Lightlines

Security skills for AI agents

These Skills look at software the way an attacker or an auditor would. The biggest groups review code for injection and broken access rules, probe systems in penetration tests and audit cloud accounts against best practices; smaller sets cover sign-in, malware analysis and forensics.
Sort by

794 Skills

Showing Skills in Security.

A Skill only gives the agent instructions and tools; it carries no permission of its own. Penetration testing and reverse engineering Skills are written for systems you own or are authorised to test, and the agent works with the access you give it.

A review reads the code or configuration and points out weaknesses. A penetration test runs the system and tries to exploit them, which shows whether a weakness can actually be used.

Tests check that the code does what it should; security Skills check what it should never do, such as letting one user read another's data. The two catch different mistakes.