Image: credrelay.com · Letting AI loose in My Homelab
Worked Proof shown
Claude Code with 150 offensive security tools on a home lab
A write-up of letting Claude Code attack the author's own home lab.
—
Credits
Earliest dated source:
The full story
An offensive security engineer with OSCP and CRTO certifications pointed Claude Code, armed with more than 150 offensive-security MCP tools, at his own home lab. It got root on an Ubuntu VM in about 30 seconds, then falsely claimed to have escaped a Docker container. He also had it build nine AI agent skills in one night. The newsletter issue is a frank look at both the power and the unreliability.
How it was done
Tools
Themes
Loading themes…
Sources
From the maker's own write-up or code. We summarise what they report; we haven't tested it ourselves.
- Hacker News: I let Claude Code with 150 offensive security MCP tools loose on my homelab
"I Let Claude Code Hack My Homelab": the full account.
- Hacker News discussion (2 points)HN submitter: jeffafPosted