Google

Google's command-line coding agent, built around their Gemini models. It's strong on multi-step planning and large context windows, which is where Google's research advantages show up most clearly. The CLI ergonomics are clean and the open-source posture means you can read what it's actually doing under the hood. Worth trying when you want to compare reasoning styles with Anthropic's or OpenAI's agents on the same task. I'd use it specifically when a problem needs a lot of context held at once and the other agents are losing the thread.