DEV Community

Phillip Austin Green profile picture

Phillip Austin Green

What can your agent do that you never granted? I find the holes in agent infrastructure and write up how to close them. Offensive security at Stonepath IT; memory integrity research at Botzr Research.

Education

USAF, University of Oklahoma

Work

Founder, Stonepath IT

How ECHOS discloses a vulnerability

How ECHOS discloses a vulnerability

Comments
2 min read
Your agent's tool surface is your attack surface

Your agent's tool surface is your attack surface

Comments
2 min read
Why your MCP server needs a red team

Why your MCP server needs a red team

Comments
2 min read
What an adversarial test report should look like

What an adversarial test report should look like

Comments
2 min read
Three questions a security lead asks about your AI agent

Three questions a security lead asks about your AI agent

Comments
2 min read
The SOC 2 gap your AI agent falls through

The SOC 2 gap your AI agent falls through

Comments
4 min read
The attack surface most teams ignore

The attack surface most teams ignore

Comments
2 min read
Indirect prompt injection is not a user input problem

Indirect prompt injection is not a user input problem

Comments
4 min read
State-Drift

State-Drift

Comments
4 min read
Six silent memory failures

Six silent memory failures

Comments
9 min read
Find the tool servers you left open

Find the tool servers you left open

Comments
3 min read
Can memory search say 'not here'?

Can memory search say 'not here'?

Comments
12 min read
Individuation Architecture: agent identity as development, not configuration

Individuation Architecture: agent identity as development, not configuration

Comments
2 min read
Fixing a wrong number in one place does not fix it

Fixing a wrong number in one place does not fix it

Comments
3 min read
loading...