Docs
Quickstart
Sign up, connect an n8n instance, add an alert channel. About five minutes, and the free tier needs no card.
Three steps, and the third is the one people skip. The free tier has all ten detectors and needs no card.
1. Create an account
Sign up at app.obsevo.com. You will get a verification email — the workspace is created for you, and the account that creates it owns it.
2. Connect your n8n
In the dashboard, Connect instance. You will be asked to choose how Obsevo reads n8n, and that choice matters more than anything else on this page — connecting an instance makes the recommendation and explains the trade-off.
The short version:
| Mode | What runs | Use it when |
|---|---|---|
| Agent | one container next to n8n | you want the most resilient setup, and execution payloads never to leave your host |
| API key | nothing on your side | you want to be running in two minutes, or cannot run a container beside n8n |
The agent keeps reporting when the n8n API stops answering, and it reduces execution data on your infrastructure so only metadata reaches us. API-key mode is faster to start and easier to justify to yourself at 11pm; you can switch later.
Either way, connecting gives you an instance ID and a one-time secret. Keep them — the secret is shown once.
3. Add an alert channel and test it
Do not skip this step. An instance with no alert channel detects everything and tells nobody, which looks identical to a product that does not work.
Go to Settings → Alerts, add a channel, and send a test alert:
- Slack, Discord or Telegram — an incoming webhook URL you create in that app.
- Email — no setup needed.
- Webhook — a signed JSON POST, for PagerDuty, Opsgenie or n8n itself.
Bring your own Slack or Discord credentials; the channels themselves are included on every plan, including Free.
What to expect next
Obsevo needs a little history before its two headline detectors can say anything:
- Failures are reported immediately.
- Missed runs need one observed run to establish the cadence, then fire once the next expected time passes plus a five-minute grace.
- Green-but-broken needs about five successful runs before it will judge a workflow, so it does not call a brand-new workflow broken for having no history.
If runs are not appearing at all, verify the connection — there is a short list of things that go wrong and each has a specific symptom.