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.
1.7K views
Replies
Dirac
For frontend focused:
Cursor + 21st.dev resources
For backend/functionality:
Cursor + testsprite MCP
Pretty bare-bones :)
100 Vibe Coding
I am using Cursor and Claude the usual way and trying now Google Ai Studio Build for very quick ideas.
We have shipped 100vibecoding.com to help users become better at vibecoding and learning to ship from zero.
I am a bit surprised not to see any mention of augmentcode in this thread. I started with Cursor, and it was working well in the early days of my app but it got itself overwhelmed rather quickly as the app/codebase grew in size. I was quite frustrated (am I really going to have to manually read/update this code? Noo!!!). Thats when I switched to augmentcode. It has been a significant improvement. Still make lots of mistakes, lots of shouting matches but at least the work is getting done. Within augment I have the option of Claude (Sonnet 4) and GPT-5. I started with Claude since that seems to be the go-to for everyone. But recently I've fount GPT's reasoning ability to be a big step up for what I need. So thats what I use now. Its not all that expensive although they dont allow carry-over and keep switching up the pricing structure which are both annoying but I will live with them. Hey, at least my app is working now. Do I really have to know what code it is running? (Dont answer that I am ashamed enough as it is).
Having said all that...is is slowwww as molassesss...so here I am wasting time while it does its thing. Imagine me being this slow at work??
Three days ago, while I was using Cursor + Claude Code, Cursor suddenly stopped responding. Had to go back to VSCode for a few days.
But yesterday Cursor started working again.
😂
Interesting.. reading through the replies, it feels like a clear majority right now is gravitating toward Claude + VS Code as their daily driver. Yeah, with the latest Plan phase introduced by Claude to perform much quicker file reads, reasoning, and well, planning, before making changes really changed the game. Now, it better understands user's intent and give convenience for modification before jumping straight into agent mode, which if anything, boosts user confidence at least. The set-up is intuitive and effective without feeling "over-automated".
Which made me reflect on Cursor. Cursor was ahead of its time when it launched:
clean UIUX
strong repo-wide context / robust file-reading operations
opinionated AI workflows that actually worked
For a while, it felt unmatched. But a year on, the gap seems to be narrowing fast. When you have VS Code + Claude (or GPT) + good planning primitives, a lot of Cursor’s original edge feels.. less exclusive?
It seems to me Cursor days are numbered. I think Cursor is a strong product, but it's also. a relatively small company competing against tech Behemoths who can quickly replicate most of the core functionality, scale, and even pour more resources to edge out with custom features. I might be missing some deeper Cursor-specific advantages or roadmap bets, though.
Anyway, curious what others think (and any other thoughts!):
What does Cursor still do meaningfully better today?
Where do you think their long-term moat is?
Or are most people converging toward “LLM + vanilla IDE” stacks? Would specialized IDEs be here to stay?
Wow, reading everyone else's stack makes me feel a little ... less than ... so lets just say that even though I have been actively building for months and have a live product, I didn't know that I even had a 'stack' ... LOL
My 'stack' progressed from @Gemini and a text editor to Gemini and @vscode.dev with @github for versioning my main and dev branches and then moved on to @Google AI Studio. Primary constraint was, still is, that I am bootstrapping (another new term for me) everything, so finding every possible way to do what I need to do with zero, or the least possible cost.
im building mostly with @Lovable and a host of tools they already integrate with like @supabase and IONOS (i was a loyal goDaddy user for many years) Depending on what i'm making, i might also use @Claude by Anthropic to help write and run scripts
My stack: Claude + Claude Code + Xcode Just shipped Usage4Claude v2.0 with this combo (macOS menu bar app that monitors your Claude usage quotas in real time). Using Claude to build a tool that tracks Claude usage... extremely 2025 😂 Claude Code is surprisingly strong with Swift/SwiftUI. It just gets all the quirks in Apple's ecosystem without needing tons of explanation. Went from idea to v2.0 way faster than I expected. GitHub: https://github.com/f-is-h/Usage4... Anyone else here building native desktop apps with AI? Seems like most people are focused on web stuff.
I use Notion and Replit Exclusively now