The Agent Skills Directory

[SAFE]: The skill implements logic through a Python script (scripts/calculate_exposure.py) that uses only the standard library. No external or untrusted packages are required.\n- [SAFE]: Data ingestion is limited to local JSON files produced by upstream skills. The script performs structured parsing and type conversion (e.g., converting strings to integers for scoring), which effectively mitigates indirect prompt injection risks by not passing raw external text into sensitive contexts.\n- [SAFE]: The skill handles sensitive configuration (API keys) correctly by advising the user to set them as environment variables rather than hardcoding them in scripts or configuration files.\n- [SAFE]: Output is restricted to local file system writes in a user-defined reports directory, following standard developer tool patterns.

exposure-coach