SLOs & reliability

An SLO sets a reliability target for your agents' actions — e.g. "99% of executions succeed over 24 hours" — and tracks the error budget as it burns. When the budget is exhausted, Nolixan can alert, throttle, or circuit-break.

Define an SLO

All routes are under /api/v1/slo (JWT; canManagePolicies for writes). environment_id is required.

Create

curl -X POST https://your-host/api/v1/slo/ \
  -H "Authorization: Bearer <JWT>" -H "Content-Type: application/json" \
  -d '{
    "name": "action-success",
    "environment_id": 1,
    "sli_type": "success_rate",
    "objective": 0.99,
    "window_hours": 24,
    "exhaustion_action": "alert"
  }'

• objective — target attainment as a fraction (e.g. 0.99).
• window_hours — the rolling window the budget is computed over.
• exhaustion_action — what to do when the error budget is spent: alert, throttle, or circuit_break.

The /slo/summary endpoint returns current attainment and how much of the error budget remains, so you can wire dashboards or alerts. This maps to SOC 2 A1.1 (availability) — see Compliance.