The Agent Skills Directory

[OBFUSCATION]: Multiple files utilize Base64 encoding to obfuscate payloads, primarily to demonstrate WAF and filter bypass techniques.
scripts/blind_xss_tester.py contains a Base64 string cG9zdChtZXNzYWdlLmNyZWF0ZUVsZW1lbnQoImltZyIpKTsK (decodes to post(message.createElement("img"));) inside an eval(atob()) block.
assets/svg_xss_payloads.txt includes a Base64 encoded data URI: data:image/svg+xml;base64,PHN2ZyBvbmxvYWQ9ImFsZXJ0KCdYU1MnKSI+PC9zdmc+ (decodes to <svg onload="alert('XSS')">).
assets/waf_bypass_payloads.txt contains <script>eval(atob('YWxlcnQoMSk='))</script> (decodes to alert(1)).
[COMMAND_EXECUTION]: The skill relies on executing external command-line utilities to perform its primary functions.
scripts/xss_full_scan.py uses subprocess.run() to invoke tools such as dalfox, xsstrike, and xspear.
The scan_with_dalfox and scan_with_xsstrike methods in the same file pass user-controlled target URLs directly into shell commands.
[EXTERNAL_DOWNLOADS]: The skill's documentation (SKILL.md) and reference guides recommend installing several third-party security tools from GitHub and official package registries.
Recommends installing XSStrike via pip, Dalfox via Go, and XSpear via RubyGems.
Includes links to external resources on GitHub (e.g., s0md3v/XSStrike, hahwul/dalfox).
[DYNAMIC_EXECUTION]: The skill generates and tests payloads that utilize dynamic JavaScript execution functions to evade detection.
Numerous payloads in the assets/ directory use eval(), Function(), setTimeout(), and atob() to execute arbitrary code string segments.
[DATA_EXFILTRATION]: The skill provides automated mechanisms for Out-of-band Application Security Testing (OAST) which exfiltrate data from target systems.
scripts/blind_xss_tester.py generates payloads that use fetch() or image requests to send data (including document.cookie) to a user-provided callback URL.
Multiple asset files (e.g., assets/blind_xss_payloads.txt) contain hardcoded examples of exfiltration payloads.
The skill contains references to known malicious domains such as evilcdn.com (e.g., https://evilcdn.com/lib.js and https://evilcdn.com/payload.js) which are flagged by automated scanners, although they are documented as examples for testing purposes.

exploit-xss