The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill enables the agent to ingest content from external sources that are not controlled by the agent platform, such as Google Drive file contents, comments, and metadata.
Ingestion points: Untrusted data enters the context through methods like files.get (file content), files.list (filenames/metadata), comments.list, comments.get, and replies.list (user comments).
Boundary markers: The skill instructions do not specify any delimiters or safety markers to differentiate between system instructions and data fetched from Drive.
Capability inventory: The skill has significant capabilities including file creation/deletion, shared drive management, and permission modification (permissions.create, permissions.update).
Sanitization: No sanitization or validation of the ingested external content is mentioned.
[COMMAND_EXECUTION]: The skill relies on the execution of the gws command-line utility to perform its operations.
Evidence: The skill defines the use of gws drive <resource> <method> [flags] for all operations and uses gws schema for introspection.

gws-drive