PagerDuty Incident Postmortem Writer

What it does

PagerDuty Incident Postmortem Writer is an SRE-focused agent that turns PagerDuty incident data into a complete blameless postmortem draft. It collects incident details, notes, and log entries from the PagerDuty API, builds a chronological timeline, scores findings (severity, duration, escalations, responders), and synthesizes narrative postmortem sections — executive summary, impact analysis, root cause hypotheses, a 5 Whys framework, and lessons learned — using an LLM with a deterministic fallback. The draft is returned as structured JSON and can be pushed to Confluence and announced in Slack. Fully offline in mock mode — no credentials required for testing. Built for SRE and platform teams who want fast, consistent, blameless postmortems instead of manually reconstructing incident timelines.

Example prompts

• Generate a full blameless postmortem for this PagerDuty incident
• Build a timeline of events for this incident
• Give me 5 Whys root-cause questions for this outage
• Get a quick summary of this incident without a full postmortem

Tools (8)

Tools the agent exposes — your AI client calls them automatically when it needs them.

• run_full — Generate a complete blameless postmortem draft from a PagerDuty incident.
• analyze_incident — Run deterministic analysis on a PagerDuty incident and return scored findings. No LLM call.
• generate_timeline — Build a chronological timeline of events for an incident, ordered from trigger to resolution.
• build_five_whys — Generate structured 5 Whys root-cause analysis prompts for an incident (blameless, system-focused).
• get_incident_summary — Return a concise incident summary without generating the full postmortem.
• discover_intent — Route a freeform user request to the most appropriate tool.
• list_capabilities — List all available tools and their purposes.
• plan_inputs — Plan inputs for a tool: returns questions, JSON schema, and a ready-to-edit example.