The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/setup.py, scripts/google_api.py, and scripts/gws_bridge.py files utilize the subprocess module to execute pip for dependency management and the gws CLI for service interaction as part of the skill's core functionality.
[PROMPT_INJECTION]: The skill processes untrusted external data from Google services, creating an indirect prompt injection surface.
Ingestion points: Gmail messages and Google Docs are fetched and read into the agent context via functions in scripts/google_api.py.
Boundary markers: The SKILL.md file contains rules explicitly requiring user confirmation before the agent performs sensitive actions like sending emails or deleting files.
Capability inventory: The skill possesses the ability to send emails, share or delete Drive files, and modify Docs and Sheets via scripts/google_api.py.
Sanitization: No specific sanitization or filtering of external content to prevent malicious instruction execution was detected.
[EXTERNAL_DOWNLOADS]: The scripts/setup.py script facilitates the installation of official Google API client libraries from the standard Python package registry.

google-workspace