ncreighton Overview
AI-Driven Content Taxonomy and Tagging automatically analyzes and categorizes your content using advanced machine learning models. This skill transforms unorganized content into a discoverable, searchable knowledge base by applying intelligent taxonomies, hierarchical tags, and semantic metadata—all without manual effort.
Why This Matters
- Content Discoverability: Improve search rankings and user navigation across WordPress, Contentful, and Shopify
- SEO Optimization: Auto-generate schema markup, meta tags, and structured data
- Multi-Channel Publishing: Maintain consistent taxonomy across Slack, email, web, and social platforms
- Time Savings: Eliminate manual tagging workflows that consume 20+ hours/month
- Content Intelligence: Discover content gaps, trending topics, and audience intent patterns
Supported Integrations
- CMS: WordPress, Ghost, Contentful, Strapi, Webflow
- Communication: Slack, Microsoft Teams, Discord
- Databases: MongoDB, PostgreSQL, Airtable
- Search: Elasticsearch, Algolia, Meilisearch
- Cloud Storage: Google Drive, Dropbox, AWS S3
- Analytics: Google Analytics 4, Mixpanel, Segment
Quick Start
Example 1: Tag a Blog Post Automatically
Analyze this blog post and apply a comprehensive taxonomy:
Title: "How to Build a Sustainable E-commerce Business in 2024"
Content: [paste article body]
Use these taxonomy categories:
- Industry: [e-commerce, sustainability, business]
- Format: [guide, tutorial, best-practices]
- Audience: [entrepreneurs, business-owners, startups]
- Topics: [auto-detect from content]
- Reading Level: [intermediate, advanced]
Return JSON with tags, confidence scores, and SEO meta tags.
Example 2: Bulk Tag WordPress Content
Connect to my WordPress site (wp.example.com) and:
1. Retrieve all posts from the past 6 months
2. Apply AI taxonomy to each post
3. Auto-assign WordPress categories and tags
4. Generate meta descriptions and schema markup
5. Return a CSV report with tagging results and coverage %
Use my custom taxonomy: [list your categories]
Example 3: Create a Content Map
Analyze my content library (100+ articles) and:
1. Identify all unique topics and subtopics
2. Create a hierarchical taxonomy tree
3. Map content to audience personas
4. Highlight content gaps and overlap
5. Recommend new content opportunities
Export as an interactive JSON structure and Markdown outline.
Capabilities
1. Intelligent Content Analysis
- Multi-dimensional Tagging: Simultaneously categorize by topic, format, audience, industry, intent, and custom dimensions
- Semantic Understanding: Uses transformer-based NLP to understand context beyond keywords
- Confidence Scoring: Every tag includes a 0-100 confidence score for quality control
- Multi-language Support: Analyze content in 50+ languages and tag in your preferred language
2. Custom Taxonomy Management
- Build Custom Taxonomies: Define hierarchical category trees specific to your business
- Dynamic Tag Suggestions: AI learns from your existing tags and suggests new ones
- Taxonomy Versioning: Track taxonomy changes and maintain backward compatibility
- Synonym Management: Map related terms (e.g., "AI" = "artificial intelligence" = "machine learning")
3. Content Organization
- Bulk Processing: Tag 100+ articles in minutes via batch operations
- Incremental Updates: Process new content automatically as it's published
- Duplicate Detection: Identify similar content and consolidate taxonomies
- Cross-linking: Suggest related content based on shared tags
4. SEO & Metadata Generation
- Schema Markup: Auto-generate JSON-LD structured data for rich snippets
- Meta Tags: Create optimized title tags, meta descriptions, and OG tags
- Keyword Extraction: Identify primary and secondary keywords with search volume
- Readability Scoring: Assess content complexity and reading level
5. Multi-Platform Distribution
- Content Syndication: Apply consistent tags across WordPress, Shopify, and custom CMS
- API Webhooks: Push tags to external systems via REST/GraphQL APIs
- Format Conversion: Export tags as CSV, JSON, XML, or proprietary formats
- Slack Integration: Post tagging results and recommendations to team channels
Configuration
Environment Variables
# Required
export OPENAI_API_KEY="sk-..." # GPT-4 or GPT-3.5-turbo for analysis
export CONTENT_TAXONOMY_CONFIG="config.json" # Path to taxonomy definition
# Optional
export WORDPRESS_URL="https://yoursite.com" # WordPress REST API endpoint
export WORDPRESS_TOKEN="xxxxxxxx" # WordPress application password
export SLACK_WEBHOOK_URL="https://hooks.slack.com/..."
export ELASTICSEARCH_HOST="localhost:9200" # For tag indexing
export CUSTOM_TAXONOMY_FILE="taxonomy.json" # Your taxonomy rules
Taxonomy Configuration File (taxonomy.json)
{
"version": "1.0.0",
"categories": {
"industry": [
"e-commerce", "saas", "healthcare", "fintech", "education"
],
"format": [
"blog-post", "video", "infographic", "case-study", "whitepaper", "guide"
],
"audience": [
"c-suite", "managers", "individual-contributors", "entrepreneurs"
],
"intent": [
"educational", "promotional", "how-to", "comparison", "news"
]
},
"confidence_threshold": 0.75,
"max_tags_per_content": 15,
"language": "en"
}
Setup Instructions
- Obtain API Keys: Get OpenAI API key from platform.openai.com
- Define Taxonomy: Create or upload your taxonomy.json file
- Test Connection: Run
openclaw test ai-driven-content-taxonomy-and-tagging - Configure Integrations: Connect WordPress, Slack, or other platforms (optional)
- Dry Run: Process 5-10 sample articles to validate results before bulk operations
Example Outputs
Output 1: Detailed Tag Report (JSON)
{
"content_id": "post_12345",
"title": "How to Build a Sustainable E-commerce Business",
"tags": {
"industry": [
{ "tag": "e-commerce", "confidence": 0.98, "weight": 1.0 },
{ "tag": "sustainability", "confidence": 0.92, "weight": 0.8 }
],
"format": [
{ "tag": "guide", "confidence": 0.95, "weight": 1.0 }
],
"audience": [
{ "tag": "entrepreneurs", "confidence": 0.89, "weight": 0.9 },
{ "tag": "business-owners", "confidence": 0.85, "weight": 0.8 }
],
"intent": [
{ "tag": "educational", "confidence": 0.96, "weight": 1.0 }
]
},
"seo_metadata": {
"meta_title": "How to Build a Sustainable E-commerce Business in 2024",
"meta_description": "Learn proven strategies for building an eco-friendly e-commerce business. Expert guide covering sustainable sourcing, green logistics, and customer education.",
"primary_keyword": "sustainable e-commerce",
"secondary_keywords": ["eco-friendly online store", "green business practices"],
"schema_markup": {
"@context": "https://schema.org",
"@type": "BlogPosting",
"headline": "How to Build a Sustainable E-commerce Business in 2024",
"keywords": ["sustainable e-commerce", "eco-friendly"]
}
},
"reading_level": "intermediate",
"estimated_read_time": "8 minutes",
"related_content": ["post_12340", "post_12341"],
"processing_time_ms": 2340
}
Output 2: Bulk Processing Summary (CSV)
post_id,title,industry,format,audience,intent,confidence_avg,seo_score,status
12345,"How to Build Sustainable E-commerce",e-commerce,guide,entrepreneurs,educational,0.92,92,success
12346,"Q3 Financial Report",fintech,whitepaper,c-suite,promotional,0.88,85,success
12347,"Customer Success Story",saas,case-study,managers,promotional,0.91,89,success
12348,"API Documentation",saas,guide,developers,educational,0.94,91,success
Output 3: Content Gap Analysis
## Content Taxonomy Coverage Report
### Strengths
- **E-commerce**: 45 articles (excellent coverage)
- **Sustainability**: 28 articles (good coverage)
- **Guides**: 92 articles (strong format diversity)
### Gaps Identified
- **Healthcare + Sustainability**: 0 articles (opportunity)
- **Video Format**: Only 3 articles (underutilized)
- **C-Suite Audience**: 5 articles (expand for enterprise market)
### Recommendations
1. Create 3-5 healthcare sustainability case studies
2. Produce 10-15 video tutorials on core topics
3. Develop 5 executive briefings for C-suite audience
4. Consolidate 8 overlapping articles on "green logistics"
Tips & Best Practices
1. Taxonomy Design
- Start Simple: Begin with 3-5 main dimensions (industry, format, audience, intent, topic)
- Avoid Overlap: Ensure categories are mutually exclusive where possible
- Use Hierarchies: Organize tags in parent-child relationships (e.g., "Technology > AI > LLMs")
- Regular Audits: Review and refine your taxonomy quarterly as your content evolves
2. Confidence Thresholds
- Strict (>0.90): Use for critical content, compliance-heavy industries
- Balanced (0.75-0.90): Default for most content; catches 95% of correct tags
- Permissive (<0.75): Use for discovery/brainstorming; requires manual review
3. Bulk Operations
- Test First: Always run 5-10 articles before processing entire library
- Batch Size: Process 100-500 articles per batch to avoid API rate limits
- Monitor Progress: Check Slack notifications or logs for errors in real-time
- Schedule Off-Peak: Run bulk jobs during low-traffic hours
4. Content Maintenance
- Incremental Tagging: Tag new content within 24 hours of publication
- Seasonal Updates: Refresh tags quarterly to capture trending topics
- Remove Obsolete Tags: Archive taxonomy versions no longer in use
- Document Changes: Maintain a changelog of taxonomy modifications
5. Integration Best Practices
- Webhook Validation: Verify WordPress/Slack webhooks are receiving data
- Error Handling: Set up alerts for failed tag updates across platforms
- Backup Original Tags: Keep manual tags as fallback if AI tagging fails
- A/B Test Results: Compare AI tags vs. human-created tags for 30 days
Safety & Guardrails
What This Skill Will NOT Do
- Override Manual Tags: AI tags are suggestions; existing manual tags are never auto-deleted
- Make Sensitive Decisions: Does not classify content as harmful, illegal, or NSFW without human review
- Modify Content: Only adds metadata; never rewrites article text or titles
- Store Sensitive Data: Does not retain or log content body text; only processes for tagging
- Guarantee Accuracy: AI tagging is 90-95% accurate; always review high-stakes classifications
Limitations & Boundaries
- Language Support: Optimal performance in English; other languages may have 5-10% lower accuracy
- Domain-Specific Jargon: May struggle with highly specialized terminology; use custom synonym mapping
- Content Length: Works best with 300+ word articles; very short content (< 100 words) may be undertagged
- Rate Limits: OpenAI API has 3,500 requests/minute limit; batch processing respects this automatically
- Cost Implications: Typical cost is $0.02-0.05 per article depending on length and API model
Privacy & Compliance
- GDPR Compliance: Does not store PII; processes content in memory only
- SOC 2 Alignment: Compatible with enterprise security requirements
- Data Retention: Tagging results stored locally; no external logging by default
- Audit Trail: Maintains version history of all taxonomy changes for compliance
Troubleshooting
Q: Why are some tags showing low confidence scores (<0.70)?
A: Low confidence indicates ambiguous content or domain-specific terminology not in the AI model's training data. Solutions:
- Add custom synonyms to your taxonomy.json (e.g., "ML" → "machine learning")
- Review the content; it may genuinely be multi-topic
- Increase confidence threshold to filter low-quality tags
- Provide training examples for niche topics
Q: Bulk tagging is slow. How can I speed it up?
A: Default settings use GPT-4 for accuracy. Solutions:
- Switch to GPT-3.5-turbo for 3x faster processing (slight accuracy trade-off)
- Increase batch size from 100 to 500 articles
- Disable schema markup generation if not needed
- Run processing during off-peak hours (fewer concurrent API requests)
Q: My WordPress tags aren't syncing. What's wrong?
A: Common causes: expired authentication token, incorrect REST API endpoint, or insufficient permissions. Solutions:
- Verify WordPress URL is correct:
https://yoursite.com/wp-json/wp/v2/posts - Regenerate application password in WordPress: Settings > Application Passwords
- Check user role has "edit_posts" capability
- Test with:
curl -u user:password https://yoursite.com/wp-json/wp/v2/posts - Enable debug logging:
export DEBUG=openclaw:*
Q: How do I prevent over-tagging (too many tags per article)?
A: Set max_tags_per_content in your taxonomy.json configuration:
{
"max_tags_per_content": 12,
"tag_priority": "confidence_desc"
}
This keeps only the 12 highest-confidence tags per article. Adjust based on your CMS and user experience preferences.
Q: Can I use this for non-English content?
A: Yes, but with caveats. Supported languages: Spanish, French, German, Chinese, Japanese, Portuguese.
- Accuracy: 85-90% (vs. 95% for English)
- Set language in config:
"language": "es"for Spanish - Consider using language-specific OpenAI models if available
- Provide more training examples for non-English taxonomies
Q: What if my content is behind a paywall or login?
A: The skill processes content you provide directly. Options:
- Copy/paste article text directly into the prompt
- Export content from your CMS API (WordPress, Contentful, etc.)
- Use browser automation to fetch login-protected content
- Provide raw content files (CSV, JSON, Markdown)
Q: How do I maintain consistency across multiple content creators?
A: Implement a taxonomy governance process:
- Centralized Taxonomy: Store taxonomy.json in version control (GitHub)
- Training: Show team examples of correctly tagged articles
- Approval Workflow: Have editors review AI tags before publishing
- Feedback Loop: Update custom synonyms based on rejected tags
- Monthly Audits: Spot-check 50 articles for consistency
Q: Can I export tags in a specific format for my CMS?
A: Yes. The skill supports multiple export formats: ```bash