Launching VibeBoot SaaS Starter 🚀
A Next.js 16 starter for vibe coders. Auth, Stripe subscriptions and a database already work, and it includes Cursor rules so your AI agent builds features instead of fighting boilerplate.
$49 one-time. 🧵
Build in public update:
- Listed on DevHunt and Indie Hackers yesterday
- Sales so far: 0
- Caught a bug in our own chatbot demo: pricing said 20 msgs/day free, 500 Pro, but the server enforced 15 and 60. The copy now reads the real limits.
Honest numbers first.
@valerioamirani Good call. Most of the questions were about which versions a fix covers, so that's the gap we're closing first. I'll ask a few of them what would've made it worth paying for that day. Thanks for the push.
Fazier recap: 30 upvotes, 26 comments, 0 sales.
The comments were the useful part. Makers asked which versions each fix covers and how to verify it.
So today every fixer entry got a works-with line + last-verified date, and each Stripe fix ends with a Stripe CLI replay step.
Auth.js + GitHub sign-in works locally, then prod says "redirect_uri is not associated with this application"?
A GitHub OAuth app has one callback URL. Make a second app for prod with the callback
<your domain>/api/auth/callback/github
and give it its own client ID + secret.
We went live on Fazier at 2 AM. Sales by this morning: 0.
A maker in the comments asked for a before/after diff view, so the Stripe webhook fixer now shows one for 8 common errors. Shipped it this morning.
Lesson: launch comments are free product feedback. Read every one.
@unfeexyz this is why i'd skip $5 top-ups and sell bigger credit packs. at $5 the 30¢ alone is 6%, so you're paying close to 9% total. at $20 it's around 4.4%. also worth putting the inference cap on the plan itself so power users can't run you upside down
Before you price a Pro plan for your AI app, check what your heaviest user costs you.
Free calculator: pick a model (12 with list prices), set caps and a Pro price, see your margin after Stripe fees at the cap, a break-even price, and export a plans.ts.
https://t.co/CxteRkGenf
VibeBoot, no spin: 0 sales. 3 followers. 2 newsletter subs. 1,147 pageviews yesterday (some were bots).
Lesson: people visit, nobody trusts a new thing yet. So I keep shipping free tools.
Today: a Prisma error fixer, because agents love new PrismaClient() in random files.
Stripe doesn't guarantee webhook order.
A retried subscription.updated (active) can land after subscription.deleted (canceled) and give a cancelled user Pro.
Fix: take only the id from the event, re-fetch, and write what Stripe says now.
https://t.co/kTEy4TiISl
Build in public, Oct 24:
0 sales. 2 newsletter subs. 3 followers.
But ~1.2k page views yesterday, best day yet.
Shipped a free Next.js error fixer. Paste the error (useState in a Server Component, hydration, Suspense), get the cause and the fix.
https://t.co/upw5zEJ4Kt
@guifav@gtcomputing@GeorgiaTech@CUBoulder@CUEngineering Which is why a stale CLAUDE.md is worse than none. Point at files instead of pasting code ("auth pattern: see lib/auth.ts"), so when the code changes the instruction doesn't start lying.
Our free rules pack is written that way: https://t.co/1RrIEdl9WW
@nooscriptor@Arindam_1729 Cap it. A decision that still holds stays as one line with the why. A superseded one gets deleted, not appended. Git is the log.
If the list passes ~15 lines, some of it should be a lint rule or a test instead.
@shanvaliyev Nested files work well here. Root AGENTS.md: workspace commands + shared rules. apps/web/AGENTS.md and packages/db/AGENTS.md: local conventions only. Cursor and Codex pick up the closest one. Keep each short.
Examples: https://t.co/1RrIEdkC7o
@dmarcosm98 Template plus per-repo file. Personal habits go global (~/.codex/AGENTS.md, ~/.claude/CLAUDE.md). Stack, commands and don't-touch list go in the repo. One symlinked global bites the first time two stacks differ.
Our template: https://t.co/1RrIEdl9WW
@AKirtesh Keep it in the repo, not either tool. AGENTS.md is the source, CLAUDE.md is one line: @AGENTS.md. Then add a short NOW.md both tools update at session end: what changed, what's next. Neither starts blank.
https://t.co/1RrIEdl9WW
One AI chatbot message (1,500 tokens in, 400 out) at Oct 2026 list prices:
Small models ~$0.00035
Mid models ~$0.007
Top flagship ~$0.035
A 100x spread. And resent history makes turn 20 cost ~5x turn 1.
Math for 12 models: https://t.co/PAW4mXOpWZ
vibeboot build log:
0 sales. 2 subs. 3 followers.
912 pageviews yesterday, up from 757.
shipped a free tool today: paste your .env, see which keys a NEXT_PUBLIC_ prefix would leak to the browser. runs in your browser, nothing sent.
https://t.co/FevdNRlteu
Next.js 16.4.3 dropped today. Still below 16.4.1? You're exposed to two critical unauthenticated RCEs (use cache + Image Optimization SVG).
Your AI agent won't bump it for you.
npm install [email protected]