The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted user input (screening results) and data from the local data/screening-tracker.json file. It lacks protective measures to prevent malicious instructions embedded within medical data from influencing the agent's logic during risk assessment or data updates.
Ingestion points: User-supplied arguments (e.g., HPV/TCT results) and historical screening data read from data/screening-tracker.json.
Boundary markers: The instructions do not define delimiters or specific warnings to ignore instructions embedded in the medical data, relying solely on keyword parsing.
Capability inventory: The skill uses Read and Write tools to modify the local file system based on its analysis of input data.
Sanitization: No input validation or sanitization logic is implemented to filter out potential injection strings before the agent processes the health records.

screening