n8n AI Agent Complete Tutorial: Build Automation Workflows from Zero (2026)

n8n is the #1 open source AI workflow platform with 196K GitHub stars. This complete Chinese tutorial walks you through 6 hands-on examples — from email auto-reply to multi-agent collaboration — all running self-hosted and free.

💡 What You Will Learn

n8n is the #1 open source AI workflow platform with 196K GitHub stars. This complete Chinese tutorial walks you through 6 hands-on examples — from email auto-reply to multi-agent collaboration — all r

---

• OpenAI (GPT-4o, GPT-4.1)
• Anthropic Claude (Opus, Sonnet)

docker run -it --rm \
  --name n8n \
  -p 5678:5678 \
  -v ~/.n8n:/home/node/.n8n \
  n8nio/n8n

You are a professional email assistant. Analyze incoming emails and generate professional, polite reply drafts. Keep it concise, under 200 characters.

 5 - []

User Input → AI Agent (Researcher) → Code Node (Clean Data) → AI Agent (Writer) → Output

{
  "agent": "Researcher",
  "tools": ["Web Search", "HTTP Request"],
  "systemPrompt": ""
}

// 
const research = $input.first().json.data;
const cleaned = research.output
  .replace(/\[\d+\]/g, '')  // 
  .split('\n')
  .filter(line => line.trim().length > 0);
return [{ json: { cleanedNotes: cleaned.join('\n') } }];

{
  "agent": "Writer",
  "tools": [],
  "systemPrompt": " 500 Use"
}

Schedule (7:00) → 
  ├─ HTTP Request (Fetch News API) → 
  ├─ AI Agent (Summarize) → 
  ├─ Code Node (Format) → 
  └─ Slack + Email (Publish)

const summary = $input.first().json.summary;
const date = new Date().toLocaleDateString('zh-TW');
const formatted = `# 📰 AI  - ${date}\n\n${summary}\n\n---\n* n8n AI Agent *`;
return [{ json: { formatted, subject: `AI  ${date}` } }];

flowchart LR
    A[Webhook] --> B[AI Agent<br/>→]
    B --> C[Code Node<br/>]
    C --> D[AI Agent<br/>→]
    D --> E[Code Node<br/>]
    E --> F[Notion<br/>]

const items = $input.all();
const zh = items[0].json.translated;
const en = items[1].json.translated;
const ja = items[2].json.translated;

return [{
  json: {
    original: items[0].json.original,
    zh, en, ja,
    generatedAt: new Date().toISOString()
  }
}];

Input → 
  Agent: Strategy Planner →
    ├─ Agent: Researcher → 
    ├─ Agent: Data Analyst →
    └─ Agent: Writer → 
  Aggregate Output

#  n8n  Docker  Ollama
docker run -d --name ollama \
  -v ollama:/root/.ollama \
  -p 11434:11434 \
  ollama/ollama

# 
docker exec ollama ollama pull qwen2.5-coder:7b