Tools: A dev coding setup in 2026

Everyone is talking about AI in 2026 — here’s what actually matters as a programmer ## How I ended up using AI for frontend (yes, even as a backend dev) ## Why Codex (and not just “another AI tool”) ## How I’m using Codex in practice ## Frontend + native setup → codex-cli ## Backend work → IntelliJ + Codex ## MCP: the underrated superpower ## A reality check (very important) ## But let’s be fair… Everyone keeps talking about AI in 2026 like it’s some abstract future thing. But honestly, the most important skill right now is very simple: 👉 You need to know how to use AI well. Not “know that it exists”. Not “sometimes ask ChatGPT something”. But actually integrate AI into your daily workflow in a way that saves time and mental energy. Over the last month, I’ve been working on a personal project. I’m a backend software engineer, and to be very honest: I don’t love frontend work. I can do it, but it’s not where I want to spend most of my energy. So I started looking for a way to automate as much of it as possible. That’s when I landed on OpenAI Codex. Codex is one of the best AI coding agents I’ve used so far, mainly because: No huge setup, no weird abstractions. It feels like a tool, not an experiment. For frontend and native parts of the project, I’m using codex-cli. The biggest value for me isn’t just “writing components”. Basically: everything that steals time but doesn’t deserve brainpower. For backend development, my setup is: And for me, this works better than the Windsurf plugin. Because Codex is natively supported inside IntelliJ, which makes a big difference in stability. With Windsurf, I had a lot of friction: With Codex, it’s the opposite: That alone improves the experience a lot. One feature that deserves more attention is MCP. When you install MCP via the CLI, you can also use it inside Codex in IntelliJ. What does that give you? Honestly, it improves the collaboration between the tools so much that it just makes the day… lighter 😄 One thing I need to say: ⚠️ You cannot blindly trust everything AI gives you. Sometimes the output is: You still need judgment. You still need to review. You still need to think. ✨ The code is pure gold. It massively reduces development time, removes friction, and lets you focus on decisions instead of typing. AI won’t replace developers. But developers who know how to use AI properly will absolutely move faster than the rest. And in 2026… that’s not optional anymore. Templates let you quickly answer FAQs or store snippets for re-use. Are you sure you want to hide this comment? It will become hidden in your post, but will still be visible via the comment's permalink. Hide child comments as well For further actions, you may consider blocking this person and/or reporting abuse - it’s easy to install - easy to configure - and, most importantly, easy to actually use - create build configs - generate environment setup scripts - scaffold project structure - automate repetitive boilerplate - handle basic but annoying setup tasks - trouble reviewing changed files - weird diffs - moments where I just didn’t trust what was happening - changes are clear - reviews are clean - everything feels predictable and simple - better communication between CLI AI and IDE AI - shared context - smoother transitions between tasks - overcomplicated - or just straight-up trash