easy-query-expert
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No malicious instructions or behavior overrides were detected. The content is strictly educational and instructional.\n- Data Exposure & Exfiltration (SAFE): No hardcoded credentials or unauthorized network requests. External references point to legitimate repositories and documentation sites.\n- Obfuscation (SAFE): No encoded or hidden content (Base64, zero-width chars, etc.) was identified across the skill files.\n- Unverifiable Dependencies & Remote Code Execution (SAFE): No remote script downloads or piped shell executions are present. References to standard Java build tools like Maven are appropriate for the skill's context.\n- Privilege Escalation (SAFE): No commands requesting elevated system permissions or modifying security policies were found.\n- Persistence Mechanisms (SAFE): No mechanisms for maintaining unauthorized access (cron, startup scripts) were detected.\n- Metadata Poisoning (SAFE): The skill metadata correctly describes its purpose and origin without deceptive instructions.\n- Indirect Prompt Injection (SAFE): The skill provides documentation templates and does not interpolate untrusted external data into dangerous execution sinks.\n- Time-Delayed / Conditional Attacks (SAFE): No logic gating malicious behavior based on dates or environmental triggers was found.\n- Dynamic Execution (SAFE): The proxy generation described is a standard compile-time APT process, posing no runtime risk.
Audit Metadata