Security · Data Protection
Which sensitive files agents are reading into context.
The problem
Agents quietly open secrets, credentials, and config files and pull them into model context. Nobody knows which sensitive paths were touched, by which agent, or how big the blast radius is.
The value
A ranked map of every sensitive path an agent opened — by category, by owner, by blast radius. Each row is a containment decision, so you can scope retrieval and add deny rules before a secret leaves the building.
6
sensitive paths opened
2
high-severity categories
691
file writes observed