staging-subdomain-hunt
Fail
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: HIGHCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands using tools like curl and httpx to perform network reconnaissance and automated vulnerability scanning against target domains.
- [DATA_EXFILTRATION]: The skill is designed to search for and extract sensitive information from remote hosts, including environment files (.env) that often contain credentials, and internal user lists via WordPress REST APIs.
- [PROMPT_INJECTION]: The instructions include aggressive directives such as 'Exploit the staging security gap' and 'Install takeover,' which encourage the agent to move from passive reconnaissance to active unauthorized exploitation of discovered vulnerabilities.
- [DATA_EXPOSURE]: The skill probes for and potentially displays sensitive data such as PHP configuration (info.php) and configuration setup pages, which can lead to the exposure of internal server details.
Recommendations
- AI detected serious security threats
Audit Metadata