Prajwal Tomar Profile picture
Oct 8 12 tweets 3 min read Read on X
I just reverse-engineered how Lovable's top users build apps 10x faster.

Turns out, It's not about writing longer prompts.

It's about this structured prompting system nobody talks about ↓ Image
@lovable_dev 1/ Create a Knowledge Base before you build

Include these in your project settings:

• Project Requirements Document (PRD)
• User flow explanation
• Tech stack details
• Design guidelines
• Backend structure

The clearer your context, the better your results.
2/ Master the 4 levels of prompting

Level 1: Training Wheels

Use labeled sections in your prompts:
- Context (what you're building)
- Task (what you want)
- Guidelines (how to do it)
- Constraints (what to avoid)

Example:
Bad: "Build me a login page"

Good:
Context: I'm building a SaaS app for small businesses
Task: Create a login page with email/password
Guidelines: Use React, make it mobile-friendly
Constraints: Don't use any external auth services

Structure helps AI understand exactly what you want.

Level 2: No Training Wheels (conversational)

Level 3: Meta Prompting (use AI to improve your prompts)

Level 4: Reverse Meta (document solutions for future use)
3/ Use the "Diff & Select" approach

Don't let Lovable rewrite entire files.

Add this to prompts: "Implement modifications while ensuring core functionality remains unaffected. Focus changes solely on [specific component]."

Fewer changes = fewer errors.
4/ Always start with a blank project

Build gradually instead of asking for everything at once.

Follow this order:

• Front-end design (page by page, section by section)
• Backend using Supabase integration
• UX/UI refinements
5/ Chat Mode vs Default Mode

Chat Mode: Planning, debugging, asking questions

Default Mode: High-level feature creation

Use Chat mode to think through problems.

Use Default mode to execute solutions.
6/ Debug like a pro

When errors happen:

→ Use "Try to Fix" button
→ Copy error to Chat mode first
→ Ask: "Use chain-of-thought reasoning to find the root cause"
→ Then switch to Edit mode
7/ Mobile-first prompting

Add this to every prompt:

"Always make things responsive on all breakpoints, with a focus on mobile first. Use shadcn and tailwind built-in breakpoints."

Most users are on mobile anyway.
8/ Step-by-step beats everything at once

Don't assign 5 tasks simultaneously.

The article specifically says: "Avoid assigning five tasks to Lovable simultaneously! This may lead the AI to create confusion."

One task at a time = fewer hallucinations.
9/ Lock files without a locking system

Add to prompts: "Please refrain from altering pages X or Y and focus changes solely on page Z."

For sensitive updates: "This update is delicate and requires precision. Examine all dependencies before implementing changes."
10/ Refactoring that works

When Lovable suggests refactoring:

"Refactor this file while ensuring UI and functionality remain unchanged. Focus on enhancing code structure and maintainability. Test thoroughly to prevent regressions."
Want the full breakdown?

The complete Lovable Prompting Bible covers:

• Advanced debugging strategies (10+ specific prompts)
• Integration with make and n8n
• Stripe setup prompts
• 20+ copy-paste prompts

Read it here: lovable.dev/blog/2025-01-1…

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Prajwal Tomar

Prajwal Tomar Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @PrajwalTomar_

Oct 2
If you’re using @lovable_dev to build apps, read this first

This is everything I wish I knew before starting ↓ Image
1. Nail your first prompt

I always start inside my custom GPT, SnapPrompt, and get the full prompt for my landing page first.

This includes layout, structure, typography, and design style. I just copy-paste that into Lovable with a design reference attached and it gives me a clean starting point.
2. Always prep your technical docs before starting in Lovable

Don’t dive in blind.

Have your DB design, UI Dev plan, MVP plan, and implementation plan ready.

Keep it simple, generate them during the planning phase using GPT or Gemini.

Then just paste them as .md files into Lovable.

That way, Lovable has full context about your product from the start.
Read 15 tweets
Oct 1
Claude with CodeRabbit + TaskMaster is crazy good.

Now you can Plan with AI, Code with AI, Review with AI.

Here’s the full workflow I use to build MVPs fast and stress-free ↓ Image
1/ The workflow (quick summary)

This is the exact system I follow:
• Plan clearly (PRD) with @ChatGPT
• Break it down using TaskMaster
• Execute one task at a time using Claude Code inside Cursor
• Review everything with @coderabbitai

Let’s go deeper ↓
2/ Vibe coding used to slow me down

I’d give a vague prompt like “build a landing page”
Claude would generate something, but it’d be a mess:

• Half-baked logic
• Features I never asked for
• Debugging spiral

Fast output. Slow progress.
Read 12 tweets
Sep 29
Want to turn your MVP into a mobile app without rebuilding the backend?

START HERE.

Most founders waste months rebuilding for iOS + Android.

This is the exact workflow I used to take BrainDump from web (Lovable + Supabase) → mobile (Rocket) in hours. Image
1/ Start with a working web app

I built BrainDump’s MVP using Lovable + Supabase.

Auth, DB, UI, RLS policies, all live and working.

Your web app is the foundation. Don’t touch the backend yet. Image
2/ Store your docs + schema

Export your Supabase schema and store the docs you built after the planning phase (MVP plan, UI dev plan, etc).

Example for BrainDump:
• Create a dump
• AI generates insights + to-dos
• Display streaks + dashboard

This spec becomes your blueprint.
Read 11 tweets
Sep 28
Want your SaaS to survive launch day?

START HERE.

The biggest problem with vibe coding is security. Too many people ship unsecure, unoptimized apps that get hacked or run painfully slow.

Here’s how to make sure your SaaS is launch-ready before you hit publish ↓ Image
1. Rate limit your endpoints

If you skip this, bots or bad actors can hit your backend 100s of times per second.

This can:
• Crash your database
• Drain your Supabase usage
• Spike costs or open you to attacks

Tools to use:
• Supabase Edge Functions with a rate limiter
• Vercel Middleware
• Basic IP throttling with Next.js middleware
2. Enable Row-Level Security (RLS)

If you’re using Supabase, turn on RLS on every table from day one.

Without it, users can query other people’s data.
And yes, this happens way more than you’d think.

To set it up:
• Go to Table → RLS → Enable
• Use policies like user_id = auth.uid()

No RLS = no data security.

Pro Tip: Try asking Cursor for these policies based on your DB design and PRD. It will help you write them correctly.
Read 11 tweets
Sep 23
My full vibe coding workflow to build MVPs faster

I don’t use Figma.
I don’t write every line of code.
I don’t need big teams.

Here’s the exact AI-powered system I use to plan, build, and launch real MVPs in 3 weeks for clients ↓ Image
1. Plan using ChatGPT

Before building, I plan fast:
- ChatGPT voice to brainstorm the client idea
- Generate docs (PRD, UI Dev Plan, DB design)
- MoSCoW Method to define essentials

This gives:
- A clear feature list
- A lean scope
- Zero ambiguity
2. Skip traditional design, go straight to dev

Most devs waste weeks designing in Figma.

I use @lovable_dev instead.

It:
- Turns text into full responsive UIs
- Connects real data
- Handles auth, forms, routing

Build 70-80% of your MVP inside Lovable, depending on the complexity of your project, and then switch to cursor/claude code.
Read 12 tweets
Sep 20
CodeRabbit + MCP = game-changing code reviews 🤯

They just cracked the biggest problem in AI code reviews.

Here’s how MCP integration changes everything ↓ Image
1/ The problem with most AI reviews

AI tools usually check code in isolation.
They can flag syntax, style, and structure, but they don’t know:

• Why the feature exists
• How it fits into your product
• What rules your team follows

It’s like having a reviewer who has never seen your app.
2/ How @coderabbitai + MCP fixes this

MCP connects CodeRabbit to every tool that holds knowledge about your code.

Technical context: docs, schemas, API specs, architecture decisions
Business context: requirements, feature specs, compliance info, user stories
Organizational context: coding standards, team decisions, security policies

Same AI. Full context. Completely different results.Image
Read 11 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us!

:(