The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill provides a surface for indirect prompt injection by ingesting and storing untrusted data (user preferences, health conditions).
Ingestion points: Data is ingested via the set_preferences and update_preference functions in scripts/preferences_manager.py and stored in ~/.claude/nutritional_preferences.json.
Boundary markers: No boundary markers or 'ignore' instructions are present to delimit the stored data when it is retrieved.
Capability inventory: The skill is capable of reading and writing to the local file system using Python's open() and pathlib (file operations in scripts/preferences_manager.py).
Sanitization: No sanitization or validation of the input data is performed before it is saved to the JSON database.
[Data Exposure] (SAFE): Although the script accesses sensitive data related to health and allergies, this data is created and managed by the skill itself for its primary purpose. The storage location (~/.claude/) is a standard practice for CLI-based agents and no unauthorized access to system-level sensitive files (like SSH keys or AWS credentials) was detected.

nutritional-specialist