The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill retrieves and processes untrusted data from Gmail, Google Drive, Docs, and Sheets, creating a surface for indirect prompt injection attacks.
Ingestion points: Data is ingested via commands like gog gmail search, gog drive search, gog sheets get, and gog docs cat as described in SKILL.md.
Boundary markers: The instructions do not define specific delimiters or "ignore instructions" markers when handling content retrieved from these external sources.
Capability inventory: The skill has the capability to perform high-impact actions including sending emails (gog gmail send), modifying spreadsheets (gog sheets update), and creating calendar events (gog calendar create).
Sanitization: There is no evidence of content sanitization or validation before the retrieved data is processed or used in further actions.
[COMMAND_EXECUTION]: The skill operates by executing shell commands via the gog CLI tool. This involves passing user-supplied or externally-sourced arguments to the command line, which requires careful handling of parameters.
[CREDENTIALS_UNSAFE]: The skill's setup instructions reference the use of sensitive credential files, specifically mentioning a path to a client_secret.json file for OAuth authentication. While standard for this tool's operation, accessing such files is a sensitive action that requires user oversight.

gog