable headers.
4. Incident severity levels, response targets, and post-incident communication rules.

Until those are published, clients should obtain commercial service commitments from their VinQuery contact.