Getting Started with Fulling – Build a Full-Stack App in Minutes with AI

🧪 Fulling Tutorial – Build Your First Full-Stack App in 10 Minutes

I remember my first time with a new framework — endless documentation, conflicting tutorials, and three hours later I still haven't rendered "Hello World". With Fulling, the experience is refreshingly different. Let me walk you through building an actual app from scratch.

📋 Prerequisites

Before we start, make sure you have:

• ✅ Docker installed (Docker Desktop or Docker Engine)
• ✅ A Claude API key from Anthropic
• ✅ Node.js 18+ (if running locally without Docker)
• ✅ 5 minutes — seriously, that's about how long this takes

🔧 Step 1: Install Fulling

The fastest way to get started is with Docker. I ran this command and it pulled the image in about 30 seconds:

docker pull sertdev/fulling-runtime:latest

If you prefer running from source (which gives you more control):

git clone https://github.com/FullAgent/fulling.git
cd fulling
cp .env.example .env
# Add your Claude API key to .env
docker compose up -d

⚠️ Heads-up: The docker-compose.yml uses sertdev/fulling-runtime:latest and maps port 8080. If 8080 is already taken (it happens), change it to 3000 or whatever you like.

🚀 Step 2: Launch Fulling

Once the container is running, open your browser to http://localhost:8080.
If you see "Fulling Runtime" — congrats, it's working. If not, check docker logs fulling for errors — I once forgot to set the API key and spent 10 minutes debugging.

💬 Step 3: Build Your First App

Here's where the magic happens. Let's build a team task board. In the Fulling interface, I typed:

Build a team task manager with:
- PostgreSQL database with users, projects, and tasks tables
- A Kanban board UI (drag-and-drop between columns)
- Real-time updates via WebSocket
- User authentication (email + password)
- API routes for CRUD operations
- Deploy with Kubernetes manifests

Fulling took about 2 minutes to process this. Here's what came out:

• 🧱 Database schema — 3 PostgreSQL tables with foreign keys and indexes
• 🎨 Next.js frontend with shadcn/ui components — beautiful, responsive Kanban board
• 🔌 WebSocket handlers — real-time updates when tasks move between columns
• 🔐 Authentication system — sessions, password hashing, protected routes
• ☸️ K8s deployment YAML — ready to apply with kubectl apply -f k8s/

🧹 Step 4: Iterate with Natural Language

This is my favorite part. After the initial build, I wanted to add a feature. I typed:

Add a comments section to each task. Users should be able to
discuss tasks in real-time, with @mentions and notifications.

Fulling analyzed the existing code, added the comments table to the database, created the UI components, wired up the WebSocket events, and updated the API routes — all without me touching a single file.

💡 Pro tip: Fulling remembers the context of your project. You don't need to re-explain everything. Just say what you want to add or change, and it understands the existing architecture.

📊 Resource Usage (from my test run)

❌ Common Issues I Ran Into

1. "Connection refused" on port 8080
Double-check the container is running: docker ps | grep fulling. If it's not there, Fulling probably crashed on startup. Run docker logs fulling to see what went wrong.

2. "Claude API key not configured"
Fulling needs a valid Claude API key in the .env file. This is probably the most common mistake — don't ask how I know.

3. PostgreSQL connection issues
Fulling expects PostgreSQL to be running. If you're using the Docker Compose setup, make sure the Postgres container starts first (Compose handles this).