Gbenga Akindolani |AI Engineer | Automation Expert

Every automation expert has shipped a workflow, then found out it broke when the client did. APIs time out. Credentials expire. A node gets a value it wasn't built for. The question was never "will it fail," it's "how fast do I find out." So I built a self-fixing system for n8n. When any workflow fails, Claude Code reads the broken workflow, finds the root cause, and sends me a Slack message with 4 buttons: → Auto Fix: Claude applies the fix it proposed, validates it, and saves → Replay: direct links to re-run a workflow that just hit a temporary glitch → Custom Fix: a form where I type the change in plain English and Claude applies it → Decline: logs the error, changes nothing THE STACK - n8n Error Trigger to catch every failure - Claude Code with the n8n-mcp server for full workflow access - ngrok to connect n8n cloud to my machine - Slack as the approval layer The result: workflows that diagnose and repair themselves while I sleep, with me still approving every change before it goes live. Watch the full video here: The complete PDF guide are linked in the video description, ready to download. You can also drop a comment and I'll send them over. If this is useful, a like, follow, and subscribe goes a long way. Follow for more on building production AI automation. #n8n #Automation #AIAgents #ClaudeAI #BuildInPublic