Honestly, what stands out most is how well it keeps track of context across an entire project, not just a single file. I’ve been using it for Python tooling, and it actually seems to understand what I’m building rather than only reacting to whatever happens to be on screen. For someone doing security research and creating tools from scratch, that kind of awareness makes a real difference.
The terminal interface feels natural if you already spend most of your time in the command line, and having Opus under the hood means it doesn’t fall apart when the logic gets complex. It just… works the way you’d want an AI coding assistant to work. Reseña recopilada por y alojada en G2.com.
The cost is the most obvious downside - running on Opus adds up quickly if you use it a lot, and there isn’t really a middle ground where you still get strong reasoning without burning through credits. The other issue is that it can occasionally go off on its own and make changes you didn’t actually ask for. That’s great when it nails it, but frustrating when you have to backtrack and undo things. A clearer confirmation step before it touches multiple files would make this feel much safer. None of this is a dealbreaker, but the pricing model could be more friendly for indie developers or students who aren’t putting it on a company card. Reseña recopilada por y alojada en G2.com.
