Production incident

For an active incident:

detect
  → assess impact
  → stabilize
  → mitigate
  → verify recovery

Use production-ops for that path.

After stabilization:

implementation defect?
→ design-thinking / code-review

trust or authorization issue?
→ security-review

race / transaction / recovery proof is hard?
→ test-engineering

system ownership or guarantee is wrong?
→ engineering-design

Do not delay mitigation while users are still affected.