The Agent Skills Directory

[SAFE]: The skill wraps the official Google Gemini CLI (gemini), which is a legitimate tool provided by a trusted technology vendor.
[SAFE]: External dependencies are restricted to the official @google/gemini-cli package available on the NPM registry and gemini-cli via Homebrew.
[SAFE]: The skill implements a robust safety framework for CLI execution, requiring explicit user approval for any file modifications and defaulting to the most restrictive --approval-mode default.
[SAFE]: Command construction uses standard shell practices, including the use of the -p flag for headless operation and wrapping user prompts in quotes to prevent accidental command splitting.
[SAFE]: Preflight checks using command -v gemini ensure the CLI is correctly installed and authenticated before the agent attempts to run any commands.

gemini