The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill requires the user to provide a path to a sensitive OAuth client secret file via the command gog auth credentials /path/to/client_secret.json. This exposes highly sensitive authentication material to the agent environment.
[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install a third-party binary using a Homebrew tap (steipete/tap/gogcli). This introduces a dependency on an external developer's repository which is not managed by the platform.
[COMMAND_EXECUTION]: The skill relies on executing the gog binary for all its operations, which can read, write, and delete data across a user's Google Workspace (Gmail, Drive, Calendar, etc.).
[DATA_EXFILTRATION]: The tool includes capabilities to send emails (gog gmail send) and export documents (gog docs export), which can be used to move sensitive information from Google Workspace to external recipients or local temporary directories.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests untrusted data from external sources and has high-privilege capabilities.
Ingestion points: Commands like gog gmail messages search and gog docs cat bring untrusted data from emails and documents into the agent's context.
Boundary markers: No delimiters or instructions to ignore embedded commands are specified when reading external content.
Capability inventory: The skill has the ability to send emails, create calendar events, and export files to the local system.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from Google Workspace before it is processed by the agent.

gog