Resources · Free guide summary
AI Agent Awareness Guide
What active AI agents can do on a Windows computer, what access really means, and how to respond when you don’t recognize the activity.
This page summarizes the guide. The full PDF, with worksheets and examples, is free from our Resources page.
What is an AI agent?
Unlike a chatbot that only answers, an agent can take actions on your behalf: open files, use a browser, call tools, connect to services, or run a series of steps.
Common types of access
Files, browser, network, applications, terminal and scripts, and connected tools such as plugins and MCP servers. Access isn’t automatically bad. The question is whether you recognize the agent and meant to authorize it.
What to pay attention to
Do I recognize the agent? Does the activity match what I asked for? Is the access reasonable for its purpose? Where is it connecting? Would I be comfortable explaining it to someone else?
What to do when you don’t recognize an agent
Identify it, review what it’s doing, compare it with your intent, stop it if you didn’t authorize it, and record what happened.
Get the full guide
Want this on your PCs automatically? See how PERCARIS works or compare plans.
PERCARIS recognizes known AI tools. It won’t flag every AI agent.
