The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the gog CLI tool to interact with the Gmail API. It follows best practices by recommending the --readonly flag for non-mutating operations and --dry-run for testing writes.
[PROMPT_INJECTION]: The skill proactively addresses the risk of indirect prompt injection that can occur when processing external email data.
Ingestion points: Untrusted email message bodies and metadata retrieved via gog gmail search or get commands.
Boundary markers: The instructions explicitly mandate the use of the --wrap-untrusted flag to help the agent distinguish between tool instructions and data content.
Capability inventory: The skill can search, read, move, and send emails via the gog command-line utility.
Sanitization: It utilizes the specialized --json --wrap-untrusted output format to ensure the agent processes Google content within a secure context.

gog-gmail