What You Can Monitor
Heartbeat Monitoring
Your bot pings AtomPing every few minutes over HTTPS. No public IP, no open ports, no domain needed. Works on laptops, VPS, Docker, Raspberry Pi — anywhere OpenClaw or your bot runs.
Token & Cost Tracking
Track input/output tokens per hour and estimated cost. Set spend alerts so a runaway conversation in a public Telegram group doesn't generate a surprise bill from Anthropic or OpenAI.
AI API Health
Know when Claude or GPT goes down — before your users notice. The SDK reports AI call success/failure and latency with every heartbeat ping.
Memory & CPU Monitoring
See RAM and CPU usage over time. Catch memory leaks days before they crash your OpenClaw instance. Get alerts when RAM exceeds 90% — before OOM kill.
Message Activity Tracking
Monitor messages processed per hour and time since last message. Detect silent failures where the process is alive but Telegram long-polling has silently died.
Native OpenClaw Skill
Run openclaw install atomping-monitor and add one env var. The skill hooks into OpenClaw's lifecycle — auto-starts monitoring on boot, tracks messages and AI calls through the skill system.
Setup in 3 Steps
Create a Heartbeat Target
Add a Heartbeat monitor in the AtomPing dashboard. Give it a name, set the expected ping interval (e.g. 5 minutes), and choose your alert channels.
Install the Skill or SDK
For OpenClaw: openclaw install atomping-monitor. For any Python bot: pip install atomping-heartbeat. One line to start. Background thread, zero performance impact.
Bot Reports Its Health
Every few minutes, your bot sends a heartbeat with system metrics, message counts, AI status, and token usage. If the ping stops — you get an instant alert.
Built for Every Type of Bot
OpenClaw & ClawdBot
Your self-hosted AI assistant runs on your laptop. No public IP, no domain. When it crashes while you're AFK, heartbeat monitoring alerts you instantly. Native OpenClaw skill — one command setup.
Telegram & WhatsApp Bots
Monitor AI-powered Telegram bots, customer support bots, and WhatsApp automation. Catch long-polling failures, FloodWait cascades, and silent channel disconnects.
Discord Community Bots
Keep moderation bots, image generation bots, and AI assistants online 24/7. Detect WebSocket gateway disconnects and event loop freezes that Discord's own reconnection misses.
AI SaaS & API Backends
Run multiple AI bots for clients? Monitor them all from one dashboard. Track per-bot token usage, cost, and uptime. White-label status pages included.
AI Bot Monitoring FAQ
My OpenClaw runs on a laptop behind NAT. Can I still monitor it?
Yes. Heartbeat monitoring works entirely outbound — your bot sends pings to AtomPing over HTTPS. No inbound connections, no port forwarding, no static IP, no domain required. If your laptop can reach the internet, your OpenClaw instance can be monitored.
How quickly will I know if my bot crashes?
Within one heartbeat interval. If your bot pings every 5 minutes and misses a ping, AtomPing confirms from multiple regions and sends an alert within 6-7 minutes total. For faster detection, set 1-2 minute intervals.
Does the monitoring SDK affect my bot's performance?
No. The SDK runs in a lightweight background thread (or async task), sends one small HTTPS POST every few minutes, and catches all errors silently. It uses ~2MB of RAM and negligible CPU. If AtomPing is unreachable, it simply retries next cycle.
Can I track how many tokens each bot consumes?
Yes. Add monitor.track_ai_call(tokens_in=X, tokens_out=Y) after each AI API call. The SDK aggregates tokens over rolling 1-hour windows and reports them with each heartbeat. See tokens per hour, totals, and estimated cost in the dashboard.
What platforms are supported?
OpenClaw has a native skill (one command install). Any Python-based bot works with the SDK: aiogram, python-telegram-bot, discord.py, Slack Bolt, custom frameworks. Node.js SDK is planned. For any language, the heartbeat API is a single POST request you can implement in 5 lines of code.
How is this different from Healthchecks.io?
Healthchecks.io monitors cron jobs with dead man's switch pings. AtomPing does that too, but adds AI-specific metrics: token tracking, AI API health, cost alerts, and memory monitoring. Plus multi-region verification, incident detection with hysteresis, and the same alerting stack (Telegram, Slack, Discord, webhooks) you use for your websites.