discharge-summary
Pass
Audited by Gen Agent Trust Hub on Oct 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill is a documentation-formatting aid that does not request access to any sensitive tools, perform network operations, or execute code. It focuses on structuring provided text into a standard clinical format.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input which represents a theoretical attack surface. 1. Ingestion points: Patient admission details and hospital course notes provided by the user in response to SKILL.md instructions. 2. Boundary markers: None; the skill does not instruct the agent to use specific delimiters for user input. 3. Capability inventory: No high-risk tools or system-level permissions are requested. 4. Sanitization: Not provided; the skill assumes the clinician provides valid data and relies on standard model safety guardrails.
Audit Metadata