xss-testing
Fail
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: HIGHCOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
- [COMMAND_EXECUTION] (HIGH): The skill contains specific shell command templates for 'dalfox', a tool used to automate XSS discovery. An agent with terminal access could execute these commands against unintended targets, potentially violating terms of service or laws.
- [DATA_EXFILTRATION] (HIGH): The document provides functional JavaScript payloads designed for credential and cookie theft, specifically using 'document.location' to send sensitive data to an external 'attacker.com' domain.
- [REMOTE_CODE_EXECUTION] (MEDIUM): It describes methods to execute arbitrary JavaScript code using 'eval()' and 'atob()' to decode Base64 strings, which are common techniques used to bypass security filters and execute malicious payloads.
- [PROMPT_INJECTION] (LOW): While primarily focused on web security, the inclusion of multiple encoding techniques (Base64, HTML entities, URL encoding) for the purpose of bypassing filters represents a risk factor for obfuscating malicious instructions directed at the agent itself.
Recommendations
- AI detected serious security threats
Audit Metadata