PipeLab

Security tools for AI agents.
Infrastructure that tests them.

AI Agent Security

Open-source tools that stop AI agents from leaking your secrets. Pipelock is the flagship — a scanning proxy for Claude Code, Cursor, and other coding agents.

Infrastructure

On-prem Kubernetes, network hardening, GPU compute, NVR systems, automated backups. Built and managed for people who want control over their stack.

Security Research

Published research on AI agent attack vectors and OWASP contributions. If agents can break it, the goal is to find out how — and ship the fix.

Plumbing

Both kinds of infrastructure. PipeLab started in the trades and still handles pipes alongside packets.

Pipelock

Open-source agent firewall

brew install luckyPipewrench/tap/pipelock
go install github.com/luckyPipewrench/pipelock/cmd/pipelock@latest
docker pull ghcr.io/luckypipewrench/pipelock:0.2.6