DEV Community: Leo Corbett

5 n8n Workflow Patterns I Use in Every Project

Leo Corbett — Sun, 22 Feb 2026 00:18:26 +0000

After building hundreds of n8n workflows, certain patterns keep proving their worth. Here are five I now use in almost every project.

1. Error Handler Sub-Workflow

Don't scatter error handling across your main workflow. Create a dedicated error handler:

Main Workflow → On Error → Call Error Handler Sub-Workflow

The error handler:

Logs the full error context
Sends alerts (Slack/email)
Stores failed items for retry
Tracks error frequency

One place to manage all error logic.

2. Config Node at Start

First node in every workflow: a Set node with configuration.

{
  "env": "production",
  "batchSize": 100,
  "retryAttempts": 3,
  "alertChannel": "#ops-alerts",
  "dryRun": false
}

Change behavior without editing logic. Toggle dry-run mode. Adjust batch sizes. All in one place.

3. Idempotency Checks

Before processing any item, check if it's already been handled:

// Check processed items table
const alreadyProcessed = await checkDatabase(item.id);
if (alreadyProcessed) {
  return []; // Skip
}

This prevents:

Duplicate emails
Double charges
Repeated API calls

Essential for workflows that might retry.

4. Batch + Delay Pattern

When hitting APIs with rate limits:

Split In Batches (10 items) → Process → Wait (1 second) → Loop

Simple but prevents 90% of rate limit issues. Adjust batch size and delay based on the API's limits.

5. Health Check Workflow

Separate workflow that monitors your other workflows:

Check execution history for failures
Verify expected runs happened
Test critical API connections
Alert if something's off

Run it every hour. Know about problems before users do.

Bonus: Logging Pattern

Every workflow should log:

Start time and trigger info
Items processed count
Any skipped items and why
Duration and outcome

Store in a simple database or even a Google Sheet. Invaluable for debugging.

Implementation Tips

Start simple. Don't add all patterns to every workflow. Add them as you need them.

Use sub-workflows. Error handlers, logging, and notifications should be reusable.

Document decisions. Sticky notes in n8n explaining WHY something works a certain way.

Version control. Export workflows regularly. Use git.

The Result

Workflows that:

Handle failures gracefully
Are easy to debug
Don't break under load
Can be modified safely

That's the difference between a prototype and production.

More production patterns at mardenseo.com/n8n/tutorials

n8n Pro Tip: Why HTTP Request Nodes Beat Integrated Nodes

Leo Corbett — Sun, 22 Feb 2026 00:14:31 +0000

Here's something that took me way too long to learn: in production n8n workflows, HTTP Request nodes are often better than the built-in integration nodes.

Sounds counterintuitive. Let me explain.

The Problem with Integrated Nodes

n8n has hundreds of pre-built nodes — Slack, HubSpot, Google Sheets, etc. They're great for getting started. But in production:

1. Limited API coverage
Integrated nodes expose maybe 20% of an API. Need a specific endpoint? You're stuck.

2. Update lag
APIs change. The node might be months behind the actual API.

3. Opaque errors
When something breaks, you get a generic error. Good luck debugging.

4. Rate limit handling
Most nodes don't handle rate limits gracefully.

HTTP Request Node: Full Control

The HTTP Request node gives you direct API access:

Method: POST
URL: https://api.service.com/v2/specific-endpoint
Headers: 
  Authorization: Bearer {{$credentials.apiKey}}
  Content-Type: application/json
Body: {{JSON.stringify($json)}}

Benefits:

Access ANY endpoint
See exact request/response
Custom error handling
Rate limit retry logic
Latest API features immediately

Real Example: Slack

The Slack node lets you post messages. Basic.

With HTTP Request, you can:

Use Block Kit for rich formatting
Schedule messages
Access the entire Slack API
Handle rate limits properly

POST https://slack.com/api/chat.postMessage
{
  "channel": "C0123456",
  "blocks": [...],
  "metadata": {...}
}

When to Use Integrated Nodes

They're still useful for:

Quick prototyping
Simple operations
OAuth flows (easier setup)
Learning what's possible

My Rule

Prototype with integrated nodes. Production uses HTTP Request.

The extra 5 minutes setting up HTTP Request saves hours debugging weird node behavior later.

Error Handling Pattern

Pair HTTP Request with proper error handling:

// In a Function node after HTTP Request
const response = $input.first().json;

if (response.error) {
  // Handle API error
  throw new Error(`API failed: ${response.error.message}`);
}

// Check rate limits
if (response.headers?.['x-ratelimit-remaining'] === '0') {
  // Trigger retry logic
}

return $input.all();

Credentials Tip

Store API keys in n8n credentials (Header Auth or Custom Auth), not hardcoded. Reference with:

{{$credentials.myApi.apiKey}}

Keeps secrets out of your workflow JSON.

Summary

Integrated nodes: good for learning, prototyping
HTTP Request: good for production, edge cases, full API access
Always add error handling
Use n8n credentials for secrets

Once you get comfortable with HTTP Request, you'll wonder why you ever relied on integrated nodes.

More n8n production patterns at mardenseo.com/n8n/tutorials

I Replaced $300/Month in Zapier Fees with Self-Hosted n8n

Leo Corbett — Sun, 22 Feb 2026 00:07:25 +0000

Last year I was paying $300/month for Zapier. Today I pay $5/month for a VPS running n8n. Here's exactly how I made the switch.

The Breaking Point

Zapier's pricing scales with tasks. When my workflows started processing 10,000+ tasks monthly, costs exploded. The final straw was hitting a rate limit during a critical data sync.

I needed:

Unlimited tasks
No rate limits
Full control over my data
Better error handling

n8n checked every box.

The Migration Plan

Week 1: Audit Existing Workflows

I documented every Zapier workflow:

Trigger type
Actions performed
Data transformed
Error handling (or lack thereof)

Key insight: 80% of my Zapier usage was 5 workflows. Start there.

Week 2: Set Up n8n

Option 1: n8n Cloud (easier)

Managed hosting
Still cheaper than Zapier at scale
Good for testing

Option 2: Self-hosted (what I did)

$5/month VPS (Hetzner)
Docker deployment
Full control

The setup:

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

That's it. n8n running in 30 seconds.

Week 3-4: Rebuild Core Workflows

The five workflows I rebuilt first:

1. Lead capture → CRM sync
Zapier: Typeform → Zapier → HubSpot
n8n: Webhook → Transform → HubSpot API

2. Invoice notifications
Zapier: Stripe → Zapier → Slack + Email
n8n: Stripe webhook → Split → Slack + SMTP

3. Content publishing
Zapier: Google Docs → Zapier → WordPress
n8n: Drive trigger → Markdown convert → WordPress API

4. Data backup
Zapier: Daily schedule → Airtable → Google Sheets
n8n: Cron → Airtable → Sheets (unlimited rows!)

5. Support ticket routing
Zapier: Email → Zapier → Categorize → Assign
n8n: IMAP → AI classification → Ticket system

What's Better in n8n

Unlimited Everything

No task limits. No throttling. Run 100,000 tasks/month on a $5 VPS.

Better Error Handling

n8n's error workflows are game-changing. When something fails:

Automatic retry with backoff
Error branch for custom handling
Detailed logs with full payloads

Zapier's error handling felt like an afterthought.

Self-Hosted Data

My data never leaves my server. For GDPR compliance and general paranoia, this matters.

HTTP Request Node

This single node replaced dozens of Zapier's premium integrations. Any API works:

Method: POST
URL: https://api.whatever.com/endpoint
Headers: Authorization: Bearer {{$credentials.apiKey}}
Body: {{JSON.stringify($json)}}

No waiting for Zapier to build an integration.

Code When You Need It

When visual nodes aren't enough, add JavaScript:

const items = $input.all();
return items.map(item => ({
  ...item.json,
  processedAt: new Date().toISOString(),
  score: calculateScore(item.json)
}));

What's Harder in n8n

Initial Learning Curve

Zapier's UI is more polished for beginners. n8n requires understanding data flow between nodes.

My tip: Spend 2 hours on the official tutorials before migrating anything.

Self-Hosting Maintenance

You're responsible for:

Backups
Updates
SSL certificates
Uptime monitoring

Most of this is automated with proper setup, but it's not zero-effort.

Some Integrations Need Work

Zapier has 5,000+ integrations. n8n has fewer native nodes, but HTTP Request covers the gap.

The Math

Before (Zapier):

Professional plan: $299/month
50,000 tasks included
Overage charges when I exceeded

After (n8n self-hosted):

Hetzner VPS: $5/month
Unlimited tasks
Better features

Annual savings: $3,528

Even n8n Cloud at $50/month would save $2,988/year.

Should You Switch?

Switch to n8n if:

You're hitting Zapier's limits
You process 10,000+ tasks monthly
You need better error handling
Data privacy matters
You're comfortable with slight technical overhead

Stay with Zapier if:

You run <5 simple workflows
Zero technical maintenance is essential
You need very specific native integrations

Getting Started

Try n8n Cloud free — No commitment, see if you like it
Audit your Zapier workflows — Which 5 matter most?
Rebuild one workflow — Start with the simplest
Run parallel — Keep Zapier active while testing
Cut over gradually — Don't rush the migration

The switch took me a month of part-time work. The savings pay for themselves in 2 weeks.

Need help migrating from Zapier to n8n? Marden SEO builds production n8n workflows for businesses.

How to Use MCP Servers with n8n for AI-Powered Automation

Leo Corbett — Sat, 21 Feb 2026 23:59:56 +0000

The Model Context Protocol (MCP) is reshaping how AI agents interact with external tools. Major cloud providers — AWS, Azure, Google Cloud — now offer MCP servers, and n8n users can leverage this for genuinely intelligent automation.

Here's how MCP works and why it matters for your n8n workflows.

What is MCP and Why Should n8n Users Care?

MCP is a standardized protocol that lets AI models discover and use tools dynamically. Instead of hardcoding every API call, you give an AI agent access to an MCP server, and it figures out which tools to use based on the task.

Traditional n8n workflow:

Trigger → HTTP Request (hardcoded API) → Transform → Output

MCP-enabled AI agent:

Trigger → AI Agent → [dynamically selects tools from MCP server] → Output

The difference: your workflow adapts to new requirements without rebuilding it.

The MCP Server Landscape (2026)

Major players now offer MCP servers:

Cloud Provider MCP Servers

AWS MCP: S3, Lambda, DynamoDB tool access
Azure MCP: Office 365, Azure Functions integration
Google Cloud MCP: BigQuery, Cloud Functions, Workspace

Specialized MCP Servers

Apify MCP: Web scraping, data extraction, browser automation
Playwright MCP: Headless browser control for testing and scraping
Amazon Ads MCP: Campaign management, reporting, bid optimization

Enterprise MCP

Microsoft Work IQ: Full M365 context for AI agents
Virtana MCP Server 2.0: Infrastructure observability tools

Integrating MCP with n8n

Method 1: HTTP Request Node

For simple MCP calls, use n8n's HTTP Request node:

URL: https://mcp-server.example.com/tools/list
Method: GET
Headers: Authorization: Bearer {{your_api_key}}

The response lists available tools. Then make subsequent calls to execute specific tools.

Method 2: AI Agent Node + Custom Tools

The more powerful approach — let n8n's AI Agent discover and use MCP tools:

Create a sub-workflow that calls the MCP server's tool listing endpoint
Bind it as a custom tool to your AI Agent node
Add an executor tool that runs MCP tool calls

Your agent can now browse available tools and use them based on the task.

Example: AI Research Agent with Apify MCP

Use case: Extract competitor pricing data from multiple websites.

Workflow:

Webhook trigger receives research request
AI Agent node with Apify MCP tools bound
Agent decides which scrapers to run based on target sites
Results aggregated and stored

Agent system prompt:

You are a competitive intelligence researcher.
When given a research request:
1. Identify target websites
2. Select appropriate Apify actors for each site type
3. Run extractions and compile results
4. Return structured pricing data

The agent handles site-specific logic without you hardcoding each scraper.

MCP vs Traditional Zapier-Style Automation

Aspect	Traditional	MCP-Enabled
Tool selection	Hardcoded	Dynamic
Adding new integrations	Rebuild workflow	Add to MCP server
Error handling	Per-step logic	Agent-level reasoning
Maintenance	High	Lower
Best for	Predictable flows	Variable tasks

When to use traditional workflows:

High-volume, identical operations
Strict compliance requirements
Simple trigger → action patterns

When to use MCP:

Research and analysis tasks
Multi-tool operations where the path varies
Rapidly evolving integration needs

Practical Use Cases

Automated Ad Campaign Management

Connect Amazon Ads MCP to your AI agent. It can:

Pull performance reports
Identify underperforming campaigns
Adjust bids based on rules you define
Generate optimization recommendations

Document Processing Pipeline

Combine multiple MCP servers:

Google Drive MCP: Access source documents
OCR MCP: Extract text from images/PDFs
Translation MCP: Handle multilingual content
Your database: Store processed results

One agent orchestrates the entire flow.

Infrastructure Monitoring

Using Virtana MCP Server 2.0:

Agent monitors system metrics
Detects anomalies using its reasoning
Pulls relevant logs via MCP tools
Compiles incident reports with context

Getting Started

Pick an MCP server relevant to your use case (Apify is good for web tasks)
Test the API manually to understand the tool schema
Build a simple n8n workflow calling one MCP tool
Wrap it in an AI Agent once you understand the patterns
Iterate — add more tools as needed

MCP isn't magic, but it does shift complexity from workflow design to prompt engineering. For variable, multi-step tasks, that's often a better tradeoff.

Need help integrating MCP servers with your n8n automation? Marden SEO builds AI-powered workflows for businesses.

Building AI Agent Workflows in n8n: The 2026 Complete Guide

Leo Corbett — Sat, 21 Feb 2026 23:59:28 +0000

AI agents aren't just hype anymore — they're production-ready tools that can handle complex, multi-step tasks autonomously. n8n's native AI Agent node makes building these systems accessible without writing code.

This guide covers everything from basic agent concepts to advanced production patterns.

AI Agents vs Traditional Workflows: What's the Difference?

Traditional n8n workflows are sequential: trigger → process → output. You define every step explicitly.

AI agents are autonomous: you give them a goal and tools, and they figure out the steps themselves.

When to use traditional workflows:

Predictable, repeatable processes
High-volume batch operations
Strict compliance requirements

When to use AI agents:

Variable input requiring judgment
Multi-step research tasks
Complex decision trees that would be painful to hardcode

n8n's AI Agent Node: Architecture Deep Dive

The AI Agent node in n8n consists of four core components:

1. The Brain (LLM Connection)

Connect to OpenAI GPT-4, Anthropic Claude, or local models via Ollama. The model handles reasoning and decides which tools to use.

Supported providers:
- OpenAI (GPT-4, GPT-4 Turbo)
- Anthropic (Claude 3 Opus, Sonnet, Haiku)
- Azure OpenAI
- Ollama (local models)
- Google Gemini

2. Memory (Context Management)

Agents need memory to maintain context across interactions:

Window Memory: Keeps last N messages (simple, token-efficient)
Buffer Memory: Stores full conversation (good for short sessions)
Vector Store Memory: Semantic search over history (best for long-running agents)

3. Tools (Agent Capabilities)

Tools are what make agents powerful. n8n provides:

HTTP Request: Call any API
Code: Execute JavaScript/Python
Calculator: Math operations
Wikipedia: Knowledge lookup
Custom Tools: Define your own via sub-workflows

4. System Prompt (Agent Personality)

The system prompt defines behavior, constraints, and output format. This is where you shape the agent's personality and rules.

Step-by-Step: Build Your First AI Agent

Let's build a lead qualification agent that:

Takes a company name
Researches the company online
Scores the lead
Returns a structured assessment

Prerequisites

n8n 1.20+ (AI nodes included)
OpenAI API key (or alternative LLM)

The Workflow

1. Trigger (Webhook)
Receives company name via POST request.

2. AI Agent Node
Configure:

Model: GPT-4 Turbo
Tools: HTTP Request (for web lookups)
Memory: Window (5 messages)

System Prompt:

You are a B2B lead qualification specialist. 

When given a company name:
1. Research the company's industry, size, and recent news
2. Score the lead 1-10 based on fit for automation services
3. Return a structured assessment

Output format:
{
  "company": "...",
  "industry": "...",
  "employee_count": "...",
  "score": 8,
  "reasoning": "...",
  "next_steps": "..."
}

3. Parse & Store
Extract the JSON from agent output and store in your CRM.

Advanced Agent Patterns

Multi-Tool Agents

Give agents multiple tools and let them choose:

Tools available:
- search_web: Find current information
- query_database: Look up customer records  
- send_email: Contact the customer
- create_ticket: Log issues

The agent decides the sequence based on the task.

Agent Chains

Connect multiple specialized agents:

Research Agent → gathers information
Analysis Agent → processes findings
Writer Agent → creates final output

Each agent focuses on what it's best at.

RAG Integration

Combine agents with retrieval-augmented generation:

User query comes in
Vector search finds relevant documents
Agent receives query + context
Agent reasons over the combined information

This pattern is powerful for customer support, documentation Q&A, and knowledge work.

Performance & Cost Optimization

AI agents can get expensive fast. Here's how to control costs:

Token Monitoring

Track usage per workflow run:

Log input/output tokens
Set up alerts for anomalies
Review high-cost runs weekly

Caching Strategies

Cache common lookups:

Store API responses in Redis
Use n8n's built-in caching for static data
Implement TTL based on data freshness needs

Model Selection

Not every task needs GPT-4:

GPT-4: Complex reasoning, nuanced judgment
GPT-3.5 Turbo: Simple classification, extraction
Claude Haiku: Fast, cheap, good for filtering

Route tasks to the cheapest model that can handle them.

Real-World Use Cases

Customer Support Triage

Agent reads incoming tickets, categorizes urgency, routes to correct team, and drafts initial responses for common issues.

Content Research

Agent takes a topic, searches multiple sources, synthesizes findings, and outputs a structured brief with citations.

Data Transformation

Agent handles messy input data — varying formats, missing fields, inconsistent naming — and normalizes it for downstream systems.

Monitoring & Alerting

Agent monitors metrics, detects anomalies, investigates root causes via tool calls, and sends contextual alerts.

Getting Started

The fastest way to learn is to build something real:

Pick a repetitive task you do manually
Break it into steps
Identify what tools the agent needs
Start simple — one tool, basic prompt
Iterate based on failures

AI agents aren't magic. They're tools that require good prompts, appropriate guardrails, and iteration. But when built right, they handle complexity that would be painful to automate traditionally.

Need help building AI agent workflows for your business? Marden SEO specializes in n8n automation and AI integration.