Act as a no-code automation specialist. I have an automation in tool: Zapier, Make, n8n or another that does describe what it does and connects list the apps involved. It runs frequency and has already failed in these situations describe the errors you have seen. Follow these steps: 1) List every point where this automation can break, separating connection failures, data in an unexpected format, rate limits, duplicated items and steps that ran only halfway. 2) For each point, give the right handling: retry with progressive backoff, alternate path, validation before writing, and an idempotency check so records are not duplicated. 3) Describe how I find out when something fails, including the alert channel, what the message must contain and what is not worth notifying. 4) Build a recovery plan explaining how to reprocess what was missed without repeating what already succeeded. 5) Suggest one simple test per risk, including the input I should use to trigger the failure on purpose. Format: a table with Risk, Symptom, Handling and Test, followed by a checklist of what must be in place before I let this automation run unattended.