Health Monitor

Axiom can ping the active LLM provider on a schedule, mark it healthy / degraded / down, and automatically fall back to another provider when the active one misbehaves. This panel controls that loop.

URL: /settings?tab=healthMonitor

Enabled

Master toggle. When off, no health checks run and the topbar health badge turns neutral.

{
  "healthMonitor": {
    "enabled": true
  }
}

Health check interval

How often each provider gets pinged, in minutes. Default: 5.

{
  "healthMonitor": {
    "intervalMinutes": 5
  }
}

Fallback trigger

When to consider a provider bad enough to switch away from it.

Value	Meaning
down	Only fall back when the provider is fully unreachable / erroring.
degraded	Fall back on slow responses or partial failures too.

Choose degraded if your users notice latency, down if you'd rather tolerate some slowness than bounce between providers. Default: down.

{
  "healthMonitor": {
    "fallbackTrigger": "down"
  }
}

Failures before fallback

How many consecutive failing checks are required before the monitor flips a provider to down and triggers the fallback. Minimum: 1. Default: 1.

Higher = slower reaction, fewer false positives.

{
  "healthMonitor": {
    "failuresBeforeFallback": 1
  }
}

Recovery check interval

Once a provider has fallen out, it gets re-checked on this shorter interval (in minutes) to see if it's back. Default: 1.

{
  "healthMonitor": {
    "recoveryCheckIntervalMinutes": 1
  }
}

Successes before recovery

How many consecutive successful checks are required before marking the provider healthy again and allowing it back into the rotation. Default: 3.

{
  "healthMonitor": {
    "successesBeforeRecovery": 3
  }
}

Notifications

Admin notifications the monitor can emit on state transitions, delivered through the agent's normal notification channels (e.g. Telegram if configured).

Each transition is a separate toggle:

Key	Transition	Default	Meaning
healthyToDegraded	Healthy → Degraded	false	Provider started responding slowly or with partial failures.
degradedToHealthy	Degraded → Healthy	false	Provider recovered from a degraded state.
degradedToDown	Degraded → Down	true	Provider went fully offline after being degraded.
healthyToDown	Healthy → Down	true	Provider went fully offline without a degraded phase.
downToFallback	Down → Fallback	true	Axiom automatically switched to another provider.
fallbackToHealthy	Fallback → Healthy	true	Original provider recovered and was switched back in.

Turn everything off if you prefer a silent, self-healing monitor.

{
  "healthMonitor": {
    "notifications": {
      "healthyToDegraded": false,
      "degradedToHealthy": false,
      "degradedToDown": true,
      "healthyToDown": true,
      "downToFallback": true,
      "fallbackToHealthy": true
    }
  }
}