n8n（n8n 自动化）

N8N Skill

Comprehensive assistance with n8n development, generated from official documentation. n8n is a fair-code licensed workflow automation tool that combines AI capabilities with business process automation.

When to Use This Skill

This skill should be triggered when:

• Building or debugging n8n workflows
• Working with the Code node (JavaScript or Python)
• Using expressions and data transformations
• Implementing AI agents, chains, or RAG workflows
• Configuring n8n deployments (Docker, npm, self-hosted)
• Setting up webhooks, credentials, or integrations
• Handling errors in workflows
• Scaling n8n with queue mode
• Creating custom nodes or white labeling n8n
• Using the n8n API programmatically
• Working with LangChain in n8n
• Migrating to n8n v1.0

Key Concepts

Core Components

• Workflows: Visual automation flows with nodes connected together
• Nodes: Individual operations in a workflow (trigger, action, logic, etc.)
• Expressions: Code snippets using {{}} syntax to access and transform data
• Credentials: Secure storage for API keys and authentication
• Executions: Individual workflow runs with their data and status

Data Structure

• n8n passes data between nodes as items (array of JSON objects)
• Each item has a json property containing the main data
• Binary data is stored separately in the binary property
• Use $json to access current item's data in expressions

AI Capabilities

• Agents: AI that can use tools and make decisions
• Chains: Predefined sequences of AI operations
• Memory: Store conversation history for context
• Tools: Functions that AI agents can call
• Vector Databases: Store and retrieve embeddings for RAG

Quick Reference

Example 1: Basic Expression to Access Data

// Access data from the current item
{{ $json.name }}

// Access data from a specific node
{{ $node["HTTP Request"].json.response }}

// Access all items from a node
{{ $("HTTP Request").all() }}

Example 2: HTTP Request with Authentication

When working with the HTTP Request node, handle errors and rate limits:

// In HTTP Request node settings:
// - Enable "Retry on Fail"
// - Set Max Tries to 3
// - Set Wait Between Tries (ms) to 1000

// For rate limiting, use Batching:
// - Items per Batch: 10
// - Batch Interval (ms): 1000

Example 3: Code Node - Transform Data (JavaScript)

// Access input data
const items = $input.all();

// Transform each item
return items.map(item => {
  return {
    json: {
      fullName: `${item.json.firstName} ${item.json.lastName}`,
      email: item.json.email.toLowerCase(),
      timestamp: new Date().toISOString()
    }
  };
});

Example 4: Code Node - Filter Data (Python)

# Filter items based on a condition
output = []

for item in items:
    if item['json']['status'] == 'active':
        output.append({
            'json': {
                'id': item['json']['id'],
                'name': item['json']['name']
            }
        })

return output

Example 5: Expression - Date Handling with Luxon

// Current date
{{ $now }}

// Format date
{{ $now.toFormat('yyyy-MM-dd') }}

// Add 7 days
{{ $now.plus({ days: 7 }) }}

// Parse and format custom date
{{ DateTime.fromISO($json.dateString).toFormat('LLL dd, yyyy') }}

Example 6: JWT Authentication Credential

For APIs requiring JWT authentication:

// Use JWT credential with:
// - Key Type: Passphrase (for HMAC) or PEM Key (for RSA/ECDSA)
// - Secret: Your secret key
// - Algorithm: HS256, RS256, ES256, etc.

// The JWT credential automatically generates tokens
// Use it in HTTP Request node > Authentication > JWT

Example 7: Handle Errors in Workflow

// In Code node, use try-catch:
try {
  const result = $json.data.someField.toUpperCase();
  return [{ json: { result } }];
} catch (error) {
  // Return error information
  return [{
    json: {
      error: error.message,
      originalData: $json
    }
  }];
}

// Or set up Error Workflow in Workflow Settings
// to catch all failures and send notifications

Example 8: Pagination in HTTP Request

// Use pagination to fetch all pages
// In HTTP Request node > Pagination:

// Type: Generic Pagination
// Request URL: {{ $url }}&page={{ $pageNumber }}
// Complete When: {{ $response.body.hasMore === false }}
// Next Page URL: Automatic

Example 9: AI Agent with Tools

// In AI Agent node:
// 1. Connect a Chat Model (OpenAI, etc.)
// 2. Add tools (Calculator, HTTP Request, etc.)
// 3. Configure memory if needed

// The agent can:
// - Analyze user input
// - Decide which tools to use
// - Execute tools and process results
// - Return final answer

Example 10: Environment Variables and Static Data

// Access environment variables
{{ $env.MY_API_KEY }}

// Store workflow static data (persists across executions)
const staticData = getWorkflowStaticData('global');
staticData.lastRun = new Date().toISOString();
staticData.counter = (staticData.counter || 0) + 1;

// Retrieve static data
{{ $workflow.staticData.counter }}

Reference Files

This skill includes comprehensive documentation in references/:

• llms-txt.md - Complete n8n documentation formatted for LLMs

- Installation and setup guides - Node reference documentation - API documentation - Code examples and patterns - Configuration options - Troubleshooting guides

• llms-full.md - Extended documentation with deep technical details

- Advanced configuration - Scaling and performance - Security and authentication - Custom node development - White labeling and embed options

Use view to read specific reference files when detailed information is needed.

Working with This Skill

For Beginners

1. Start with basic workflow creation:

- Trigger nodes (Webhook, Schedule, Manual) - Action nodes (HTTP Request, Set, Edit Fields) - Learn expression syntax with simple {{$json.field}} access

1. Understand data structure:

- Each node outputs an array of items - Use the data inspector to see item structure - Practice with the Edit Fields node for data transformation

1. Common patterns:

- Webhook → HTTP Request → Set → Respond to Webhook - Schedule → Code → HTTP Request → Conditional - Manual → Loop Over Items → Process Each

For Intermediate Users

1. Master the Code node:

- JavaScript mode for complex transformations - Access $input.all() for all items - Return properly formatted items with json property

1. Work with expressions:

- Use built-in methods: .first(), .last(), .item - Date manipulation with Luxon - JMESPath for complex JSON queries

1. Error handling:

- Use Try-Catch in Code nodes - Set up Error Workflows - Configure Retry on Fail for API calls

1. Data operations:

- Merge data from multiple sources - Split and filter items - Loop over items for batch processing

For Advanced Users

1. AI and LangChain:

- Build AI agents with custom tools - Implement RAG with vector databases - Use memory for conversational workflows - Chain multiple AI operations

1. Scaling and performance:

- Configure queue mode for distributed execution - Optimize database settings - Use execution data pruning - Configure task runners

1. Custom development:

- Create custom nodes - White label n8n for embedding - Use the n8n API for workflow management - Implement external secrets with AWS/Azure/GCP

1. Advanced patterns:

- Sub-workflows for reusability - Webhook authentication and validation - Complex data transformations with JMESPath - Real-time data processing with SSE/WebSockets

Common Issues and Solutions

HTTP Request Errors

• 400 Bad Request: Check query parameters and array formatting
• 403 Forbidden: Verify credentials and API permissions
• 429 Rate Limit: Use Batching or Retry on Fail options
• 404 Not Found: Verify endpoint URL is correct

Expression Errors

• Workflows now fail on expression errors in v1.0+
• Set up Error Workflows to catch failures
• Test expressions in the expression editor
• Check for undefined values before accessing properties

Data Type Issues

• Use .toString(), .toNumber() for type conversion
• Handle null/undefined with {{$json.field || 'default'}}
• Binary data requires special handling with buffers

Migration to v1.0

• New execution order (depth-first instead of breadth-first)
• Python support in Code node (Pyodide)
• Mandatory user management (no more BasicAuth)
• WebSocket push backend is now default
• Node 18.17.0 or higher required

Environment Configuration

Docker Setup

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

# With environment variables
docker run -it --rm \
  --name n8n \
  -p 5678:5678 \
  -e N8N_BASIC_AUTH_ACTIVE=true \
  -e N8N_BASIC_AUTH_USER=admin \
  -e N8N_BASIC_AUTH_PASSWORD=password \
  -v ~/.n8n:/home/node/.n8n \
  n8nio/n8n

Key Environment Variables

• N8N_HOST: Hostname (default: localhost)
• N8N_PORT: Port (default: 5678)
• N8N_PROTOCOL: http or https
• WEBHOOK_URL: External webhook URL
• N8N_ENCRYPTION_KEY: Encryption key for credentials
• DB_TYPE: Database type (sqlite, postgres)
• EXECUTIONS_MODE: queue or main (queue for scaling)

Resources

Official Resources

• n8n Documentation
• n8n Community Forum
• n8n Workflow Templates
• n8n GitHub Repository

Learning Paths

• Level One Course: Basic workflow building
• Level Two Course: Advanced data handling and error management
• Video Courses: Visual learning resources
• AI Tutorial: Build AI workflows from scratch

API and Development

• API Reference
• Creating Custom Nodes
• Node Development

Notes

• This skill was automatically generated from official n8n documentation
• Code examples use proper language tags for syntax highlighting
• Examples are extracted from real-world patterns in the docs
• Focus on practical, actionable patterns for immediate use

Updating

To refresh this skill with updated documentation:

1. Re-run the scraper with the same configuration
2. The skill will be rebuilt with the latest information
3. Review Quick Reference section for new examples