web-research
Pass
Audited by Gen Agent Trust Hub on Feb 25, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: No malicious behavior detected. The skill logic is entirely contained within structured markdown instructions and utilizes built-in agent capabilities for research.
- [PROMPT_INJECTION]: No prompt injection or safety bypass patterns were identified. The skill maintains a focus on factual synthesis and research methodology.
- [DATA_EXFILTRATION]: No sensitive data access or credential leakage patterns were found. The skill operates on public web information.
- [COMMAND_EXECUTION]: No unauthorized command execution or privilege escalation attempts were detected.
- [SAFE]: Potential indirect prompt injection surfaces are addressed through rigorous verification and source quality hierarchies. Ingestion points: External web pages via WebSearch. Boundary markers: Cited URLs and mandatory source evaluation notes. Capability inventory: Parallel sub-agent creation via the Task tool. Sanitization: Implementation of a credibility hierarchy (Official docs over practitioners) and instructions to explicitly flag contested information.
Audit Metadata