The Agent Skills Directory

[SAFE]: The skill provides instructional content for security researchers (red teaming) focused on client-side web vulnerabilities. While it includes commands for data exfiltration and reconnaissance (e.g., curl, grep), these are directed at target web applications for vulnerability identification purposes rather than the agent's local environment or the user's sensitive data.
[COMMAND_EXECUTION]: The skill includes shell commands using curl and grep to analyze target websites for specific DOM signals and configurations. These commands are standard reconnaissance techniques for the stated purpose of the skill.
[DATA_EXFILTRATION]: The skill contains proof-of-concept JavaScript and CSS code designed to demonstrate data exfiltration from a target web application to an Out-of-Band (OOB) listener (e.g., https://OOB.example/c). These are placeholders for security testing and do not target the user's local credentials or system files.
[REMOTE_CODE_EXECUTION]: The skill discusses techniques like Service Worker abuse and DOM-XSS, which involve executing code within the context of a target web application's origin. It does not contain patterns for executing arbitrary remote code on the local system.

hunt-dom