How to Build an AI-Powered Content Pipeline with n8n and Claude

DOCKER COMPOSE SETUP

# docker-compose.yml
version: "3.8"
services:
  n8n:
    image: n8nio/n8n:latest
    restart: always
    ports:
      - "5678:5678"
    environment:
      - N8N_BASIC_AUTH_ACTIVE=true
      - N8N_BASIC_AUTH_USER=admin
      - N8N_BASIC_AUTH_PASSWORD=your-secure-password
      - N8N_HOST=your-domain.com
      - N8N_PROTOCOL=https
      - WEBHOOK_URL=https://your-domain.com/
      - GENERIC_TIMEZONE=America/New_York
    volumes:
      - n8n_data:/home/node/.n8n
volumes:
  n8n_data:

n8n CREDENTIAL SETUP

Step 1: Get your API key
  - Go to console.anthropic.com
  - Navigate to API Keys
  - Create a new key, copy it immediately

TOKEN COST CALCULATOR

WORKFLOW NODE CHAIN

Trigger (Cron: daily at 6 AM)
  ↓
Google Sheets Node (read next topic from content calendar)
  → Reads: keyword, target_audience, content_type, word_count
  ↓
HTTP Request #1 -- OUTLINE (Claude Sonnet)
  System prompt: "You are an expert content strategist..."
  User prompt: "Create a detailed outline for an article
    about [keyword] targeting [audience].
    Include H2/H3 structure, key points per section,
    and internal linking opportunities."
  ↓
Split In Batches Node (split outline into sections)
  ↓
HTTP Request #2 -- SECTION DRAFT (Claude Sonnet, looped)
  System prompt: "You are an expert content writer.
    Match this brand voice: [voice_doc].
    Write only the section described. No intro/outro."
  User prompt: "Write section: [section_title]
    Key points: [section_points]
    Target word count: [section_words]
    Previous sections for context: [prior_sections]"
  ↓
Merge Node (combine all sections into full article)
  ↓
HTTP Request #3 -- POLISH (Claude Sonnet)
  System prompt: "You are a senior editor..."
  User prompt: "Polish this article for flow and
    consistency. Fix transitions between sections.
    Ensure the introduction hooks the reader and
    the conclusion provides clear next steps.
    Article: [full_draft]"

SEO OPTIMIZATION NODES

Full Article Draft (from Step 3)
  ↓
HTTP Request -- KEYWORD OPTIMIZATION (Claude Haiku)
  Prompt: "Analyze this article for the target keyword
    '[keyword]'. Return JSON:
    {
      'keyword_count': number,
      'keyword_density': percentage,
      'keyword_in_h2': boolean,
      'keyword_in_first_100_words': boolean,
      'suggested_lsi_keywords': string[],
      'missing_keyword_placements': string[]
    }"
  ↓
IF Node (keyword_density < 0.8% or > 2.5%)
  → Yes: HTTP Request -- REOPTIMIZE (Claude Haiku)
    "Naturally integrate the keyword '[keyword]'
     into these sections: [missing_placements]"
  → No: Continue
  ↓
HTTP Request -- META TAGS (Claude Haiku)
  Prompt: "Generate SEO meta tags for this article.
    Target keyword: [keyword]. Return JSON:
    {
      'meta_title': max 60 chars with keyword,
      'meta_description': max 155 chars with keyword
        and call-to-action,
      'og_title': string,
      'og_description': string,
      'slug': kebab-case URL slug
    }"
  ↓
HTTP Request -- SCHEMA MARKUP (Claude Haiku)
  Prompt: "Generate JSON-LD schema markup for this
    article. Include Article schema and FAQPage
    schema if the article contains FAQ content.
    Return valid JSON-LD."
  ↓
HTTP Request -- INTERNAL LINKS (Claude Haiku)
  Prompt: "Given this list of existing articles on
    the site: [article_list]. Suggest 3-5 internal
    links to add to this article. Return JSON:
    {
      'links': [{'anchor_text': '', 'target_url': '',
        'placement_context': ''}]
    }"

QUALITY SCORING WORKFLOW

SEO-Optimized Article (from Step 4)
  ↓
HTTP Request -- QUALITY SCORE (Claude Haiku)
  System prompt: "You are a senior content editor.
    Score this article objectively."
  User prompt: "Score this article 1-10 on each
    criterion. Return JSON only:
    {
      'readability': number,
      'factual_confidence': number,
      'originality': number,
      'structure': number,
      'seo_compliance': number,
      'overall_score': number (average),
      'issues': string[] (specific problems found),
      'low_confidence_claims': string[]
    }

WORDPRESS REST API PUBLISH NODE

HTTP Request Node -- PUBLISH TO WORDPRESS
  URL: https://your-site.com/wp-json/wp/v2/posts
  Method: POST
  Authentication: Basic Auth (Application Password)
  Headers:
    Content-Type: application/json
  Body (JSON):
  {
    "title": "{{ $json.meta_title }}",
    "content": "{{ $json.article_html }}",
    "status": "draft",
    "slug": "{{ $json.slug }}",
    "categories": [{{ $json.category_id }}],
    "meta": {
      "_yoast_wpseo_title": "{{ $json.meta_title }}",
      "_yoast_wpseo_metadesc": "{{ $json.meta_description }}"
    },
    "date": "{{ $json.scheduled_date }}"
  }

CMS COMPARISON FOR AUTO-PUBLISHING

MONITORING WORKFLOW

Trigger (Cron: weekly, Monday 9 AM)
  ↓
Google Sheets Node (read published articles list)
  → Reads: article_url, publish_date, target_keyword
  ↓
HTTP Request -- GSC API (for each article)
  URL: https://searchconsole.googleapis.com/
    webmasters/v3/sites/[site]/searchAnalytics/query
  Body: {
    "startDate": "[7_days_ago]",
    "endDate": "[today]",
    "dimensions": ["query", "page"],
    "dimensionFilterGroups": [{
      "filters": [{
        "dimension": "page",
        "expression": "[article_url]"
      }]
    }]
  }
  ↓
Function Node -- ANALYZE PERFORMANCE
  // Categorize articles:
  // - High CTR (>5%): Winning format, replicate
  // - High impressions, low CTR: Rewrite title/meta
  // - Low impressions: Needs backlinks or topic shift
  // - Position 4-10: Optimize for featured snippets
  ↓
IF Node (needs_optimization == true)
  → Yes: Trigger content update workflow
    → Fetch current article from CMS
    → Claude Sonnet: "This article ranks position
      [pos] for [keyword]. Current title: [title].
      Impressions: [imp], clicks: [clicks], CTR: [ctr].
      Suggest improvements to increase CTR and ranking."
    → Update article via CMS API
  → No: Log performance data
  ↓
Google Sheets Node -- UPDATE PERFORMANCE LOG
  → Write: impressions, clicks, CTR, avg_position
  ↓
HTTP Request -- TOPIC INSIGHTS (Claude Haiku)
  Prompt: "Based on this performance data for our
    published articles: [performance_summary].
    Which topics should we double down on?
    Which keywords show opportunity (high impressions,
    low CTR)? Suggest 10 new article topics
    informed by this data."
  ↓
Google Sheets Node -- ADD TO CONTENT CALENDAR
  → Write suggested topics to content calendar