Skip to content
NEWMarketplace and your team: sell what you build, and get plans and changes that are tested before you see them. See what is new

Library / Monitoring, Ops & Incidents

Monitoring, Ops & Incidents

Monitor n8n workflow errors and OpenAI spend with Slack, Gmail and Sheets

Checks n8n execution health and LLM cost every hour, pauses runaway workflows, and uses an AI agent to write root-cause reports to Slack, Gmail and Sheets.

What it does

## 🩺 n8n Instance Health & Cost Monitor v3 For teams running a busy n8n instance, where one expired credential or a looping AI agent can fail hundreds of times overnight. By morning the on-call engineer is staring at fifty red executions with no idea which one matters. ### How it works Every hour the workflow reads recent executions for every active workflow through the n8n API and scores error rate, volume and LLM spend. Runaway workflows are switched off. For each incident, an AI agent reviews the failing node's configuration, past incidents and the weekly baseline, then writes a root cause and fix steps. That goes to Slack with Re-enable, Acknowledge and Snooze buttons. Serious cases page on-call by email. Every Monday it compares the week with earlier weeks, clusters similar errors, draws an 8-week trend chart and emails an AI-written health report. ### Setup steps 1. Create an n8n API key and add it as Header Auth (X-N8N-API-KEY). 2. Create a Google Sheet with Guard_Log and Metrics_History tabs. 3. Connect Google Sheets, Slack, Gmail and OpenAI. 4. Set your instance URL in both Set Monitor Config and Set Action Lane Config, plus thresholds and emails. 5. Tag must-not-stop workflows no-auto-disable, run the guard once manually, then activate.

The steps

  1. Every Hour (Guard)
  2. Every Monday 9 AM (Digest)
  3. Set Guard Mode (1h Window)
  4. Set Digest Mode (7-Day Window)
  5. Set Monitor Config
  6. Fetch Active Workflows
  7. Build Workflow Index
  8. Fetch Executions per Workflow
  9. Compute Health & Cost Metrics
  10. Route by Mode
  11. Read Guard Log Sheet
  12. Read Metrics History Sheet
  13. Classify & Dedupe Incidents
  14. Build Failure Sample List

51 steps in total.

Similar automations