Cross-Site Scripting and HTML Injection Testing

Warn

Audited by Gen Agent Trust Hub on Nov 6, 2026

Risk Level: MEDIUMDATA_EXFILTRATIONOBFUSCATIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill provides explicit JavaScript payloads designed to exfiltrate session cookies and capture keystrokes, specifically directing the stolen data to a remote host (e.g., attacker.com).
  • [OBFUSCATION]: Documents and provides examples of filter bypass techniques using Base64 encoding (atob), HTML entity encoding, and hex encoding to hide malicious payloads from basic security scanners.
  • [DYNAMIC_EXECUTION]: Instructs the agent to utilize dangerous JavaScript functions such as eval(), Function(), setTimeout(), and innerHTML to process and execute injected code during exploitation.
  • [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted data from web applications (URLs, headers, form fields), providing a significant ingestion point for indirect prompt injection where malicious content on a target site could influence the agent's behavior.
  • Ingestion points: Target web application URL, user input fields, URL fragments, HTTP headers (defined in SKILL.md).
  • Boundary markers: None present in the instructions to delimit untrusted web content.
  • Capability inventory: Not applicable to the skill itself as it provides payloads for the user to execute via browser or proxy tools.
  • Sanitization: No sanitization or escaping mechanisms are provided for the ingested external content.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 6, 2026, 03:30 PM