David Palfery

I started with Cursor and it was great early on, but as I gained more experience with AI-assisted development, the cracks showed. The recurring command window stall bug became frustrating, and the shifting pricing model—unlimited, then limited, then adjusted again—eroded trust. I want transparency and predictability in usage and cost. I need clear visibility into tokens or message limits so I know when I’m close to hitting a wall. Paying for unused capacity makes no sense. Lately, I’ve found the pay-per-token model in some open-source VS Code extensions to be a far better fit.

Kilo Code has become my go-to AI coding assistant. Unlike other tools, it gives me real control: rules, modes, and memory banks make it possible to align the model with how I actually work. I can keep context clean, tune responses per mode, and manage cost with precision. The VS Code extension is stable and efficient, giving me visibility into tokens and ensuring I know exactly what’s being sent. This makes it easier to optimize prompts, enforce standards, and prevent token waste. Kilo Code stands out because it’s not just about generating code—it’s about governing how code is created. For developers who want structure instead of noise, this is the product that delivers.