AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them.

Description and use

<p>A coding agent with terminal access can install a package, run a script, start a child process and connect to a server. Add skills, hooks and MCP tools, and a request like “fix the build” can become a chain of actions with access to your files and credentials.</p> <p>Those actions need security decisions. Which command may run? Which tool call did I actually approve? What should happen when an agent touches a sensitive path?</p> <p>I've spent eight months building <strong>AEGIS: an open-source security and protection app for local AI agents</strong>. “Antivirus for AI agents” is how I descr

Problem

<p>A coding agent with terminal access can install a package, run a script, start a child process and connect to a server. Add skills, hooks and MCP tools, and a request like “fix the build” can become a chain of actions with access to your files and credentials.</p> <p>Those actions need security decisions. Which command may run? Which tool call did I actually approve? What should happen when an agent touches a sensitive path?</p> <p>I've spent eight months building <strong>AEGIS: an open-source security and protection app for local AI agents</strong>. “Antivirus for AI agents” is how I descr

Outcome

<p>A coding agent with terminal access can install a package, run a script, start a child process and connect to a server. Add skills, hooks and MCP tools, and a request like “fix the build” can become a chain of actions with access to your files and credentials.</p> <p>Those actions need security decisions. Which command may run? Which tool call did I actually approve? What should happen when an agent touches a sensitive path?</p> <p>I've spent eight months building <strong>AEGIS: an open-source security and protection app for local AI agents</strong>. “Antivirus for AI agents” is how I descr

Evidence

AI Agents Have the Keys to Your Computer. I Spent 8 Months Building an Antivirus for Them.

Source