n8n Silent Failure & Retry Guardrails Checklist

• Name the failure modes before building.
• Classify each node as read, safe write, risky write, or irreversible.
• Set retry rules by risk tier.
• Add idempotency keys to every write.
• Create a dead-letter review path.
• Alert only when owner action is needed.
• Save handoff evidence.

Error workflow blueprint

Main workflow → Validate → Idempotency check → Risk-tiered action → Success log
Error workflow → Classify failed node → Dead-letter record → Alert or digest → Resume instructions

Dead-letter record fields

• workflow_name, execution_id, source_system, source_id
• failed_node, risk_tier, error_summary
• last_successful_step, customer_visible, duplicate_key
• recommended_next_action, owner, created_at

Delivery acceptance tests

• Happy path succeeds and logs evidence.
• Duplicate input does not duplicate risky write.
• Timeout/500 routes correctly.
• Missing input creates human-review item.
• Risky failure alerts owner instead of blind retry.