VelistBuild Fullstack Apps

One Command. Full Stack. Production Ready.

# One command = Fullstack app with auth, DB, testing
bun create velist my-app
cd my-app && bun dev

// features/users/api.ts
export const usersApi = createProtectedApi('/users')
  .get('/', async (ctx) => {
    const users = await userService.getAll()
    return ctx.inertia.render('users/Index', { users })
  })
  .post('/', async (ctx) => {
    const user = await userService.create(ctx.body)
    return ctx.inertia.redirect('/users')
  })

<!-- features/users/pages/Index.svelte -->
<script>
  import AppLayout from '$shared/layouts/AppLayout.svelte'
  let { user, users } = $props();
</script>

<AppLayout title="Users" {user} path="/users">
  <div class="max-w-7xl mx-auto p-6">
    <h1 class="text-2xl font-bold mb-4">Users</h1>
    <div class="grid gap-4">
      {#each users as u}
        <div class="p-4 bg-white rounded-lg shadow">
          {u.name}
        </div>
      {/each}
    </div>
  </div>
</AppLayout>

🚀 Build 99% AI-Made Apps (You Just Review) ​

Velist is the first framework built for AI agents. Not AI-assisted. AI-made.

The 5-Agent Assembly Line ​

Every agent stops for your approval — no black box, no surprises.

How It Works (Real Example) ​

You type:

@workflow/agents/product.md

Build a task management app for teams.
Features: create tasks, assign members, set deadlines, 
mark complete, filter by status/priority.

AI does:

1. ✅ Product Agent → Writes PRD, User Stories, Roadmap
2. ✅ You review → Approve or request changes
3. ✅ Tech Lead → Designs DB schema, API routes, page structure
4. ✅ You review → Approve architecture
5. ✅ Developer → Codes all features (backend + frontend)
6. ✅ You review → Test the app
7. ✅ QA Agent → Writes & runs unit + E2E tests
8. ✅ You review → Final approval
9. ✅ DevOps → Deploys to production

Result: Production-ready app. You wrote 0 lines of code. You just reviewed and approved.

📖 Learn the AI Workflow

Why Developers Choose Velist ​

🎯 Stop Wrestling With Configuration ​

Traditional fullstack development means configuring 10+ tools. With Velist, everything works out of the box:

🤖 Built for the AI Era ​

Velist is the first framework designed specifically for AI agents:

• Structured prompts for each agent role (Product, Tech Lead, Developer, QA, DevOps)
• Mandatory review points - agents can't skip human approval
• Feature isolation - each feature is self-contained, perfect for parallel agent work
• Clear contracts - types define boundaries between agents

🧪 Testing Strategy (QA Agent Workflow) ​

Two-level testing approach as defined in QA Agent workflow:

# Unit tests (fast, isolated)
bun run test
bun run test:watch

# E2E tests (critical flows only)
bun run test:e2e

Test Priority:

1. Service layer (business logic, edge cases)
2. API routes (happy path, auth checks, validation errors)
3. Repository (if complex queries)

🚀 Deployment (DevOps Agent Workflow) ​

Production-ready VPS deployment workflow:

✅ BUILD → ✅ DEPLOY → ✅ VERIFY → ✅ MONITOR

Native VPS deployment (default):

• SSH-based deployment
• PM2 process manager
• SQLite database (embedded, no separate DB server)

Docker optional - For teams that prefer containers (note: 30-70s build time vs 3-5s with PM2):

# Docker deployment (alternative)
docker-compose up -d

Deployment Comparison:

⚡ Developer Experience That Actually Delivers ​

• Hot reload that actually works - server and client
• Type safety from database to UI props with TypeScript
• Database migrations auto-generated by Drizzle from schema.ts
• Dark mode included in the starter template
• Shared utilities ($shared/lib) - CSV, debounce, toast, UUID, image processing
• Modern input styling - Clean focus states, consistent dark mode support
• Active path navigation - Menu highlighting out of the box

🛠️ Our Stack (The Real Deal) ​

We don't hide our dependencies. Here's what Velist actually uses:

🔐 Security & Real-time Features ​

Velist includes enterprise-grade features that usually require hours of setup: