The Agent Skills Directory

[SAFE]: The skill serves as a technical reference for the gws CLI tool, specifically adapted for proxy-authenticated headless environments. It provides guidance that limits the agent's exposure to credentials by using a transparent proxy model.
[EXTERNAL_DOWNLOADS]: The skill references the official GitHub repository for the Google Workspace CLI (github.com/googleworkspace/cli) for community feedback and issue tracking. This repository belongs to a well-known organization and is used for project maintenance.
[COMMAND_EXECUTION]: The skill provides templates for executing local commands using the gws binary. The authentication model is designed to be transparent, removing the need for the agent to manage or view sensitive OAuth tokens or API keys, which reduces the risk of credential exposure during command invocation.
[PROMPT_INJECTION]: The skill describes a surface for indirect prompt injection as it interacts with external Google Workspace data (emails, files).
Ingestion points: Google Workspace API responses processed via commands such as gws gmail +read or gws drive files list (SKILL.md).
Boundary markers: None specified for delimiting raw command output in the prompt.
Capability inventory: Shell execution of the gws binary with the ability to read, write, and send data (e.g., gws gmail +send) across Workspace services.
Sanitization: The skill explicitly instructs the agent to use the --sanitize flag with Model Armor for PII and content safety screening, which mitigates the risk of processing malicious embedded instructions.

gws-shared