Automated Monitoring for OpenClaw Bots: A Practical Guide

Manual monitoring means you find out your bot is down when a user complains. Automated monitoring means the crash is fixed before most users notice. Here's how to set it up — from simple cron scripts to a full repair agent.

The monitoring gap in OpenClaw

OpenClaw doesn't include built-in uptime monitoring. When the gateway process dies, nothing alerts you. You find out via:

• A user DMing "your bot isn't working"
• Checking Telegram and noticing the silence
• Looking at the server and seeing the process is dead

The average developer finds out 4–6 hours after the crash. By then, multiple users have already bounced.

Option 1: Simple cron watchdog (free, DIY)

The cheapest approach: a cron job that checks the process and restarts it if dead.

#!/bin/bash
# Simple OpenClaw watchdog — add to cron: */5 * * * * /path/to/watchdog.sh

LOG="/var/log/openclaw-watchdog.log"
TS=$(date '+%Y-%m-%d %H:%M:%S')

if ! pgrep -f "openclaw-gateway" > /dev/null; then
  echo "[$TS] Gateway not running — restarting" >> "$LOG"
  systemctl restart openclaw-gateway 2>> "$LOG" || \
    openclaw gateway restart 2>> "$LOG"
  
  # Optional: send Telegram alert
  # curl -s "https://api.telegram.org/bot${BOT_TOKEN}/sendMessage" \
  #   -d chat_id="${CHAT_ID}" \
  #   -d text="⚠️ Gateway restarted at $TS" > /dev/null
else
  echo "[$TS] Gateway OK" >> "$LOG"
fi

# Run every 5 minutes
*/5 * * * * /usr/local/bin/openclaw-watchdog.sh 2>&1

Option 2: systemd with auto-restart (slightly better)

If you're on Linux, systemd can restart the gateway on crash with a configurable delay:

[Unit]
Description=OpenClaw Gateway
After=network.target

[Service]
User=your-username
WorkingDirectory=/home/your-username/.openclaw/workspace
ExecStart=openclaw gateway start --foreground
Restart=on-failure
RestartSec=10s
# Stop auto-restart after 5 tries in 5 minutes (prevents crash loops)
StartLimitInterval=300
StartLimitBurst=5

[Install]
WantedBy=multi-user.target

This catches simple crashes. But it still won't tell you what happened, and it won't fix config corruption or disk issues.

Option 3: Full automated repair agent

This is what Mechanic Bot does. Instead of a dumb restart, it:

What "automated repair" actually means

When Mechanic detects the gateway is down, it runs through a decision tree:

The key difference: step 3. Dumb restart fails silently if the config is broken. A repair agent fixes the underlying cause.

Setting up Mechanic Bot

One command installs the monitoring agent on your Linux VPS or Mac:

curl -fsSL https://mechanicbot.io/onboard.sh | sudo bash -s -- \
  --token YOUR_TOKEN \
  --email you@example.com

curl -fsSL https://mechanicbot.io/onboard-mac.sh | bash -s -- \
  --token YOUR_TOKEN \
  --email you@example.com

The agent runs as a non-root systemd service (Linux) or launchd service (Mac), connects to the Mechanic hub via WebSocket, and starts monitoring your gateway within 30 seconds.