Google has launched Gemini CLI, a new open-source command-line interface (CLI) AI agent built on top of its Gemini 2.5 Pro reasoning model that is designed to function as a local AI assistant, bringing coding, content generation, task management, and research capabilities directly into developers’ terminals.
According to the tech giant, it provides “lightweight access” to Gemini, giving users the “most direct path” from their prompts to the model. While the company claims Gemini CLI excels at coding, it can be utilized for a wide range of complex developer workflows, such as code refactoring, document generation, executing shell commands, running scripts, and editing files.
Gemini CLI Allows Developers to Generate Code, Images, and Videos
Gemini CLI is integrated with Gemini Code Assist, an AI coding assistant aiding developers to write code more efficiently, and includes built-in Model Context Protocol (MCP) and Google Search support. It also allows developers to generate images and videos using Google’s Veo and Imagen tools.
The agent is freely available to all developers through a Gemini Code Assist license that can be obtained via a personal Google account. This will provide access to Gemini 2.5 Pro’s massive 1 million-token context window, with a usage limit of 60 model requests per minute and 1,000 requests per day. Google claims this to be the “largest allowance” offered across the industry.
Developers can feed the multimodal foundation model large codebases, documentation, and file trees for multi-step analysis or transformation. They can use the agent to analyze a project’s structure, suggest changes, and then apply edits to specific files, all via commands in natural language on the system’s terminal.
For professional-grade developers who need to run multiple agents simultaneously or prefer to use specific models, the company is offering a Google AI Studio or Vertex AI key for usage-based billing or a Gemini Code Assist Standard/Enterprise license.
Google is Encouraging Developers to Support & Expand Gemini CLI Ecosystem
Because Gemini CLI is fully open-source under the Apache 2.0 license, developers can inspect the code to understand how it works by reviewing the prompt lifecycle, file operations, and function execution, and verify its security implications. While it defaults to using Gemini 2.5 Pro, the agent can be configured to work with local models, alternative APIs, or organization-specific tools.
Google has welcomed the global developer community to contribute to the project by reporting bugs, suggesting features, improving security practices, and submitting code improvements. Everything from its system prompts to its memory store can be customized or swapped out using Multimodal Composable Functions (MCPs), which is a Python-based interface for adding custom behavior and connecting the agent to other systems.
Gemini Code Assist now shares the same technology as Gemini CLI and can be deployed on Microsoft’s open-source and lightweight code editor, Visual Studio Code, where users can place a prompt into the chat window in agent mode, and Code Assist will work on their behalf to write tests, fix errors, build features, and migrate code. Based on user prompts, the agent will build a multi-step plan, auto-recover from failed implementation paths, and recommend solutions.
Gemini CLI is available on GitHub, with complete documentation and example use cases. Google said it will continue developing the tool and has encouraged the developer community to explore, extend, and improve the ecosystem.
With Gemini CLI, Google now joins a growing ecosystem of AI developer tools that place openness and composability front and center. While many AI assistants are embedded into proprietary platforms or locked behind closed APIs, Gemini CLI invites collaboration, scrutiny, and modification.
