HTML Injection Testing

Warn

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: MEDIUMPROMPT_INJECTIONDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill defines triggers that encourage the agent to assist with malicious objectives like "deface web applications" and "perform HTML injection attacks."
  • [DATA_EXFILTRATION]: Includes specific payloads for exfiltrating user credentials via fake login forms and capturing session cookies through CSS-based tracking techniques targeting external domains.
  • [COMMAND_EXECUTION]: Features a Python-based fuzzing tool that uses the requests library to automate the injection of malicious payloads into web applications.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Mar 10, 2026, 06:51 AM