Update to 33CRM

{
  "description": "Review 33CRM conversation workload by channel and operational status, including active, waiting, escalated, resolved, timestamps, and message counters. Use for inbox workload, escalation volume, channel mix, stale-conversation checks, or conversation operations.",
  "included_files": [
    {
      "relative_path": "agents/openai.yaml",
      "size_in_bytes": 459
    }
  ],
  "name": "conversation-operations-review",
  "skill_md_contents": "    ---\n    name: conversation-operations-review\n    description: Review 33CRM conversation workload by channel and operational status, including active, waiting, escalated, resolved, timestamps, and message counters. Use for inbox workload, escalation volume, channel mix, stale-conversation checks, or conversation operations.\n    ---\n\n    Use this skill for conversation operations, not conversation-content analysis.\n\nWorkflow:\n1. Resolve the period.\n2. Call `get_conversation_report` for volume by current channel and status for conversations created in the window.\n3. Call `list_conversations` for operational records filtered by channel, status, or activity dates.\n4. Call `get_conversation` for one conversation's metadata and total/inbound/outbound/per-sender counters.\n5. Use pagination when needed.\n\nOutput:\n- Conversation volume\n- Channel mix\n- Status mix\n- Escalated / waiting workload\n- Specific conversation counters when requested\n- Operational attention items\n- Caveats\n\nRules:\n- Never claim to have read message text: message bodies, attachments, and media are not returned.\n- Current status is an as-of snapshot and may differ from status during the requested window.\n- Do not infer customer intent, medical information, sentiment beyond the explicit stored field, or content not returned.\n- Reply in the user's language.\n- Read-only only.\n"
}