n8n Marketing Flows Skill

Skill by ara.so — Marketing Skills collection.

What It Is

n8n-marketing-flows is a collection of 79+ ready-to-import n8n workflow templates for marketing automation. It covers:

Social media publishing and hashtag generation
Sentiment monitoring (PTT, social platforms)
News aggregation and AI summaries
Ad management (auto-pause low performers)
AI content generation (posts, images, video)
Performance analytics and reporting

Each workflow is a JSON file you import into n8n (self-hosted or cloud), connect your API keys, and activate.

Two versions per workflow:

*.json — Standard (uses Claude/OpenAI + Email/Slack)
*.local-ollama.json — Free local (uses local Ollama, no API keys needed)
*.skeleton.json — Credential template (architecture ready, you fill in API keys)

Installation

1. Set Up n8n

Option A: Docker (self-hosted)

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

Access at http://localhost:5678

Option B: n8n Cloud Sign up at https://n8n.io (free tier available)

Option C: npm

npm install -g n8n
n8n start

2. Clone the Template Repository

git clone https://github.com/YuriCrystal/n8n-marketing-flows.git
cd n8n-marketing-flows

3. Import a Workflow

In n8n UI: ⋯ menu (top right) → Import from File
Select a JSON file from templates/
Configure credentials (see below)
Click Save and Activate

Project Structure

n8n-marketing-flows/
├── templates/
│   ├── 01-weekly-ai-news-digest.json
│   ├── 01-weekly-ai-news-digest.local-ollama.json
│   ├── 02-social-hashtag-generator.json
│   ├── 04-ptt-sentiment-monitor.local-ollama.json
│   ├── 90-viral-social-post.json
│   ├── 91-webhook-bridge.json
│   └── [59 more *.skeleton.json templates]
├── docs/
│   ├── 01-weekly-ai-news-digest.md
│   └── 91-webhook-bridge.md
├── ROADMAP.md
├── CREDITS.md
└── README.md

Key Workflow Types

1. Free Local Workflows (19 files, `*.local-ollama.json`)

Uses Ollama (free, runs locally) + public data sources (RSS, web scraping). No API keys required.

Example: Weekly AI News Digest

# Start Ollama locally
ollama pull llama3.2

# Import templates/01-weekly-ai-news-digest.local-ollama.json
# Configure:
# - RSS feeds: TechCrunch, Hacker News, etc.
# - Schedule: Every Monday 9am
# - Output: Discord webhook or n8n display

2. Standard Workflows (uses Claude/OpenAI)

Requires Anthropic or OpenAI API key.

Setup credentials in n8n:

Settings → Credentials → Add → Anthropic/OpenAI
API Key: <your key from env>
Default Model: claude-haiku-4-5 (or gpt-4o-mini)

3. Platform Integration Skeletons (59 files, `*.skeleton.json`)

Pre-built architecture for:

Meta Graph API (Facebook, Instagram, Threads)
Google Sheets (data export)
YouTube Data API (analytics)
WordPress REST API (auto-publish)
Meta Marketing API (ad management)

These require YOUR platform API keys — see credential sections below.

Core Workflows

Social Media Posting

Workflow 02: Generate Hashtags for Post

// Input: Post text + platform (Instagram/Facebook/Threads)
// Output: Optimized hashtags

// Usage in n8n:
Webhook → HTTP Request (LLM) → Set (format) → Respond

Example prompt to LLM node:

// In "Prompt" field of HTTP Request node:
`Generate 5-10 hashtags for this {{ $json.platform }} post:

"{{ $json.postText }}"

Rules:
- Instagram: Mix popular + niche, under 30 chars each
- Facebook: 3-5 broad hashtags
- Threads: Conversational, 5-7 hashtags

Return only hashtags, space-separated.`

Import templates/02-social-hashtag-generator.json and customize the prompt.

Sentiment Monitoring

Workflow 04: PTT Forum Monitor

Scrapes PTT (Taiwan's largest forum), detects mentions, sends alerts.

// In HTTP Request node (scrape PTT):
{
  "method": "GET",
  "url": "https://www.ptt.cc/bbs/{{ $json.board }}/index.html",
  "headers": {
    "Cookie": "over18=1"  // Age verification
  }
}

// In Code node (filter mentions):
const keyword = "{{ $json.brandName }}";
const posts = $input.all();
return posts.filter(p => 
  p.json.title.includes(keyword) || 
  p.json.content.includes(keyword)
);

Import templates/04-ptt-sentiment-monitor.local-ollama.json

AI Content Generation

Workflow 90: Viral Social Post Generator

Uses反主流 (counter-mainstream) formula from Hao's research (see CREDITS.md).

// LLM prompt structure:
`You are a social media expert. Generate a viral post using this formula:

1. **Hook** (first line): Controversial or counter-intuitive
2. **Contrast**: "Everyone says X, but actually Y"
3. **Story**: Personal anecdote or data
4. **CTA**: "What's your experience?"

Topic: {{ $json.topic }}
Platform: {{ $json.platform }}
Tone: {{ $json.tone }}

Output 3 variations under 280 chars each.`

Import templates/90-viral-social-post.json or .local-ollama.json version.

Webhook Bridge for External Tools

Workflow 91: Webhook Bridge

Allows AI agents, Zapier, Make.com, or scripts to call n8n workflows via HTTP.

# Activate workflow 91, get webhook URL:
# https://your-n8n.com/webhook/marketing-flow

# Call from code:
curl -X POST https://your-n8n.com/webhook/marketing-flow \
  -H "Content-Type: application/json" \
  -d '{
    "action": "generatePost",
    "topic": "AI automation trends",
    "platform": "linkedin"
  }'

# Returns JSON with generated content

In n8n workflow:

// Webhook node receives POST
// Switch node routes by $json.action
// LLM node generates content
// Respond to Webhook node returns JSON

Configuration Patterns

Switching LLM Providers

Claude (default)

// HTTP Request node:
{
  "url": "https://api.anthropic.com/v1/messages",
  "method": "POST",
  "headers": {
    "x-api-key": "{{ $credentials.anthropicApiKey }}",
    "anthropic-version": "2023-06-01"
  },
  "body": {
    "model": "claude-haiku-4-5",
    "messages": [{"role": "user", "content": "{{ $json.prompt }}"}]
  }
}

OpenAI

{
  "url": "https://api.openai.com/v1/chat/completions",
  "method": "POST",
  "headers": {
    "Authorization": "Bearer {{ $credentials.openaiApiKey }}"
  },
  "body": {
    "model": "gpt-4o-mini",
    "messages": [{"role": "user", "content": "{{ $json.prompt }}"}]
  }
}

Local Ollama (free)

{
  "url": "http://localhost:11434/api/generate",
  "method": "POST",
  "body": {
    "model": "llama3.2",
    "prompt": "{{ $json.prompt }}",
    "stream": false
  }
}

Notification Channels

Email (SMTP)

// Send Email node:
To: process.env.ALERT_EMAIL
Subject: [Alert] {{ $json.title }}
Body: {{ $json.summary }}

Discord Webhook

# Get webhook URL from Discord server settings
# In HTTP Request node:
POST https://discord.com/api/webhooks/YOUR_WEBHOOK_ID/TOKEN
{
  "content": "🚨 New mention detected",
  "embeds": [{
    "title": "{{ $json.title }}",
    "url": "{{ $json.link }}",
    "color": 15258703
  }]
}

Slack

// Slack node (requires Slack app):
Channel: #marketing-alerts
Message: :warning: {{ $json.alert }}

Scheduling

// Cron node examples:
// Every Monday 9am:
0 9 * * 1

// Every 6 hours:
0 */6 * * *

// Weekdays at 10am:
0 10 * * 1-5

Credential Setup

Anthropic (Claude)

# Get key: https://console.anthropic.com/settings/keys
# In n8n: Credentials → Anthropic API
API_KEY=your_anthropic_key

OpenAI

# Get key: https://platform.openai.com/api-keys
# In n8n: Credentials → OpenAI
API_KEY=your_openai_key

Meta Graph API (Facebook/Instagram)

# Create app: https://developers.facebook.com
# Get access token with pages_read_engagement, pages_manage_posts
# In n8n: Credentials → HTTP Request (Bearer Token)
ACCESS_TOKEN=your_meta_token

Use in workflow:

// Get Instagram posts:
GET https://graph.facebook.com/v18.0/me/media
  ?fields=id,caption,like_count,timestamp
  &access_token={{ $credentials.metaToken }}

Google Sheets

# In n8n: Credentials → Google Sheets API → OAuth2
# Follow OAuth flow, authorize scopes:
# https://www.googleapis.com/auth/spreadsheets

Common Use Cases

1. Weekly News Summary

# Import: templates/01-weekly-ai-news-digest.local-ollama.json

# Modify RSS feeds in RSS Read node:
- https://techcrunch.com/feed/
- https://news.ycombinator.com/rss
- https://www.theverge.com/rss/index.xml

# Change schedule in Cron node: 0 9 * * 1 (Monday 9am)

# Output to Discord webhook:
POST https://discord.com/api/webhooks/YOUR_ID/TOKEN
{
  "content": "📰 This week's AI news:\n\n{{ $json.summary }}"
}

2. Auto-Pause Low-Performing Ads

# Import: templates/fb-ad-auto-pause.skeleton.json

# Configure Meta Marketing API credentials
# Set threshold in IF node:
{{ $json.ctr }} < 0.01  // Click-through rate below 1%

# HTTP Request node:
POST https://graph.facebook.com/v18.0/{{ $json.adId }}
{
  "status": "PAUSED",
  "access_token": "{{ $credentials.metaToken }}"
}

3. Multi-Platform Post Scheduler

# Import: templates/05-multi-platform-post-draft.json

# Input: Topic + platforms (array: ["facebook", "instagram", "linkedin"])

# LLM generates platform-specific versions:
# - Facebook: Conversational, 280 chars
# - Instagram: Visual-first, 10 hashtags
# - LinkedIn: Professional, 500 chars

# Output: JSON array of drafts → Save to Google Sheets or send to Slack

Troubleshooting

"Missing credentials" Error

Solution: Click nodes with 🔑 icon, fill in API keys.

# Check required credentials for each workflow:
grep -r "credentials" templates/01-*.json

# Common ones:
- Anthropic API (Claude)
- OpenAI API
- SMTP (Email)
- Meta Graph API
- Google OAuth2

Ollama Connection Failed

# Ensure Ollama is running:
ollama serve

# Test connection:
curl http://localhost:11434/api/generate -d '{
  "model": "llama3.2",
  "prompt": "Test"
}'

# Pull model if missing:
ollama pull llama3.2

RSS Feed Not Updating

# Check feed URL in RSS Read node
# Validate feed: https://validator.w3.org/feed/

# Increase "Fetch Interval" in Cron node:
0 */2 * * *  # Every 2 hours instead of daily

Rate Limits (Meta, OpenAI)

Solution: Add Wait node between API calls.

// After HTTP Request node:
Wait node → Time: 1 second

// Or use Batch node to process in chunks:
Batch Size: 10

LLM Returns Empty Response

# Check prompt in HTTP Request body:
- Ensure {{ $json.prompt }} variable exists
- Test prompt in Expression editor (=)
- Increase max_tokens in API call:
  "max_tokens": 500

Advanced Patterns

Dynamic Prompt Engineering

// Use Code node to build context-aware prompts:
const platform = $json.platform;
const tone = $json.tone || 'professional';

const prompts = {
  instagram: `Create a visual-first post with emojis...`,
  linkedin: `Write a thought leadership piece...`,
  twitter: `Craft a controversial take under 280 chars...`
};

return {
  json: {
    prompt: prompts[platform] + `\n\nTone: ${tone}\nTopic: ${$json.topic}`
  }
};

Multi-Step AI Workflows

// Chain multiple LLM calls:
1. Generate outline (Claude)
2. Expand each point (GPT-4)
3. Add SEO keywords (local Llama)
4. Format for platform (Claude)

// Use Set node to pass data between steps:
Set node after each LLM → Store in {{ $json.step1Output }}

Error Handling

// Add Error Trigger node:
On Error → Send alert to Slack → Retry workflow

// In HTTP Request node:
{
  "continueOnFail": true,
  "retry": {
    "maxRetries": 3,
    "waitBetweenRetries": 5000
  }
}

Environment Variables

Set in n8n Docker:

docker run -e ANTHROPIC_API_KEY=sk-xxx \
  -e OPENAI_API_KEY=sk-xxx \
  -e ALERT_EMAIL=team@company.com \
  n8nio/n8n

Reference in workflows:

// In expressions:
{{ $env.ANTHROPIC_API_KEY }}
{{ $env.ALERT_EMAIL }}

Resources

Full documentation: https://github.com/YuriCrystal/n8n-marketing-flows/tree/main/docs
n8n docs: https://docs.n8n.io
Ollama models: https://ollama.com/library
Workflow examples: /templates folder (79 files)

License

MIT — Freely use, modify, and distribute. Attribution required.

n8n-marketing-flows