Experimenting with Claude Code CLI in the Home Lab 🛡️
2026-05-23 · AI Pentesting HomeLab Ludus RedTeaming
I've been wanting to get familiar with the Claude Code CLI for a while now, so I finally dedicated some time to put it to the test in my home lab.
Inspired by some of the cool work I've seen by @alleem regarding pentesting, I decided to see if I could teach Claude to act as an "internal pentester." My goal was to have it generate reports based on the HackSmarter example report format.
The Workflow
I used a Proxmox + Ludus setup to host the LeHack 2024 environment. I built a Kali machine and gave Claude access to it to run commands, train, and validate the creation of a custom NetExec skill.
My Takeaways
The Learning Curve
It was incredibly insightful, though I definitely had to correct the model quite a bit to keep it on the right workflows.
Structure Matters
I realized that Claude performed much better once I created a standardized folder structure including existing scripts. This prevented it from having to "reinvent the wheel" every time we hit a familiar objective.
Security Hygiene
I also experimented with having Claude sanitize my secrets automatically — a massive win to ensure I don't accidentally expose credentials when I git push.
The Verdict
The result was some pretty neat reports, but it required a lot of manual guidance and correction. While AI is a powerful assistant, I think I still enjoy the manual side of pentesting the most.
Reports
I've attached the two reports I generated from my testing of the LeHack 2024 Ludus environment:
Has anyone else been playing around with CLI agents for security tasks?
#InfoSec
#Pentesting
#ClaudeCode
#AI
#CyberSecurity
#HomeLab
#Ludus
#RedTeaming