Workflow: Production Failure & Blast Radius Audit

When to use

• Before merging a PR
• Before deploying to production
• Any change touching APIs, persistence, queues, or external services

Goal

Identify failure modes that only show up under real production conditions: retries, partial failures, concurrency, and restarts.

---

Prompt (paste into Cascade)

You are reviewing this code as if it is already running in production under real load.

Assume the following WILL happen:

• requests will be retried (clients, load balancers, frameworks)
• downstream services will timeout or partially fail
• duplicate requests or events will occur
• the process may crash or restart mid-execution

Analyze this code and identify:

1. Failure modes (what can go wrong)
2. Blast radius (how far the failure spreads)
3. Data safety risks (duplication, corruption, loss)

For each issue, classify it as:

• BLOCKER – can cause production incidents or data loss
• RISK – works now but unsafe at scale
• OK – failure is contained or self-healing

For each BLOCKER or RISK:

• describe a concrete failure scenario
• propose the smallest possible fix

Focus on real behavior, not style or formatting.