The Agent Skills Directory

[SAFE]: The skill is designed for administrative and productivity tasks within Google Workspace. It performs documented operations and does not contain hidden malicious logic or obfuscated code.
[EXTERNAL_DOWNLOADS]: The skill requires several official Google Ruby gems (google-apis-docs_v1, google-apis-drive_v3, google-apis-sheets_v4, google-apis-calendar_v3, google-apis-people_v1, and googleauth) for API interaction. These are trusted dependencies provided by Google.
[COMMAND_EXECUTION]: The skill includes two Ruby scripts, docs_manager.rb and drive_manager.rb, which provide a command-line interface for the agent to interact with Google APIs. These scripts are used for legitimate document and file operations.
[PROMPT_INJECTION]: The skill has an indirect prompt injection surface because it allows the agent to read content from external Google Docs which could contain instructions intended to influence the agent's behavior.
Ingestion points: The read operation in scripts/docs_manager.rb extracts full text content from Google Docs.
Boundary markers: The current implementation does not include specific delimiters or warnings for the agent to ignore instructions embedded in the document text.
Capability inventory: The skill can perform file system operations (via uploads/downloads), network requests (to Google APIs), and modify document content and permissions.
Sanitization: No sanitization or filtering is applied to the content read from external documents.
[DATA_EXFILTRATION]: No unauthorized data exfiltration patterns were detected. All network communication is conducted with official Google API endpoints (googleapis.com) using valid OAuth 2.0 tokens.

google-docs