The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes content from Google Drive and external URLs which could contain malicious instructions for the AI agent.
Ingestion points: download_file_to_storage, search_files, and upload_file_from_storage (via source_file_url).
Boundary markers: No explicit boundary markers or instructions to ignore embedded prompts are provided in the skill instructions.
Capability inventory: The skill has extensive capabilities including file movement, sharing, permanent deletion, and trashing files across My Drive and shared drives.
Sanitization: No sanitization or filtering logic for external content is described.
[EXTERNAL_DOWNLOADS]: The upload_file_from_storage action includes a source_file_url parameter that allows the agent to fetch content from arbitrary public URLs. This provides a mechanism for network requests to external domains, which can be used to ingest untrusted data into the Google Drive environment.

google-drive