ctf-rev
Pass
Audited by Gen Agent Trust Hub on Oct 22, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data (binaries, scripts, bytecode) which serves as an injection surface. Maliciously crafted data could attempt to subvert the agent's analysis or instructions.
- Ingestion points: Binaries, scripts, bytecode, and packed artifacts (SKILL.md).
- Boundary markers: Absent.
- Capability inventory: Reconstructing and executing scripts (SKILL.md).
- Sanitization: Absent.- [DYNAMIC_EXECUTION]: The skill instructs the agent to reconstruct logic from untrusted inputs into scripts and validate them. This involves executing code generated from potentially malicious sources.
Audit Metadata