The Agent Skills Directory

[SAFE]: The skill and its bundled scripts are restricted to local file operations using only the Python standard library. No external dependencies, network libraries, or third-party packages are used.
[SAFE]: Strict path validation in scripts/_common.py prevents access to non-local paths, network shares, and symlinks, ensuring the tool only interacts with intended local files.
[SAFE]: Input JSON files are subject to rigorous structural checks, including schema validation, recursion depth limits, and node count bounds to prevent resource exhaustion and malformed data processing.
[SAFE]: The skill explicitly prohibits clinical decision-making, diagnosis, or medication recommendations, maintaining a clear boundary as a documentation-only aid through both natural language instructions and deterministic script logic.
[SAFE]: A specialized privacy check identifies and flags prohibited direct-identifier field names (e.g., SSN, MRN) to help prevent the accidental inclusion of identifiable patient data in the documentation packages.
[SAFE]: Atomic file write operations with restrictive permissions (0o600) and directory creation with 0o700 ensure that output data remains local and accessible only to the executing user.

treatment-plans