Google just dropped something actually useful for builders: Gemini CLI (Gemini command line interface). In this video, I walk you through how to install and set up Gemini CLI for free, straight from the official blog + repo. The big win? The usage limits are wild compared to the old days—60 model requests per minute and 1,000 requests per day—and it’s open-source. I show the exact flow in terminal: you install it (Node.js 18+ required), pick your mode (yes, I went dark mode), and authenticate either with your Google account or a Gemini API key from AI Studio. After setup, I stress-test it the only way I know: by building. First I generate a simple responsive to-do app with HTML/CSS/JS and verify core actions like add + delete. Then I push it into a more complex to-do list where it asks permission to use React, installs dependencies, and scaffolds the whole project. It doesn’t go perfectly (toggleDarkMode wasn’t defined), but that’s the point—I copy the error, paste it back into Gemini CLI, and it fixes the code. By the end, the advanced app has dark/light mode, search, edit, due dates, sorting, priority colors, and delete working. If you get stuck, just feed the error back to the agent and keep moving fast.