aegis-file-handling
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection because its primary purpose is to process untrusted data (source code) within an audited repository.
- Ingestion points: The instructions direct the agent to analyze files within a target repository using search tools and language-specific sink detection.
- Boundary markers: The skill includes explicit warnings that reproduction procedures should only be executed against authorized lab environments.
- Capability inventory: The skill recommends using shell-based searching (
ripgrep) and network-based verification (curl) for security auditing. - Sanitization: The methodology requires mapping findings to specific lines of code and manual verification to ensure accuracy and prevent accidental execution.
- [EXTERNAL_DOWNLOADS]: The skill references documentation and educational materials from well-known and trusted security organizations, including OWASP, MITRE (CWE), and PortSwigger's Web Security Academy, as well as official Python documentation. These references are used for informational purposes and do not facilitate the execution of untrusted external code.
Audit Metadata