What’s Your Vibe Coding Stack in 2025?
by•
AI dev tools are evolving crazy fast , every few weeks there’s a new “must-try” for vibe coders.
Some people are building full products with @ChatGPT by OpenAI and @Replit , others swear by @Cursor and @Claude by Anthropic , and a few are mixing @Lovable + @v0 by Vercel + @bolt.new to ship apps in record time.
I’ve been refining my own vibe stack lately, trying to find that sweet spot between speed, control, and creativity.
It made me wonder ,what does your setup look like right now?
Share your current “Vibe Stack”:
Your go-to AI tools
How you connect them together
What kind of projects you’re shipping
Let’s crowd-source the best combos for 2025 . Who knows, maybe we’ll spot the next trend before it blows up.
2.1K views
Replies
Claude Code inside VSCode, 3x MCPs (Sequential thinking, Context 7 and Augment's Context Engine MCP), Codex for front end (seems to be better currently), Vercel, MongoDB or Superbase, GitHub Desktop.
It is an interesting question. I have not been doing this long. Started out comparing ChatGPT, Grok, and Claude.ai. I chose claude dot ai as it seem to be better at writing code. My problem is I was using claude dot ai to build a platform, and it had issues. Then I learned about project files and handoff's and these help but there was no true way to for it to accurately monitor the platform's code base. The I learned about Claude Code. I installed that in a Linux development environment and it is all I have used since. It was everything I was looking for in what I wanted it to do. I worked my way up to the MAX plan as I was and still am using Claude Code heavily and claude dot ai comes off of that max plan too. I joined here a few days ago and saw Cursor and it looks nice, as it has an interface for what I already do so I don't see the need for it in my use case and pay another 200/month when I am already paying that much for Claude Code. While I am coding though I will have other tools ChatGPT, Gemeni and so-on to make some text for me. Outside side of that I have not used anything else, I would be interested in how they compare though.
Build Check
I was trying Cursor and VS + Copilot, and the first one is better by far. Starting now to test Claude
It's Claude and VS Code for me. I do iOS development, so I need Xcode as well, but most stuff is happening in VS Code.I'm busing with two apps currently (one will launch tomorrow!) but also some backend stuff: i created a pipeline for easy screenshot management for the app stores. Basically a script that will remotly control my apps in the iphone simulator and take screenshots, then generates titles and subtitles for nano banana etc - and after a few minutes I have a dozent quality screens in a folder for uploading to Apple.
Windsurf for me, bouncing freely between a number LLMs and reasoning levels depending on the task and complexity. The other thing I've been doing is using external Gemini Pro to review most of the code I vibe, then feed back to the IDE agent iteratively until I get the code clean. Over the past year I've written two SaaS products with nearly 200K lines of code between them. One runs the charity my wife and I founded and run and one is a product I'm considering launching here, if I think there'd be interest...
For native macOS apps, my stack is: Swift + SwiftUI + Claude (API and Claude Code in terminal). No web frameworks, no Electron.
I built TokenBar (a menu bar app that tracks AI token usage across 20+ providers) with this stack. Claude handles the boilerplate, SwiftUI layout, and especially parsing different API response formats. I still write the core logic by hand, but Claude probably handles 60-70% of the typing.
The workflow: describe what I want in natural language, Claude generates Swift code, I review and iterate. For a solo dev shipping a native Mac app, this is incredibly fast. Went from idea to shipped product in about 3 weeks.
If you are building for macOS, go native. SwiftUI + Claude is a killer combo because Apple has solid documentation that the models were trained on.
My 2026 vibe coding stack has evolved quite a bit:
Claude Code for the heavy lifting (architecture, complex features, refactoring)
- Lightweight linting + AI-powered test generation as the quality gate
- Git hooks that run AI-suggested tests before every commit
The biggest addition to my stack this year: automated quality checks specifically for AI-generated code. I found that AI output needs different validation than human code — it's syntactically perfect but sometimes makes assumptions about APIs or data shapes that don't match reality.
The stack that works isn't just about code generation speed — it's about maintaining confidence that what ships actually works. Speed without quality just means you ship bugs faster.
Gemini CLI and VS Code for me. We released 2 products in production using that stack.
While all the agentic development happens inside Gemini CLI, I use VS Code for reviewing the code / writing instruction markdowns for gemini to pickup. That has been working beautifully. Never felt need of a specialized design tool. I just explain what I am imagining to gemini and it nails it perfectly in 2-3 iterations.
Surprised to see Gemini not coming up at all in this thread.
I use a combination of Replit, ChatGPT, and Cursor. I have also found that now ChatGPT Codex offers much higher limits than Claude. That's clearly their play to gain market share. I have a question: What's everyone used to test or validate their prototypes?
Perhaps an unpopular option here :) Claude Code in the terminal, used manually. I pick what to delegate, review everything that comes back, and write the interesting parts myself. Slower than going full auto but I actually know what's in my own codebase, which feels important to me