The Agent Skills Directory

[SAFE]: The skill instructions and associated files define legitimate logic for medical appointment management, including registration, querying, modification, and cancellation. No malicious intent, hidden behaviors, or security bypasses were detected.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input to extract appointment details (names, dates, IDs). It provides specific guidelines in rules.md and solver_rules.md for date normalization and grounding parameters in prior observations, which serves as a safeguard against malicious or invented data. The included validator.py further enforces these constraints through programmatic checks.

Appointment lifecycle disambiguation