{
  "checklist_items": [
    {
      "id": "chk_debugging_via_evidence_feeding",
      "item": "Security Engineering and Data Infrastructure teams debug production issues by feeding Claude Code stack traces and screenshots directly, rather than describing the problem from memory.",
      "category": "documentation_runbook",
      "source_title": "How Anthropic teams use Claude Code",
      "source_url": "https://claude.com/blog/how-anthropic-teams-use-claude-code",
      "summary": "Feeding raw evidence (stack traces, screenshots) into context produces more reliable debugging runbooks than paraphrased problem descriptions.",
      "created_at": "2026-07-02 06:02:48",
      "cite_as": "https://subagentchecklists.com/api/checklist-items/chk_debugging_via_evidence_feeding"
    },
    {
      "id": "chk_runbook_consolidation",
      "item": "The Security Engineering team has Claude \"ingest multiple documentation sources to create markdown runbooks and troubleshooting guides\" -- condensed docs become context for debugging, which they found more efficient than searching full knowledge bases.",
      "category": "documentation_runbook",
      "source_title": "How Anthropic teams use Claude Code",
      "source_url": "https://claude.com/blog/how-anthropic-teams-use-claude-code",
      "summary": "Consolidating scattered docs into a single dated runbook (with citations back to sources) beats searching a full knowledge base each time.",
      "created_at": "2026-07-02 06:02:48",
      "cite_as": "https://subagentchecklists.com/api/checklist-items/chk_runbook_consolidation"
    }
  ]
}