Google launches extensions system for its command-line coding tool (techcrunch.com)

Google has rolled out an extensions system for its command-line AI tool, Gemini CLI, letting third parties plug functionality directly into the terminal. Dubbed Gemini CLI Extensions, the launch includes integrations from Figma, Stripe and others, plus Google’s own Nanobanana image generator as the first example. Unlike OpenAI’s ChatGPT apps, which are curated, Google’s approach is intentionally open: extensions are published to public GitHub repositories and installed manually by developers, and Google does not vet or endorse published extensions. Google frames the move as turning Gemini CLI into “an extensibility platform” and “a conduit to other tools” used in developer toolchains. For the AI/ML community this is significant because it extends model-driven workflows into everyday developer environments and accelerates third-party innovation, while shifting trust and security responsibilities to users and repo maintainers. Technically, extensions let developers call external services or run instructions directly from the CLI (e.g., generate images from the terminal), enabling tighter integration with CI/CD, automation scripts, and code maintenance flows. With Gemini CLI already claiming over one million users—largely developers—and reportedly used internally to manage Google’s code, the open-extensibility model could rapidly broaden practical, composable AI tooling but will also raise questions about supply-chain security, permissioning, and quality control.