The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is entirely built around executing the gog CLI tool to interact with Google services. It provides instructions for the agent to run shell commands that can perform sensitive operations, including adding Gmail forwarding addresses, managing account delegates, sharing Drive files with external emails, and performing batch deletions of messages.
[INDIRECT_PROMPT_INJECTION]: The skill has a significant attack surface for indirect prompt injection because it allows the agent to ingest and process data from untrusted external sources (emails, documents, chat messages) while maintaining powerful capabilities to act on that data.
Ingestion points: The agent can read untrusted content via gog gmail thread get (references/gmail.md), gog chat messages list (references/chat.md), gog docs cat (references/docs-slides.md), and gog sheets get (references/sheets.md).
Boundary markers: The instructions do not define any delimiters or warnings to help the agent distinguish between its system instructions and instructions potentially embedded within the data it reads.
Capability inventory: The skill provides commands for gog gmail send, gog gmail forwarding add, gog gmail delegates add, gog drive share, gog drive delete, and gog classroom coursework create across its reference files.
Sanitization: There is no mention of sanitization, validation, or escaping of the external content before it is processed or used in subsequent commands.

gog-cli