The Agent Skills Directory

[OBFUSCATION]: Several reference files, including reference/dom-xss-advanced.md and reference/scenarios/dom-vulnerabilities/waf-filter-bypass.md, provide examples of Base64 encoded JavaScript payloads (e.g., eval(atob('YWxlcnQoMSk='))) to demonstrate how attackers bypass Web Application Firewalls (WAFs).
[REMOTE_CODE_EXECUTION]: The documentation in reference/scenarios/prototype-pollution/server-side-pollution.md details methods for achieving remote code execution (RCE) via Node.js prototype pollution, providing payloads that abuse execArgv and including a reverse shell connection string as a demonstration of high-impact exploitation.
[DATA_EXFILTRATION]: Multiple files, such as reference/scenarios/xss/data-exfiltration.md and reference/scenarios/xss/cookie-theft.md, include JavaScript and Python scripts for harvesting cookies, session storage, and page content, exfiltrating the data to external endpoints (e.g., Burp Collaborator) as part of vulnerability impact demonstration.
[PERSISTENCE]: In reference/scenarios/xss/beef-integration.md, the skill describes the use of the BeEF (Browser Exploitation Framework) to hook victim browsers and maintain persistent control across sessions.
[INDIRECT_PROMPT_INJECTION]: The skill covers extensive techniques for identifying and exploiting vulnerabilities where untrusted data is processed by an application, which represents the primary attack surface for indirect prompt injection in agent-enabled web environments.

client-side