clinical-reports
Pass
Audited by Gen Agent Trust Hub on Nov 6, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and validate clinical reports provided by users, which represents an untrusted data source. This creates a potential surface for indirect prompt injection attacks, especially since the skill has permissions for the Bash and Write tools.
- Ingestion points: External clinical reports are read and parsed by the validation scripts in the scripts/ directory.
- Boundary markers: The provided scripts do not implement specific boundary markers or instructions to isolate untrusted input from the agent's instructions.
- Capability inventory: The skill is granted access to the Read, Write, Edit, and Bash tools.
- Sanitization: The input data is processed using regular expressions for extraction and validation, but no sanitization is applied to specifically neutralize potential instructions embedded within the clinical text.
Audit Metadata