The Agent Skills Directory

Prompt Injection (HIGH): The skill is highly vulnerable to Indirect Prompt Injection. It retrieves untrusted content from external sources and provides tools to perform impactful actions based on that content.
Ingestion points: The skill reads external data via gog gmail search, gog drive search, gog sheets get, and gog docs cat.
Boundary markers: There are no specified boundary markers or delimiters to isolate untrusted content from agent instructions.
Capability inventory: The skill allows for high-impact write operations including gog gmail send (email exfiltration/spam), gog calendar create (scheduling), and gog sheets update (data manipulation).
Sanitization: No sanitization or validation mechanisms are described for the data retrieved from Google APIs before it is processed by the agent.
External Downloads (MEDIUM): The installation process relies on a third-party Homebrew tap (steipete/tap/gogcli). While the developer may be known in other contexts, this repository is not in the predefined list of trusted sources, meaning the binary integrity cannot be verified by this analysis.
Command Execution (LOW): The skill is designed to execute a local binary (gog) to perform its tasks. While inherent to the skill's design, this grants the agent the ability to execute shell commands with the user's Google Workspace permissions.

gog