The Agent Skills Directory

[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface by ingesting untrusted data that is later processed into reports and assessments.
Ingestion points: User input for family member details, medical conditions, and clinical notes (processed in SKILL.md).
Boundary markers: The instructions utilize markdown headers and JSON blocks to structure data, but they lack explicit instructions to the agent to disregard commands embedded within the health data fields.
Capability inventory: The skill is configured with Read and Write tool access to manage the data/family-health-tracker.json file (SKILL.md).
Sanitization: No text sanitization or input validation is defined to prevent the agent from interpreting malicious instructions hidden within user-provided medical notes or disease names.

family