Getting Started with Gemini CLI ✨

Experience the power of Gemini directly from your terminal with Gemini CLI!

For developers, the command line interface (CLI) isn’t just a tool; it’s home. The terminal’s efficiency, ubiquity and portability make it the go-to utility for getting work done. And as developers’ reliance on the terminal endures, so does the demand for integrated AI assistance. — Gemini CLI launch blog

Gemini CLI is now live! 🚀🎉

Let’s explore and answer all the important questions like “What is it?”, “How do I get started?”, and “What can it do?”.

Gemini CLI start up screen.

What is Gemini CLI?

Gemini CLI is an open-source AI agent that brings the power of Gemini directly into your terminal. It provides lightweight access to Gemini, giving you the most direct path from prompt to model. While it excels at coding, Gemini CLI can do much more. It’s a versatile, local utility you can use for a wide range of tasks, from content generation and problem solving to deep research and task management.

With the Gemini CLI you can:

• Query and edit large codebases in and beyond Gemini’s 1M token context window.
• Generate new apps from PDFs or sketches, using Gemini’s multimodal capabilities.
• Automate operational tasks, like querying pull requests or handling complex rebases.
• Use tools and MCP servers to connect new capabilities, including media generation with Imagen, Veo or Lyria.
• Ground your queries with the Google Search tool, built in to Gemini.

How to get started?

Install NodeJS 18+, then run:

npm install -g @google/gemini-cli

You now have Gemini CLI installed, run it using the gemini command:

gemini

When prompted, just sign in with your Google account! 🔑

Why should you care?

Developers live in the command line, integrated AI assistance is a natural choice to improve your velocity and access to powerful tools. ✨

Unmatched Free Tier

To use Gemini CLI free-of-charge, you can simply login with a personal Google account and get a free Gemini Code Assist license. That free license gets you access to Gemini 2.5 Pro and its massive 1 million token context window. To ensure you rarely, if ever, hit a limit during this preview, Gemini CLI offers the industry’s largest allowance: 60 model requests per minute and 1,000 requests per day at no charge.

5 Features of Gemini CLI You Need to Know

Here are a few awesome Gemini CLI features that you should be aware…

1. GEMINI.md context files

Pass custom context (persona, build processes, style guide rules, etc.) to Gemini CLI using GEMINI.md context files.

Example — Have Gemini CLI automatically lint/format all Python code 🐍

# Example Python dev GEMINI.md file
...

## Mandatory Tooling
To ensure all Python code adheres to these standards, the following commands **must** be run before committing any `.py` files. These commands will automatically fix many common issues and flag any that require manual intervention.

When creating or modifying any `.py` Python files, you **must** run the following commands from the root of the project:

1.  **Check and fix linting issues:**
    ```bash
    uvx ruff@latest check --fix .
    ```
2.  **Format the code:**
    ```bash
    uvx ruff@latest format .
    ```

You can view the combined context from your context files by running /memory show in the Gemini CLI.

GEMINI.md context shown using /memory show command.

To put it to the test, ask Gemini CLI to create a main.py file with the following code:

import sys,os
import asyncio
import requests

def hello_world(name:str='World'):
    print(f'Hello, {name}!')
    unused_str = 'random string'

if __name__=='__main__':




    hello_world()

There are several poorly formatted pieces of the code that the ruff commands added to the GEMINI.md file should be able to fix (spacing, double quotes instead of single quotes, etc. ).

Gemini CLI creating a properly formatted code file using GEMINI.md context.

The code is now properly formatted. ✅

2. Built-in tools (/tools)

Gemini CLI comes with many powerful pre-built tools (like Google Search) that can be viewed by running the /tools command.

/tools — Lists built-in Gemini CLI tools.

Gemini CLI is successfully able to ground it’s responses using the Google Search tool! 🏀

3. MCP support (/mcp)

Gemini CLI is launching with Model Context Protocol (MCP) support already integrated! 🔌

To configure MCP servers with Gemini CLI, create a .gemini/settings.json file with the mcpServers field.

To view configured MCP servers and tools from within the Gemini CLI run /mcp.

Example — GitHub MCP Server Configuration

Below is an example of a .gemini/settings.json used to configure the GitHub MCP server:

{
  "mcpServers": {
    "github": {
      "command": "npx",
      "args": [
        "-y",
         "@modelcontextprotocol/server-github"
      ],
      "env": {
       "GITHUB_PERSONAL_ACCESS_TOKEN": "${GITHUB_PERSONAL_ACCESS_TOKEN}"
      }
    }

Gemini CLI will automatically replace the ${GITHUB_PERSONAL_ACCESS_TOKEN} placeholder from the settings.json file with the GITHUB_PERSONAL_ACCESS_TOKEN environment variable (which you must first set using a command like export GITHUB_PERSONAL_ACCESS_TOKEN=<YOUR_PAT>).

/mcp — List all connected MCP servers.

4. Shell mode (terminal passthrough)

The ! prefix lets you interact with your system’s shell directly from within Gemini CLI.

Shell mode activated using “!”.

5. Gemini CLI flags

The Gemini CLI provides flexible customization through the support of many CLI flags.

You can view all supported CLI flags by running the--help command:

gemini --help

Notable flags:

• --yolo : YOLO mode, automatically accepts all actions withour prompting you for allowlisting.
• --model: Change the underlying Gemini model (default: gemini-2.5-pro)
• --prompt: Non-interactive mode, run a single query.

Example of using the Gemini CLI with --prompt flag:

gemini --prompt "what is a requirements.txt commonly used for in Python?"

📣 Gemini CLI News

• 👨‍💻 Check out the official open source Gemini CLI GitHub repository.
• 📑 Read the official Gemini CLI launch blog.
• 🧩 Complete the Gemini CLI Getting Started Codelab.

📝 Feedback Wanted

Have any questions? Run into any issues? We want to hear all about it!

Since Gemini CLI is fully open-source, you can file issues, start discussions, or even contribute code, all on GitHub!