Gemini CLI 3.0: NEW GenKit Extension! Powerful AI Coding Agent Beats Claude Code!

Key Concepts

• Gemini CLI: A powerful command-line interface enabling developers to interact directly with Gemini models for coding, automation, debugging, and AI-assisted workflows.
• Gemini 2.5 Pro: A state-of-the-art AI model accessible through the Gemini CLI.
• AI Extension Marketplace: A new framework allowing full customization of the Gemini CLI and integration with external tools.
• Context Files: Components within extensions that provide contextual information to Gemini.
• Custom Commands (Slash Commands): User-defined, reusable prompts that streamline repetitive interactions and can invoke extensions or AI agents.
• Tool Calling: The ability of an AI model to interact with and utilize external tools; the new extensions address Gemini's previous limitations in this area.
• Open Ecosystem: The collaborative environment for Gemini CLI extensions, supporting integrations from various providers (e.g., payment, database handlers).
• MCPs (Multi-Context Prompts): Bundled within extensions, these are designed to teach Gemini how to use specific tools, from local scripts to external APIs.
• GitHub URL / Local Path: Methods for installing Gemini CLI extensions directly.
• Pre-built Template: A starting point for developers to create their own custom Gemini CLI extensions.
• Public Repositories / Extension Store: Sources for discovering and installing third-party developed extensions.
• Image Gen: An example capability, demonstrated by the "Nana Banana" extension, allowing direct image generation within the terminal.
• Postman Extension: A real-world application example demonstrating how extensions can scaffold API request collections for streamlined testing and development.
• AI Agents: Intelligent components that, when combined with custom commands, can transform complex instructions into simple, one-line commands.
• Docs / Changelog: Official documentation and a record of bug fixes and minor edits, providing comprehensive guidance and update transparency.
• npm install: The Node Package Manager command used to install the Gemini CLI.
• Node.js 18+: A prerequisite runtime environment for installing and running the Gemini CLI.
• Google Account authentication: Required for free access and usage of the Gemini CLI.
• Chrome DevTools MCP: An example extension that enables the use of AI agents within the Chrome browser for tasks like navigation and file operations.

Introduction to Gemini CLI and Recent Upgrades

The Google DeepMind team has released a significant upgrade to the Gemini CLI, making it a top-tier, free command-line tool with access to advanced models like Gemini 2.5 Pro. Recent features drastically enhance its capabilities, including an AI extension marketplace, context files, and custom commands. These additions specifically address Gemini's previous lack of tool calling ability, enabling it to interact with external tools. The Gemini CLI is a powerful interface for developers, supporting coding, automation, debugging, and AI-assisted workflows, offering up to 1,000 free requests per day, making it highly accessible.

Gemini CLI Extension Framework

The most impactful new feature is the Gemini CLI extension framework, which allows for complete customization and seamless integration of the CLI with various tools directly from the command line. This eliminates the need for context switching between applications, consolidating entire workflows within the Gemini terminal.

• Core Functionality and Design:

  • Extensions are installed using the gemini extension install command, specifying a GitHub URL or local path.
  • The framework fosters an open ecosystem that supports top-tier integrations, ranging from payment providers to database handlers.
  • Extensions are designed for simplicity and power, bundling MCPs (Multi-Context Prompts), context files, and custom commands into lightweight packages. These packages effectively teach Gemini how to use any tool, whether local scripts or external APIs.
  • The workflow is straightforward: extensions can be installed directly from GitHub URLs or local folders, used instantly to enhance the CLI experience, and managed effortlessly for a clean and flexible setup.

• Building and Accessing Extensions:

  • Developers have the ability to build their own Gemini CLI extensions, either by starting with a pre-built template for common use cases or by creating one from scratch for complete customization.
  • A growing list of public repositories created by third-party developers is accessible through the Gemini CLI extension store, allowing users to easily install extensions to expand CLI functionality.

• Example: Nana Banana Extension:

  • The "Nana Banana" extension is highlighted as an example. After installation via its script and the gemini extension install command, it enables direct image generation (image gen) capabilities within the Gemini CLI terminal.

Enhanced Custom Slash Commands

Initially introduced in July, custom slash commands allow developers to define reusable prompts, significantly streamlining repetitive interactions. The latest update brings substantial improvements:

• They can now directly invoke the newly released extensions.
• They facilitate the use of AI agents to transform complex, multi-step instructions into simple, one-line commands.
• This enhancement drastically improves speed, consistency, and productivity by enabling better instruction of Gemini agents to follow rules and properly utilize tools, thereby mitigating Gemini's previous limitations in robust tool-calling.

Documentation and Changelog

Google DeepMind has introduced comprehensive documentation (docs) for the Gemini CLI, providing thorough explanations for almost every feature. This resource is crucial for new users to understand how to get started and maximize the utility of its features. Additionally, a changelog details all bug fixes and minor edits implemented over time, showcasing continuous refinement aimed at making Gemini CLI one of the best terminal agents available.

Getting Started with Gemini CLI

The process for installing and using the free Gemini CLI is straightforward:

• Installation Steps:
  1. Ensure Node.js 18 or above is installed on your computer.
  2. Open your command prompt or terminal.
  3. Execute the npm install -g @google/gemini-cli command (implied by the context of "npm install command"). This installs all necessary packages for Gemini CLI functionality.
  4. Start the Gemini CLI using the gemini command.
• Authentication: Authenticate the CLI with your Google account, which is completely free.
• Managing Extensions (Example: Chrome DevTools MCP):
  • After installation, the gemini /extensions command allows users to manage and list all installed extensions.
  • An example installation of the Chrome DevTools MCP (Multi-Context Prompt) is demonstrated. This extension is designed to help use AI agents within the Chrome browser.
  • Once successfully installed and enabled, the gemini extension command lists active extensions.
  • Using Ctrl+T (or a similar command) within the Gemini terminal provides a command list for the active MCP, detailing available tools such as listing, opening files, clicking, and opening new pages. The example shows the AI agent successfully navigating to a specific URL (the presenter's channel) within a new Chrome tab.

Conclusion/Main Takeaways

The latest upgrades to the Gemini CLI, particularly the introduction of the AI extension framework and enhanced custom slash commands, significantly boost its utility and address previous limitations in tool calling. By enabling deep integration with external tools and streamlining complex workflows, Gemini CLI offers developers a powerful, customizable, and free AI-assisted command-line environment. The comprehensive documentation and continuous updates further solidify its position as a top-tier tool for modern development, providing actionable insights and preserving specific details for enhanced productivity.