Should I vibe code
AI-first code editor with agents, composer, and model routing
Sure. Build the tool you are building it with. See you in eighteen months.
?
Their verdict, the Pro price and the build-time estimate come from their entry, MIT-licensed. Checked 2026-08-03.
?
Our verdict, the regret score and everything below it. Editorial and unsponsored — nobody can pay to be moved.
The honest answer
why the verdict is what it is
An editor with an agent loop is a demo you can reach. An editor you would actually use all day, keeping pace with model releases, is a full-time job with competitors who have raised money.
What actually breaks
not "if". the specific failures.
- Nothing important — but eighteen months of your evenings
- Model APIs changing under you, repeatedly
It is not a 3am problem, it is an eighteen-month one. You get the agent loop working in a fortnight and it genuinely feels good. Then you spend a year on the parts nobody demos — multi-cursor, undo across agent edits, large-file performance, an extension API — and every few months a model release changes what good looks like and the funded competitor ships it that week. Nothing fails. You are simply always six months behind on a project you cannot finish, using someone else's editor to work on it.
Is that you?
the verdict is a default, not a law
- It is an extension to an editor that already exists, not a new editor
- You are scratching one specific itch your current tools do not cover
- You are starting from a blank window and the word "editor"
If you build it anyway
the checklist, then the prompt that enforces it
- Build an extension, not an editor. The editor is a solved problem you would be re-solving badly.
- Abstract the model provider from day one — they change fast.
- Scope to one interaction you actually want, and finish it.
I want AI coding assistance built my way. Save me from myself: 1. Do not build an editor. Build an extension for one that exists (VS Code, Neovim). If I ask for a standalone editor, tell me how many person-years text editing alone represents and ask what I actually want. 2. Ask me to name the single interaction I want before writing anything — inline completion, or selection refactor, or repo chat. Implement exactly one, well. 3. Put the model behind a provider interface with at least two implementations, because the API I use today will change within months. 4. Stream responses. A blocking call in an editor feels broken regardless of how good the output is. 5. Never apply an edit without showing a diff I can reject. 6. Cache aggressively and show me the token cost per request, so the running total is never a surprise. 7. Respect .gitignore and never send ignored files as context. Add an explicit denylist for .env and credential-shaped files.
That one keeps you out of trouble. For the prompt that actually builds it, canivibecodeit.com has one.
their build prompt ↗Or don’t build it
the boring option, and the way back out
For the general case, yes — this is a well-funded, fast-moving category and matching it is not a side project. For one specific workflow nobody has built, your extension will beat all of them.
$20/mo is cheaper than your weekend.
It is an extension. Uninstall it.
Open-source AI code assistant; closest practical DIY substitute when paired with VS Code o
Questions
Why is this only DEMO ONLY rather than something harsher?
Because the only thing at risk is your time. Nobody else is harmed, nothing is irreversible, and you will learn a great deal. It is a time sink, not a trap.
Every week, someone ships something they shouldn’t have.
New verdicts, the worst thing that landed in the trap, and the occasional incident report. No other email, ever.
You are building a box that runs strangers’ code. That box is a cryptominer with extra steps.
You will replace a $20 bill with a pager that goes off on Christmas.
Reselling access to a platform that does not want you reselling access is a business model with a countdown.
last reviewed 2026-08-03 · verdict is editorial and unsponsored · shared entry data from canivibecodeit under MIT · not legal advice