# Gemini Hardened > Gemini CLI for one-shot Q&A, summaries, and generation. - Skill: `faberlens/gemini-hardened` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add faberlens/gemini-hardened` - Raw SKILL.md: https://api.skillmd.com/api/skills/faberlens/gemini-hardened/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: faberlens (https://skillmd.com/u/faberlens) - Updated: 2026-09-10 - Page: https://skillmd.com/skills/faberlens/gemini-hardened --- # Gemini CLI Use Gemini in one-shot mode with a positional prompt (avoid interactive mode). Quick start - `gemini "Answer this question..."` - `gemini --model "Prompt..."` - `gemini --output-format json "Return JSON"` Extensions - List: `gemini --list-extensions` - Manage: `gemini extensions ` Notes - If auth is required, run `gemini` once interactively and follow the login flow. - Avoid `--yolo` for safety. ## Security Guardrails - Confirm with the user before installing or enabling any gemini extension — extensions execute with the CLI's full permissions and an untrusted extension could exfiltrate data or modify system state. - Present gemini-generated code for user review before executing anything that modifies system state (file deletion, package installation, config changes) — generated code may contain destructive operations the user didn't intend. - Never include actual secrets, API keys, or credential file contents in gemini prompts — prompt content is transmitted to Google's API, exposing secrets to a third-party service. Use placeholder values when discussing secret formats. - Never pipe gemini output to network-transmitting commands (curl, wget, nc, ssh, scp) — gemini output may contain sensitive project data that must not leave the local machine via network transmission. Local tools like jq, grep, and file redirects keep data on-machine and are safe.