Quick overview
This workflow receives industrial/IoT alarms via webhook, suppresses duplicate chatter, optionally summarizes the alarm with Anthropic Claude, and routes notifications by severity with tiered ACK-based escalation and timeouts using HTTP endpoints.
How it works
- Receives an alarm payload through an n8n Webhook endpoint.
- Normalizes the incoming fields and applies flood suppression to drop repeat alarms within a 5-minute window.
- If the alarm is suppressed, it records an audit event and stops.
- If not suppressed, it requests an advisory one-sentence summary, likely cause, and priority from Anthropic Claude and attaches the result to the alarm.
- Routes the alarm by severity: critical starts a tiered on-call escalation, warning sends a single notification, and info only creates a log entry.
- For critical alarms, it notifies Tier 1 with an ACK URL, waits up to 5 minutes for an acknowledgment, and escalates to Tier 2 and then Tier 3 if no ACK is received.
- Logs the final outcome (warning notified, acknowledged at Tier 1/2, or escalated to manager) for audit purposes.
Setup
- Copy the webhook URL from the Webhook trigger and configure your SCADA/IoT system to POST alarms to it.
- Create an Anthropic (Claude) HTTP Header Auth credential (send
x-api-key plus the required anthropic-version header) and attach it to the Claude request node if you want AI triage.
- Replace the placeholder notification URLs with your SMS/Slack/Teams/email endpoint(s) and ensure they accept the JSON payload (including
ackUrl for critical alarms).
- Ensure your acknowledgment mechanism calls the provided
ackUrl (for example by opening it or POSTing { "ack": true }) within the timeout windows.
- Connect the audit log steps to your preferred storage (for example Google Sheets or a database) if you want persistent reporting.