radiology
Pass
Audited by Gen Agent Trust Hub on Jul 31, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides structured guidance for manuscript preparation and venue selection for the Radiology journal, focusing on reporting standards like STARD and CLAIM.- [DATA_EXFILTRATION]: The skill instructs the agent to read internal resource files at
../../resources/source-basis.mdand../../resources/official-source-map.md. These relative paths point to supplemental documentation within the skill's resource directory rather than sensitive system files.- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface as it processes untrusted research study descriptions provided by users. Mandatory evidence chain: 1. Ingestion points: Author research descriptions in SKILL.md body. 2. Boundary markers: Absent. 3. Capability inventory: None (no tools or command execution used). 4. Sanitization: Absent. The risk is minimized by the lack of exploitable capabilities.
Audit Metadata