INCIDENT RESPONSE FOR AGENTS

ARTIFICIAL INTELLIGENCE ASSISTANT PROMPTS Prompt 1: Blameless postmortem. Use this after resolving any incident, to produce a first draft for human review. [START OF PROMPT] Write a blameless postmortem from these incident traces, focused on systemic causes and fixes rather than individual blame: [PASTE LAB OR PRODUCTION INCIDENT TRACES/LOGS HERE]. Output format: a postmortem document with sections Summary, Timeline, Root Cause, Impact, Systemic Contributing Factors, Action Items (each with an owner placeholder and due-date placeholder). [END OF PROMPT] Prompt 2: Full incident-response run. Use this to practice or to structure a real response from start to finish. [START OF PROMPT] Run a full incident-response process on this incident and produce the deliverables for each stage: detection summary, severity classification, containment action, root-cause analysis, blameless postmortem, and a specific new eval/regression test that prevents recurrence: incident data = [PASTE LAB OR PRODUCTION INCIDENT DATA/LOGS HERE]. Output format: one labeled section per stage. [END OF PROMPT]