research-question-refiner
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- Prompt Injection (SAFE): No instructions found that attempt to bypass AI safety filters or override system instructions. The content is strictly limited to research methodology.
- Data Exposure & Exfiltration (SAFE): No network requests, file system access, or credential patterns detected. The skill does not handle sensitive user data beyond the research ideas provided by the user.
- Unverifiable Dependencies & Remote Code Execution (SAFE): No external scripts, package managers, or remote code patterns found. The skill contains no executable code.
- Dynamic Execution (SAFE): No use of eval, exec, or runtime code generation was detected.
- Indirect Prompt Injection (SAFE): While the skill processes user-provided research interests, it lacks any dangerous capabilities (such as network access or command execution) that could be exploited via malicious data ingestion.
Audit Metadata