🕵️

ANDERSON

The System — SRE Health Monitor

Every 5 minutes, Anderson checks the heartbeat of the entire Matrix. CPU, RAM, disk, services — if anything crashes, Anderson auto-restarts it and alerts Neo on WhatsApp.

A Real Alert

Zion crashes at 3 AM — Anderson detects, fixes, and reports

⏰ CRON every 5 minutes triggers 🕵️ ANDERSON health.py check ⚠️ zion.service = dead! auto-restart AUTO-FIX systemctl restart zion verify → is-active ✓ alert Neo 📱 WHATSAPP ALERT 🚨 [Anderson] ALERT zion.service was DEAD → restarted → ACTIVE ✓ 3:00 AM — Zion crashes (OOM) 3:05 AM — Anderson detects 3:05 AM — Restarted + alert sent

A Real Alert — Trinity-CLI Edition

Trinity-CLI task fails or hangs — Anderson catches it within 5 min and alerts Neo

📱 NEO (WhatsApp) @trinity-cli loop test 🔻 TRINITY-CLI claude -p "..." (Opus 4.7) ⚠️ multi-turn loop impossible writes status='failed' 🗄️ trinity_tasks (SQLite) #391ea62a → status='failed' reads 🕵️ ANDERSON health.py — every 5 min check_trinity_cli_ failed_tasks() 🚨 1 failed + 1 stale 📱 WHATSAPP ALERT 🚨 [Anderson] Trinity-CLI failed task #391ea62a + 1 stale claim > 5 min → Neo investigates DETECTION TYPES: ❌ status='failed' (last 24h) ⏰ status='claimed' & claimed_at > 5 min ago (stale/hung) runner crashed / Opus errored runner picked it up but never marked done — likely hung

Everything Anderson Monitors

Anderson is the sun — checks radiate across the entire system

🕵️ ANDERSON RESOURCES system vitals 🖥️ CPU usage % 🧠 RAM (24 GB total) 💾 Disk space 🌡️ Load average SERVICES systemd watchdog 🏛️ zion.service 📡 whatsapp-bridge 🔺 trinity-ws.service 📅 zion-calendar 🔻 trinity-cli tasks ACTIONS auto-remediation 🔄 Auto-restart dead 🚨 WhatsApp alert 📊 Log to health.db 🧹 nightly cleanup NETWORK connectivity checks 🔌 TCP :8080 bridge 🌐 TCP :8765 WS 📡 Internet reachable ALERTS notification channels 💬 WhatsApp to Neo 📊 health_history.db 📝 journalctl logs 🔔 alert mode = verbose SCHEDULE cron orchestration ⏱️ health every 5 min 🧹 cleanup midnight 🔄 bridge restart 5 min 📊 stagger cron jobs

Tech Stack

RuntimePython 3 — cron triggered
ScheduleEvery 5 minutes (cron)
ChecksCPU, RAM, disk, systemd services, TCP ports, Trinity-CLI failed/stale tasks
ActionsAuto-restart, WhatsApp alerts, log to SQLite
CLI Watchercheck_trinity_cli_failed_tasks() — detects status='failed' (24h) + stale 'claimed' (>5 min)
Databasehealth_history.db (incident history)
Key FileAgentAnderson/health.py
Cost$0 — cron + Python + systemd