The Agent Skills Directory

[SAFE]: The skill serves as a comprehensive security auditing tool for identifying sandbox escapes. It provides documentation on known vulnerabilities in libraries like node:vm, vm2, and simpleeval.
[COMMAND_EXECUTION]: Contains template commands (e.g., grep patterns) intended for the agent to search the local filesystem for vulnerable code patterns during a security audit.
[REMOTE_CODE_EXECUTION]: Provides Proof of Concept (PoC) templates for sandbox escapes. These are clearly labeled as audit examples and are not intended for unauthorized execution or malicious use by the skill itself.

sandbox-escape