It would seem state of the art LLM agent orchestration is not cheap for a hobbyist.
It's minimal, but I'm posting things.
AI context should look like a top-down funnel, become more and more precise and relevant to each step of the task at hand.
It would seem state of the art LLM agent orchestration is not cheap for a hobbyist.
Lately, I've seen vibe coders embrace a new term: "spec-driven development". I just find the term funny, given it's what we've already been doing for many years.
Wow! Using GPT-5-Codex I managed to fix something that has been irking me for years in the KDE Plasma graphical environment ; roughly cost me 30 cents.
Open AI's GPT-5 Codex coding agent is the closest I've come to a pleasant and smooth experience coding with AI.
Thinking of adding a /now page, like Derek Sivers.
It's healthier approaching Vim and customizing your experience with it as a hobby; it's about creating a tool which helps you enjoy your job more. It's not about being "more productive".
Own your content!