An agentic app is software that can perform multi-step actions on a user’s behalf instead of only answering a prompt. In practice, it may read files, browse folders, call tools, run commands, or move data between services with some level of automation.
This matters in cyber security because every extra capability expands the attack surface. If an attacker can influence the app’s inputs, they may steer it into opening sensitive files, sending data to the wrong place, or using credentials stored on the endpoint. Security bugs often involve weak sandboxing, poor path validation, overbroad permissions, or unsafe tool access. Defenses include least-privilege file scopes, strong isolation, short-lived credentials, user confirmation for risky actions, and logging of tool use and file access.