Tooling & stack
Which tools to build in and which tech stack to build on — the two choices that make or break a build.
Choosing tools & stack
Pick the right AI environment and a boring, popular stack the model actually knows.
4 cards →Have you picked the right vibecoding stack?
Five questions on the two choices that make or break a build — which tool you work in, and which stack you build on. No sign-up, instant score, and you can share how you did.
-
Q1 The main vibecoding tools fall into two camps. What are they?Why
The split that matters is whether you get a portable codebase or a fast hosted app. Choosing by hype instead of by that question is how people end up in the wrong camp.
-
Q2 You're committing a long-term product to a closed platform. What must you check first?Why
A hosted builder can be exactly right for a fast app — but if you can't get the code out, 'long-term product' and 'closed platform' are a bad pairing you discover far too late.
-
Q3 Why pick a 'boring', popular stack?Why
Popularity in the training data is the whole point. Chase the newest framework and you hit confident, invented APIs — and watch for the AI quietly mixing in an obscure library for one feature, creating a fragile spot you can't debug.
-
Q4 Why work in an AI IDE rather than a standalone chat window?Why
Copy-pasting between a chatbot and your editor introduces mistakes at every hop on a multi-file project. It also matters that the AI reads existing files first — otherwise it regenerates things that already exist, or conflicts with them.
-
Q5 What's the minimum setup that makes vibecoding safe rather than risky?Why
Git makes every step a checkpoint you can return to, the preview means you see what was built, and tests catch what breaks. Skipping version control until 'later' is how a bad generation becomes unrecoverable.