Add automated health monitoring via Claude Code cron job

[dennisonbertram]

Overview

Set up a periodic health check that uses Claude Code to SSH into the server, evaluate system state, and fire alerts if anything is wrong.

What to check

• GET /v1/system/health → must return {"status":"ok"}
• docker ps → Traefik, registry, and all deployed services must be running
• Disk usage → alert if > 80%
• ah.service systemd status → must be active
• HTTPS endpoint → curl -sI https://agentic.hosting/ must return HTTP 200

Implementation sketch

# /usr/local/bin/ah-health-check.sh (runs via cron on local machine or server)
claude -p "You are a health monitor for agentic-hosting server. SSH and check: (1) /v1/system/health returns ok, (2) docker ps shows traefik+registry+website running, (3) disk < 80%, (4) https://agentic.hosting/ returns 200. Output ALERT: <details> for any failure, OK if all pass." --output-format text | grep "^ALERT" | mail -s "agentic-hosting alert" ops@agentic.hosting

*/15 * * * * /usr/local/bin/ah-health-check.sh

Alert channels to support

• Email
• Slack webhook
• SMS (Twilio)
• Custom webhook

Labels

enhancement, operations, monitoring