agentusecasesAll 965 use cases
Software & DevOps

Run production deploys and port scans from Telegram with OpenClaw

A security professional uses an OpenClaw agent with read-only prod access, PR-only changes and manual approval for restarts to deploy changes and scan a VPS for exposed ports.

Done withOpenClaw

What they did
Simon Roses runs an OpenClaw agent, AgentX, on a Raspberry Pi 5 and talks to it through a Telegram bot. He asks it to deploy the latest changes to prod. It has read-only production access, can only open PRs, and needs his manual approval to restart services. It also runs nmap from a VPS over SSH to check his public IP for exposed ports, and alerts him on Telegram.
How it went
The agent ran port-exposure checks and flagged an example exposed gateway port, with a suggested firewall command. He reports about €50 in API spend after two weeks. It sometimes gets stuck and needs manual intervention.
Worth knowing
The scans and deploys needed sudo and docker group permissions. Keeping production read-only, PR-only and approval-gated for restarts is the safeguard to copy.

Try it yourself with OpenClaw

Work as my ops assistant over [Telegram] with read-only access to [production server]. When I ask, scan [VPS IP/host] for exposed ports and report findings, and propose deploy changes as pull requests on [repo] only. Do not restart services or change anything live until I approve each step.

Read the original ↗

Source: simonroses.com · Feb 1, 2026

Five of these in your inbox every morning

The best things people got an AI agent to do, each with the prompt to try it.

More like this