---
title: Provider outage alerts (multi-provider API status monitor)
---

# Provider outage alerts (multi-provider API status monitor)

## What it is

The same automatic outage alerting Claude already has, extended to the
OTHER API providers you use — but filtered so you only ever see an alert for a provider
you have actually used recently. If OpenAI, Cursor, or Groq has an outage, Omniscio tells
you; if a provider you never touch has an outage, it stays silent.

## Where to find it

Alerts arrive in two places at once: an inbox card in the Omniscio **Inbox**, and a **desktop notification**. The card itself carries the incident text, a **View status page** button and the usual **Start session** button. The switch that controls all of this sits in **Settings → Notifications**, under the entry **"Other provider outage alerts"**. Claude's own outage alerts have a separate toggle on that same page.

## How it behaves

### The rule: only providers you've used in the last 7 days

This is the whole point. Omniscio checks whether you've used a provider in the **last 7
days** — either as a coding session (Codex, Cursor, …) or through a background feature
that billed against it (OpenAI, Groq, …). A provider outside that window is **not even
polled**, and never raises a card. Start using a provider again and its monitoring resumes
on its own within a minute or two.

This gate applies to **every** provider, Claude included — though in practice you use
Claude constantly, so its alerts effectively always show.

### Which providers are watched

Omniscio only watches a provider that publishes a **reliable public status feed**. Today
that's:

- **Claude / Anthropic** (its own dedicated monitor — see [anthropic-status-monitor.md](anthropic-status-monitor.md))
- **OpenAI** (status.openai.com) — covers Codex and GPT sessions
- **Cursor** (status.cursor.com)
- **Groq** (groqstatus.com) — used by some background AI features

Providers **without** a reliable public status feed are deliberately **not** watched, and
that's expected — this includes **Google / Gemini** (no compatible status feed), **xAI /
Grok** (its feed blocks automated access), and **DeepSeek / OpenRouter** (no reachable
feed). Adding a provider later is just wiring up a verified feed.

### What an alert looks like

When a watched provider has an incident affecting its API, you get **one** inbox card and
**one** desktop notification. The card:

- names the incident and explains it's the provider's outage, not your setup;
- has a **View status page** button that opens that provider's own status page (a fixed,
  trusted address — never a link pulled from the feed);
- also offers the universal **Start session** button, like every alert;
- **auto-clears** when the provider marks the incident resolved (a "resolved" notification
  fires), and dismissing it sticks across restarts.

Only components that matter to your API usage count — a provider's marketing-site or
web-app-only incident won't alarm you. If a status page is simply unreachable, that's
treated as your network (the offline banner's job), never as a provider outage.

### Turning it on or off

On by default. **Settings → Notifications → Other provider outage alerts**
(`providerStatusMonitorEnabled`). Turning it off clears any standing cards quietly. Claude's
own alerts have their own separate toggle (Anthropic outage alerts).

## Related

Claude's own monitor is a separate, dedicated system rather than a row in this one, and [anthropic-status-monitor.md](anthropic-status-monitor.md) describes it. For how an inbox card is raised, dismissed and auto-cleared in general, see [inbox-alerts.md](inbox-alerts.md) and the wider [inbox-overview.md](inbox-overview.md). What gets through as a desktop notification, and how silence is configured, is [notifications-and-silence.md](notifications-and-silence.md). The other family of warning cards about your own AI usage rather than a vendor's outage is [ai-spend-alerts.md](ai-spend-alerts.md).