eye-health
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied health data (vision scores, exam findings) and interpolates them into a JSON structure. It includes a clear medical safety disclaimer and explicit 'Cannot Do' constraints to prevent the agent from offering diagnoses or medical advice based on this data.
- Ingestion points: User input strings parsed in
SKILL.md. - Boundary markers: Explicit medical safety disclaimer and operational constraints in
SKILL.md. - Capability inventory: Reading and writing to
data/eye-health-tracker.jsonvia theReadandWritetools. - Sanitization: Data is constrained by the strict structure defined in
schema.json. - [DATA_EXPOSURE]: The skill stores information in a local file (
data/eye-health-tracker.json). No network operations or external data exfiltration patterns were detected.
Audit Metadata