Workflow

Each run, this agent gathers the evidence for the incident given in the incident run input, reconstructs what happened, and drafts a grounded root-cause analysis that states the likely cause with its supporting evidence and confidence, plus recommended follow-ups and the questions still left open.

1. Gathers the relevant runbooks, postmortems, and timeline notes for the incident run input from Google Drive or Microsoft SharePoint.

2. Reconstructs the incident timeline from the team threads and alerts in Slack or Microsoft Teams, noting when symptoms appeared, what was tried, and when service recovered.

3. Finds the related bugs, regressions, and remediation owners in Linear or Jira.

4. Drafts the root-cause analysis for the incident run input and writes it for the Audience audience: states the likely root cause with its supporting evidence and a confidence level, recommends remediation and follow-up actions with owners, and lists the open questions and unknowns explicitly. Read-only, it never edits tickets, messages, or documents.