Gemini CLI is Google’s command-line coding agent, distributed as open source. Unlike the Gemini model or the Gemini API, it is a terminal tool that uses Gemini models to read the project, plan and execute coding tasks.
What it’s for
- Running coding tasks from the terminal, reading the local repository
- Generating, editing and explaining code and project files
- Running commands and using the result to continue the task
- Automating continuous integration flows, including with GitHub Actions
- Connecting external tools and context through MCP
How it works
- Installed via npm/npx with a Google account; a free tier is available for individual use
- The agent combines a large context window with access to files and commands in the current directory
- It supports MCP to attach tool and data servers
- Separate release channels (stable, preview and nightly) let you track versions
Where it runs
- Terminal on Linux, macOS and Windows
- Integration with editors and with GitHub Actions
- Local execution in the project directory
Availability and licensing
- CLI is open source under Apache-2.0
- Gemini models are proprietary and subject to Google’s terms
- Individual use has a free tier; limits and plans change over time
Strengths
- Google’s official tool for terminal use
- Direct integration with the Gemini ecosystem and with MCP
- Open source with predictable release channels
- A practical entry point for coding automation in CI
Things to note
- The product is the CLI; it does not replace the Gemini API for application integration
- Model availability and limits depend on the Google plan and account
- Like any terminal agent, it runs real commands and requires review