Dify — No-Code AI App Builder | AI Engineering Wiki

Dify is an open-source platform (Apache 2.0) for building AI applications without code. With 50k+ GitHub stars, Dify is the leading no-code solution for LLM apps in 2026.

Installation

Docker (Recommended)

git clone https://github.com/langgenius/dify.git
cd dify/docker
cp .env.example .env
docker-compose up -d

Access: http://localhost/

Self-Hosted Requirements

Server: 4GB RAM, 20GB disk
Database: PostgreSQL 13+
Cache: Redis 6+
Vector DB: Optional (Weaviate, Qdrant, Milvus)

Core Components

1. Workflow Editor

Visual drag-and-drop LLM workflow builder:

Trigger
  ↓
Input Processing
  ↓
LLM Node
  ↓
Output Formatting
  ↓
API Response

Supports:

LLM calls (OpenAI, Ollama, Azure, local)
Tool calling (web search, database)
Conditional logic
Loops and parallel processing

2. RAG Integration

Built-in document search:

Upload PDF/TXT/Web
Auto-chunk and embed
Vector search in LLM context

3. Agent Mode

Agent-based workflows with tool calling.

4. API Publishing

Publish as OpenAI-compatible REST API.

Practical Workflow

Create New App

Dashboard → "Create New App"
Choose "Blank App" or template
Configure:
- Icon
- Name
- Description
Open workflow editor

Configure LLM Node

├── Model: qwen2.5-7b
├── System Prompt: "You are helpful..."
├── Temperature: 0.7
├── Max Tokens: 500
└── Context: 4096

Format Output

LLM Output
  ↓
JSON Parser
  ↓
Template Formatter
  ↓
API Response

Dataset Management

Training data and context sources:

Upload Dataset

Datasets section
Create Dataset
Upload CSV
Auto-indexed for RAG

Example CSV

question,answer
What is AI?,Artificial Intelligence is...
Explain quantum computing,A quantum computer uses...

Use in RAG

Query
  ↓
Vector Search
  ↓
Top Results as Context
  ↓
LLM Response

API & Integration

REST API Example

curl -X POST http://localhost/api/workflows/run \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{"workflow_id": "abc123", "inputs": {"message": "Hello"}}'

Python Integration

import requests

response = requests.post(
    "http://localhost/api/workflows/run",
    json={
        "workflow_id": "abc123",
        "inputs": {"query": "What is AI?"}
    },
    headers={"Authorization": "Bearer YOUR_API_KEY"}
)

print(response.json()["output"])

Self-Hosting Checklist

1. Database

docker run -d --name postgres \
  -e POSTGRES_PASSWORD=secure \
  postgres:15

2. Redis

docker run -d --name redis \
  redis:7

3. Environment

DATABASE_URL=postgresql://user:pw@postgres:5432/dify
REDIS_URL=redis://redis:6379
SECRET_KEY=your-secret-key

Dify — No-Code AI App Builder