The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the user to install the @googleworkspace/cli package from the NPM registry and the google-cloud-sdk via Homebrew. It also provides instructions for cloning the official repository from GitHub. These sources are well-known technology services and organizations.
[INDIRECT_PROMPT_INJECTION]: The skill creates an attack surface by reading data from external sources such as Gmail messages and Google Drive files. The documentation addresses this risk by describing the use of a 'Model Armor' safety feature, which allows the agent to sanitize or block potentially malicious content within API responses.
Ingestion points: Content retrieved from Gmail messages, Google Drive files, and Google Docs via API calls (documented in SKILL.md and references/services.md).
Boundary markers: The skill explicitly recommends using the --sanitize flag to handle untrusted content.
Capability inventory: The skill utilizes the gws CLI to perform actions like sending emails, writing files, and modifying calendars.
Sanitization: The skill supports runtime filtering of response data through integration with Google Cloud safety templates.
[COMMAND_EXECUTION]: The skill's primary function is to execute shell commands using the gws and gcloud utilities to manage Google Workspace resources. This behavior is consistent with the skill's stated administrative purpose.

gws