7 questions before trusting an API automation

A happy-path demo proves that a workflow can run. These seven questions test whether it can fail without duplicating work, losing state, or leaving operators guessing.

1. What happens if the provider accepts the request but your client times out?

Look for: a durable operation record created before the call, a stable idempotency key, and reconciliation before any resubmit.

2. Can the same webhook arrive twice—or five times—without repeating the business effect?

Look for: event deduplication plus a unique effect record around charging, publishing, sending, or provisioning.

3. Can polling and webhooks race safely?

Look for: one shared, atomic state-transition path. The losing handler should become a no-op.

4. Which errors are actually retryable?

Look for: explicit handling for connection failures, 408, 429, selected 5xx responses, Retry-After, jitter, and a bounded attempt/time budget.

5. Can a late event move state backward?

Look for: monotonic transitions. A late running event must not overwrite success; a late success may need to reconcile a local timeout.

6. Where do unknown or invalid events go?

Look for: a dead-letter record with reason, payload hash, first-seen time, and replay status—not silent discard.

7. Can an operator answer “what happened?” without reading raw logs?

Look for: operation ID, provider reference, attempts, state history, last error, next action, and a safe manual-replay control.

Minimum failure-path test set

• Accepted request followed by client timeout
• Duplicate webhook delivery
• Webhook/poller race
• 429 with Retry-After
• Late success after local timeout
• Unknown operation and invalid signature

Need this applied to one live workflow?

FlowPatch delivers a fixed-scope $99 diagnostic for one external API/provider boundary within two business days after written scope acceptance and confirmed payment.

View the sample deliverable or email flowpatchpilot@proton.me.