incident-postmortem-assistant

What This Skill Does

The incident-postmortem-assistant is a professional SRE-focused agent skill designed to streamline the post-incident review process. It takes raw, unstructured incident data—such as Slack logs, monitoring alerts, and timeline notes—and transforms them into a structured, blameless postmortem draft. The skill strictly adheres to professional engineering standards, focusing on systemic analysis rather than individual attribution. It identifies the root cause (the primary technical failure), contributing factors (inducers), technical amplifiers (factors that worsened the impact), and provides a clear separation between immediate recovery actions and long-term preventative improvements.

Installation

To integrate this skill into your environment, run the following command in your terminal: clawhub install openclaw/skills/skills/52yuanchangxing/incident-postmortem-assistant Ensure your current working directory has access to the provided resources folder to leverage the internal template and spec validation files.

Use Cases

• Post-Incident Analysis: Quickly synthesizing raw alert data into a formal incident review document.
• SRE Workflow Standardization: Ensuring every incident report follows the same organizational structure for RCA (Root Cause Analysis).
• Educational Tooling: Assisting junior engineers in differentiating between root causes and mere symptoms during the debrief process.
• Documentation Automation: Generating Markdown drafts based on incident logs while maintaining the context of complex distributed system failures.

Example Prompts

1. "Here are the incident logs from the last 4 hours: [pasted logs]. Please generate a postmortem draft highlighting the root cause and suggested follow-up tasks."
2. "Review this timeline: [timeline]. I need an analysis identifying why the cache miss rate amplified the latency issues during our deployment."
3. "Summarize the impact of this incident based on the attached monitoring data and suggest three preventative actions."

Tips & Limitations

• Blameless Culture: The agent is hard-coded to prioritize blameless language. It will automatically filter out or flag accusatory phrasing.
• Data Privacy: Always redact PII or sensitive credentials before feeding logs into the agent.
• Verification: The agent is designed as an assistant. Always review the 'root cause' analysis against your team's internal technical debt and system architecture before finalizing reports.
• Missing Information: If data is incomplete, the agent is configured to output an 'awaiting clarification' list rather than inventing system failures.