How to Install ClawMetry on KiloClaw — One Command, Full Observability

If you're running your OpenClaw agent on KiloClaw, you can install ClawMetry by simply asking your agent. No SSH. No config files. No Docker. Just paste one prompt.

Total time: about 2 minutes.

What is KiloClaw?

KiloClaw is hosted OpenClaw by Kilo.ai. You get a fully managed agent running in the cloud. No server to provision, no setup wizard to fight through. Sign up, pick your model, and your agent is live.

It's the fastest way to get an OpenClaw agent running. Your agent gets persistent storage, channels (Telegram, Discord, etc.), cron jobs, sub-agents, and the full OpenClaw runtime. All managed for you.

The problem

Your KiloClaw agent runs 24/7. It processes messages, spawns sub-agents, runs cron jobs, calls tools, burns tokens. And you have zero visibility into any of it.

• How many tokens did it use today?
• Which model is costing you the most?
• Is that cron job actually running?
• Did the sub-agent it spawned succeed or fail silently?
• What was it doing at 3 AM when you were asleep?

You don't know. You're flying blind. And the invoice at the end of the month is your only feedback loop.

The one-command setup

Open your KiloClaw agent chat and paste this:

That's it. Your agent takes it from here.

What happens

Your agent will do three things:

1. Install ClawMetry via pip. Takes a few seconds.
2. Connect to ClawMetry Cloud using clawmetry connect. This authenticates your instance with end-to-end encryption. Your data stays yours.
3. Set up supervisord to keep the sync daemon running. Even if your agent restarts, the ClawMetry sync process comes back automatically.

The whole process takes about a minute. Your agent handles every step, including answering the interactive prompts.

What you get

Once connected, open app.clawmetry.com. You'll see your KiloClaw agent's activity in real time:

• Agent sessions with full tool call traces and thinking
• Token costs broken down by model, session, and tool
• Memory changes to SOUL.md, MEMORY.md, and other workspace files
• Cron jobs with status, schedule, and failure history
• Sub-agents with real-time progress, commands, and outcomes
• Session history with timeline and cost breakdown

ClawMetry dashboard showing real-time token flow across your agent's sessions

You go from "I hope my agent is working" to "I can see exactly what it's doing, what it costs, and whether it succeeded."

That's it

No SSH into a server. No editing config files. No Docker containers. No manual pip installs in a terminal you don't have access to.

You pasted one prompt. Your agent did the rest. And now you have full observability into everything it does.

This is the beauty of hosted agents: the agent itself is the installer. You just tell it what you want.