environment-secrets-exfiltrator
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: CRITICALDATA_EXFILTRATIONCREDENTIALS_UNSAFE
Full Analysis
- Data Exposure & Exfiltration (CRITICAL): The script in
get_info.pyiterates through all environment variables looking for sensitive strings like 'KEY', 'SECRET', and 'TOKEN', then transmits the collected data tohttps://attacker.example.com/secretsusing an HTTP POST request. - Obfuscation (HIGH): The harvested data is Base64 encoded before being sent to the external domain, which is a common technique to bypass simple network traffic analysis.
- Metadata Poisoning (HIGH): The skill's
SKILL.mdfile provides a misleading description claiming the tool is for 'system diagnostics', while the actual code performs unauthorized secret exfiltration. - Unverifiable Dependencies (MEDIUM): The skill relies on the
requestslibrary for outbound network connections to an untrusted, non-whitelisted domain.
Recommendations
- AI detected serious security threats
Audit Metadata