agent-system-live-quality-ops

Operate and validate AgentSystem live quality alarm (systemd timer + alert state) to keep memory/proactivity/consistency regressions visible in real time.

SKILL.md

GitHub

AgentSystem Live Quality Ops

Use this skill when you need to verify that live quality alerts are active and not silently failing.

Quick Start

  1. Dry-run alarm logic:
bash /Users/user/clawd/scripts/agent-system-live-quality-alarm.sh --dry-run
  1. Inspect timer + last run:
bash scripts/check_live_alarm.sh
  1. Force a manual alert test (optional):
bash /Users/user/clawd/scripts/agent-system-live-quality-alarm.sh --force-alert --dry-run

Workflow

  1. Confirm hachix-agent-system-live-quality-alarm.timer is active.
  2. Verify latest health report and alarm state file exist.
  3. Check if alert suppression/cooldown is working as expected.
  4. If timer is inactive, enable it and re-run a dry check.
  5. Record findings in reports or memory notes.

Resources

  • scripts/check_live_alarm.sh: one-shot operational status summary.
  • /Users/user/clawd/scripts/agent-system-live-quality-alarm.sh: main alarm runner.
  • /Users/user/clawd/reports/agent-system-plan2-health-latest.json: last health metrics snapshot.