LPW

Plumber by trade, open-source security by night. Building tools to keep AI agents honest. Creator of Pipelock.

USA Joined on Feb 8, 2026 https://github.com/luckyPipewrench/pipelock

Pinned

LPW

Feb 22

What is an agent firewall?

#security #ai #devops #opensource

6 min read

LPW

Mar 5

Guardrails deleted, now what?

#security #ai #opensource #go

4 min read

Want to connect with LPW?

Create an account to connect with LPW. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

LPW

Mar 5

Your MCP server's tool descriptions are an attack surface

#security #ai #mcp #opensource

6 min read

LPW

Mar 3

"CVE-2026-25253: WebSocket hijacking turns your AI agent into an attack tool"

#security #ai #websocket #mcp

5 min read

LPW

Mar 3

Your AI agent leaks API keys through DNS queries

#security #ai #webdev #devops

4 min read

LPW

Feb 26

Every protocol your agent speaks, scanned

#security #ai #mcp #websocket

4 min read

LPW

Feb 14

6 months until the EU AI Act hits. Here's what runtime security means.

#security #ai #devops #opensource

7 min read

LPW

Feb 13

The first AI agent espionage campaign, and what defenses actually matter

#devo #ai #opensource #security

6 min read

LPW

Feb 11

The v0.2 roadmap for Pipelock. GitHub Actions integration, MCP input scanning, smart DLP, and the path to Pipelock Pro.

#devops #ai #opensource #security

3 min read

LPW

Feb 10

Securing Claude Code with Pipelock

#devops #security #ai #opensource

4 min read

LPW

Feb 9

283 ClawHub Skills Are Leaking Your Secrets. VirusTotal Can't Fix This.

#ai #security #devops #opensource

4 min read

LPW

Feb 8

Lateral movement in multi-agent LLM systems

#devops #ai #security #opensource

5 min read

DEV Community

LPW

Badges

Writing Debut

Skills/Languages

Currently learning

Currently hacking on

Available for

Pinned

What is an agent firewall?

Guardrails deleted, now what?

Want to connect with LPW?

Your MCP server's tool descriptions are an attack surface

"CVE-2026-25253: WebSocket hijacking turns your AI agent into an attack tool"

Your AI agent leaks API keys through DNS queries

Every protocol your agent speaks, scanned

6 months until the EU AI Act hits. Here's what runtime security means.

The first AI agent espionage campaign, and what defenses actually matter

The v0.2 roadmap for Pipelock. GitHub Actions integration, MCP input scanning, smart DLP, and the path to Pipelock Pro.

Securing Claude Code with Pipelock

283 ClawHub Skills Are Leaking Your Secrets. VirusTotal Can't Fix This.

Lateral movement in multi-agent LLM systems