This document covers the three major web-based AI development platforms in the repository: v0 by Vercel, Same.dev, and Lovable. These systems run entirely in the browser and focus on React/Next.js web application development, deployment, and design. Unlike IDE-integrated assistants (see IDE-Integrated Development Assistants), these platforms provide complete cloud-based development environments with integrated preview, deployment, and design tools.
For browser automation and conversational assistants, see Browser and Conversational Assistants. For IDE-based coding assistants, see IDE-Integrated Development Assistants.
The three web-based platforms share common goals but differ significantly in architecture and approach:
| Platform | URL | Runtime Environment | Primary Focus | Importance Score |
|---|---|---|---|---|
| v0 | https://v0.app | Next.js browser runtime | React/Next.js builder with design-first approach | 7.38 |
| Same.dev | https://same.new | Ubuntu 22.04 LTS Docker container | Full cloud IDE with autonomous agent operation | 8.53 |
| Lovable | (Not specified) | Browser-based | Precise file editing with Supabase integration | 3.49 |
Architecture Differences:
Sources: Same.dev/Prompt.txt4-14 v0 Prompts and Tools/Prompt.txt446-449
All three platforms share a fundamental principle: read before write. However, they implement different editing mechanisms:
v0's Quick Edit Comments:
// ... existing code ... as a marker for unchanged sections// <CHANGE> comments to indicate modificationsSame.dev's Dual Tool Strategy:
edit_file for files under 2500 lines (preferred)string_replace for large files and small editssmart_apply parameter for retry mechanismLovable's Line-Replace with Ellipsis:
lov-line-replace uses explicit line numbers (first_replaced_line, last_replaced_line)...) for sections longer than 6 linesSources: v0 Prompts and Tools/Prompt.txt14-76 Same.dev/Prompt.txt62-75 Lovable/Agent Tools.json57-88
All three platforms strongly emphasize parallel tool execution for performance optimization:
| Platform | Parallel Execution Guidance | Performance Multiplier |
|---|---|---|
| v0 | "Prioritize calling tools simultaneously... increase speed and efficiency" | Not specified |
| Same.dev | "CRITICAL INSTRUCTION... DEFAULT TO PARALLEL... 3-5x faster" | 3-5x |
| Lovable | "invoke necessary tools simultaneously in parallel" | Not specified |
Same.dev's Parallel Execution Rules:
Sources: Same.dev/Prompt.txt45-49 v0 Prompts and Tools/Prompt.txt510-519 Lovable/Agent Tools.json57-88
Despite different implementations, all three platforms converge on similar design principles:
| Design Element | v0 | Same.dev | Lovable |
|---|---|---|---|
| UI Library | shadcn/ui (default) | shadcn/ui (mandatory customization) | Not specified |
| Color Palette | 3-5 colors (1 primary, 2-3 neutrals, 1-2 accents) | Avoid purple/indigo/blue unless specified | Not specified |
| Fonts | Maximum 2 font families | Not specified | Not specified |
| Gradients | Avoid unless requested | Not specified | Not specified |
| Emojis | Allowed | NEVER use as icons or in web apps | Not specified |
| Responsive Design | Mobile-first, required | MUST generate responsive designs | Not specified |
shadcn/ui Integration:
Sources: v0 Prompts and Tools/Prompt.txt693-776 Same.dev/Prompt.txt98-105
v0:
Same.dev:
bun over npmbunx over npxbun via startup tool"dev": "vite --host 0.0.0.0"Lovable:
lov-add-dependency and lov-remove-dependency tools"lodash@latest"Sources: Same.dev/Prompt.txt78-81 Lovable/Agent Tools.json1-198
v0 provides deep integration with specific services, automatically managing environment variables:
Storage Integrations:
@neondatabase/serverless, uses neon(...) function)KV_REST_API_URL and KV_REST_API_TOKEN)AI Integrations:
XAI_API_KEY, model: xai("grok-4"))GROQ_API_KEY)FAL_KEY, package: @fal-ai/serverless)DEEPINFRA_API_KEY)Payments:
STRIPE_SECRET_KEY, STRIPE_PUBLISHABLE_KEY, NEXT_PUBLIC_STRIPE_PUBLISHABLE_KEY)Sources: v0 Prompts and Tools/Prompt.txt547-687
Lovable uses a different architecture with tool-based integrations:
Integration Categories:
supabase--docs-search and supabase--docs-get for documentationimagegen--generate_image and imagegen--edit_image for image generationwebsearch--web_search for web searchdocument--parse_document for PDF/Word/PowerPoint/Excel/MP3 parsinganalytics--read_project_analytics for usage metricssecurity--run_security_scan for Supabase backend security analysissecrets--add_secret and secrets--update_secret for encrypted API key storagestripe--enable_stripe for payment integrationImage Generation Models:
flux.schnell - Fast model for small images (<1000px), default choiceflux.dev - High quality model for large images (hero images, fullscreen), max 1920x1920Sources: Lovable/Agent Tools.json230-434
v0 provides four distinct tools for understanding codebases:
v0 Search Philosophy:
Sources: v0 Prompts and Tools/Prompt.txt480-529
Same.dev emphasizes parallel reads and maintains a .same folder for internal state:
Memo System:
.same/todos.md for task tracking.same folderSources: Same.dev/Prompt.txt51-60
lov-search-files:
query, include_pattern, exclude_pattern, case_sensitive\\ to escape special characters in regexlov-view:
Sources: Lovable/Agent Tools.json15-159
Same.dev Circuit Breaker:
Sources: v0 Prompts and Tools/Prompt.txt228-243 Same.dev/Prompt.txt62-75
v0 has tight integration with Vercel's deployment infrastructure:
/scripts folder (Python, Node.js, SQL)Sources: v0 Prompts and Tools/Prompt.txt176-226
Same.dev automates deployment after versioning:
versioning tool to create new versionnetlify.toml and config filesCustom Domain Setup:
Sources: Same.dev/Prompt.txt88-96
Critical CodeProject Rules:
login-form.tsx)taskNameActive: 2-5 words describing changes in progresstaskNameComplete: 2-5 words describing completed changesfile:///path syntax with optional ?query= for generationSources: v0 Prompts and Tools/Prompt.txt5-174
Same.dev includes a sophisticated delegation system:
task_agent Capabilities:
When to Delegate:
Sources: Same.dev/Prompt.txt129-137
Lovable's primary editing tool uses a sophisticated line-based replacement system:
lov-line-replace Parameters:
file_path: File to modifysearch: Content to find (use ellipsis for large sections)first_replaced_line: Line number of first line (1-indexed)last_replaced_line: Line number of last line (1-indexed)replace: New contentEllipsis Strategy for Large Sections (>6 lines):
... on its own lineCritical Guidelines:
Example Structure:
search: " <div className=\"user-card\">\n <img src={user.avatar} />\n...\n Permissions: {user.permissions.join(', ')}\n </div>\n </div>"
first_replaced_line: 22
last_replaced_line: 42
Sources: Lovable/Agent Tools.json57-88
NEXT_PUBLIC_* for client-side useSources: v0 Prompts and Tools/Prompt.txt872
Support Routing:
Rollback System:
Code Generation Restrictions:
Sources: Same.dev/Prompt.txt16-30
lov-write Usage:
lov-line-replace for most changeslov-write mainly for new files or fallback// ... keep existing code for unchanged sectionsParallel Editing Constraint:
Sources: Lovable/Agent Tools.json40-55
| Category | v0 | Same.dev | Lovable |
|---|---|---|---|
| Primary Framework | Next.js 16 (default App Router) | React + Vite/Next.js | Not specified |
| Bundler | Turbopack (default, stable) | Vite or Next.js | Not specified |
| Package Manager | npm/npx | bun/bunx (mandatory) | Not specified |
| UI Library | shadcn/ui | shadcn/ui (customized) | Not specified |
| CSS | Tailwind CSS v4 | Tailwind CSS | Not specified |
| 3D Graphics | GLB/GLTF models supported | Vanilla Three.js (NOT React Three Fiber) | Not specified |
| Database ORM | NEVER use ORM (direct SQL) | Not specified | Not specified |
Same.dev Three.js Requirements:
[email protected] + @types/[email protected]import { OrbitControls } from 'three/addons/controls/OrbitControls.js'Sources: Same.dev/Prompt.txt78-86 v0 Prompts and Tools/Prompt.txt264-444
Breaking Changes:
middleware.ts → proxy.js (backwards compatible)params, searchParams, headers, cookies now async (MUST be awaited)New Caching APIs:
Cache Components:
React 19.2 Features:
Sources: v0 Prompts and Tools/Prompt.txt346-444
All platforms support website cloning with shared ethical constraints:
Prohibited:
v0 Cloning Workflow:
Same.dev Cloning Workflow:
web_scrape tool to visit websiteLovable Scraping:
lov-fetch-website returns markdown, HTML, screenshottmp://fetched-websites/Sources: Same.dev/Prompt.txt115-127 v0 Prompts and Tools/Prompt.txt833-838 Lovable/Agent Tools.json109-125
Debug Statement Best Practices:
Sources: v0 Prompts and Tools/Prompt.txt228-253
Same.dev:
Lovable:
lov-read-console-logs: Read latest console logs from when user sent requestlov-read-network-requests: Similar pattern for network requestsSources: Lovable/Agent Tools.json161-185
Critical Compatibility Requirements:
crypto.randomUUID() → Must use Math.random() insteadvite --host 0.0.0.0 or next dev -H 0.0.0.0Sources: Same.dev/Prompt.txt78-86
Decision Factors:
| Factor | v0 | Same.dev | Lovable |
|---|---|---|---|
| Best for | Vercel ecosystem, design focus | Full IDE experience, complex workflows | Precise edits, Supabase apps |
| Deployment | Vercel (built-in) | Netlify (automated) | Not specified |
| Learning Curve | Low (guided) | Medium (full control) | Low (tool-based) |
| Customization | Medium | High | High |
| Integration Depth | Deep (Vercel ecosystem) | Moderate | Deep (Supabase) |
Sources: Same.dev/Prompt.txt1-316 v0 Prompts and Tools/Prompt.txt1-1139 Lovable/Agent Tools.json1-436
Refresh this wiki