AI Tagger Universe

AI Tagger Universe: Easy Tag Generation & Management for Obsidian

Automatically generate intelligent tags for your Obsidian notes using AI. This plugin analyzes your content and adds relevant tags to your note's frontmatter, helping you organize and discover connections in your knowledge base.

🔌 Installation

This plugin can be installed directly from the Obsidian Community Plugins browser:

1. Open Obsidian Settings
2. Navigate to Community Plugins
3. Disable Safe Mode (if enabled)
4. Search for "AI Tagger Universe"
5. Click Install, then Enable

Alternatively, you can manually install the plugin:

1. Download the latest release from this repository
2. Extract the files to your Obsidian vault's .obsidian/plugins/ai-tagger-universe folder
3. Reload Obsidian and enable the plugin in the Community Plugins settings

✨ Key Features

🤖 Flexible AI Integration

• Use your preferred AI service:
  • Local LLMs: Ollama, LM Studio, LocalAI, or any OpenAI-compatible endpoint
  • Cloud Services: OpenAI, Claude, Gemini, Groq, Grok, Mistral, DeepSeek, Cohere, GLM (Zhipu AI), MiMo (Xiaomi), Minimax, SiliconFlow, Aliyun, Bedrock, Vertex AI, OpenRouter, and more

🏷️ Smart Tagging System

• Multiple tagging modes:
  • Generate completely new tags based on content
  • Match against your existing vault tags
  • Use predefined tags from a custom list
  • Hybrid modes combining generation with existing/predefined tags
• Nested/Hierarchical tags (NEW!):
  • Generate hierarchical tags like science/biology/genetics
  • User-configurable max depth (1-3 levels)
  • Smart prompting for category-based organization
  • Backward compatible (disabled by default)
• Batch operations for tagging multiple notes at once
• Multilingual support for generating tags in your preferred language

📊 Tag Network Visualization

• Interactive graph showing relationships between tags
• Discover connections and patterns in your knowledge base
• Search functionality to find specific tags
• Node size indicates tag frequency
• Adjustable force settings: Customize repulsion strength and link distance
• Click-to-show documents: Click any tag to see all notes containing it
• Real-time updates: Network refreshes automatically when tags change

🛠️ Advanced Management

• Generate tags from selected text portions
• Batch tag entire folders or your whole vault
• Clear tags while preserving other frontmatter
• Collect and export all tags from your vault
• Flatten hierarchical tags: Convert nested tags (a/b/c) into separate flat tags
• Tag format options: Choose between kebab-case, camelCase, PascalCase, snake_case, or original
• Debug Mode: Enhanced logging for troubleshooting tag generation
• Popular Tools Tips: Built-in guidance for common LLM setup configurations

🆕 What's New in Version 1.0.16

Major Features

• 🌐 New AI Providers:
  • GLM (Zhipu AI) - China's leading AI service
  • MiMo (Xiaomi) - Xiaomi's MiMo-V2-Flash model
  • Minimax - MiniMax-M2.1 model
• 🏷️ Tag Format Options: Choose your preferred naming convention
  • kebab-case (my-tag-name)
  • camelCase (myTagName)
  • PascalCase (MyTagName)
  • snake_case (my_tag_name)
  • Original (preserve as-is)
• 📊 Enhanced Tag Network Visualization:
  • Adjustable force settings (repulsion strength, link distance)
  • Click any tag node to see all documents containing it
  • Real-time updates when tags change in your vault
  • Manual refresh button
• 🔀 Flatten Hierarchical Tags: New commands to convert nested tags into flat tags
  • Flatten for current note, folder, or entire vault
  • Converts science/biology/genetics → science, biology, genetics

Bug Fixes

• Fixed Claude API CORS compatibility for browser-based access

📝 Version 1.0.15

Major Features

• 🌳 Nested Tags Support: Generate hierarchical tags with parent/child relationships
  • Create tags like technology/artificial-intelligence/machine-learning
  • Configurable max nesting depth (1-3 levels)
  • Smart LLM prompting for hierarchical structure
  • Fully integrated with bilingual interface

Improvements

• Added automated testing suite with 32 validation tests
• Enhanced tag formatting to preserve forward slashes
• Improved settings organization with nested tags section
• Better documentation with CLAUDE.md for future development

Bug Fixes

• Fixed icon display issue in ribbon and toolbar (replaced invalid 'network' icon with 'git-graph')

📝 Previous Updates

Version 1.0.14

• 🎉 Full Chinese Interface Support: Complete localization for Chinese-speaking users
• 🌍 Bilingual Interface: Easy language switching between English and Chinese
• 🔧 Enhanced Debug Mode: Better logging and troubleshooting capabilities
• 📋 Improved User Guidance: Tips for popular AI tools and services

🚀 Quick Start

1. Install the plugin from Obsidian Community Plugins
2. Configure your AI provider:
   • Go to Settings → AI Tagger Universe → LLM Settings
   • Choose between Local LLM (Ollama, LM Studio, etc.) or Cloud Service (OpenAI, Claude, etc.)
   • Enter your endpoint URL and API key (if required)
   • Test the connection to verify it works
3. Select your tagging mode:
   • Choose how you want tags to be generated (new tags, existing tags, or hybrid)
   • Adjust tag generation limits (0-10 tags per note)
4. Optional: Configure interface language:
   • Go to Settings → AI Tagger Universe → Interface
   • Choose between English or 中文 (Chinese)
   • Restart Obsidian for the language change to take effect
5. Start generating tags:
   • Use the ribbon icon (left sidebar) to tag the current note
   • Use the command palette (Ctrl/Cmd+P) for more options
   • View tag relationships with the Tag Network visualization

🔧 Configuration Options

LLM Settings

• Service Type: Local LLM or Cloud Service
• AI Provider: Choose from 15+ services (Ollama, OpenAI, Claude, Gemini, Groq, etc.)
• Endpoint URL: Your LLM service endpoint
• API Key: Authentication key (if required)
• Model Name: Specific model to use
• Temperature (Override): Leave empty to use provider default; set to 0 to reduce randomness for more repeatable tag output

Tag Generation

• Tagging Mode: Select how tags are generated or matched
  • Generate New: Create entirely new tags from content
  • Predefined Tags: Match against existing vault tags
  • Hybrid: Combine generation with existing tags
  • Custom: Use your own tag list from a file
• Tag Limits: Set maximum numbers for generated/matched tags (0-10)
• Tag Language: Generate tags in your preferred language
• Tag Format: Choose naming convention (kebab-case, camelCase, PascalCase, snake_case, original)
• Nested Tags:
  • Enable hierarchical tag generation (e.g., parent/child/grandchild)
  • Configure max nesting depth (1-3 levels)
  • Creates parent/child relationships for better organization

Interface & Advanced

• Interface Language: Choose between English and Chinese
• Excluded Paths: Skip specific folders during batch operations
• Debug Mode: Enable detailed logging for troubleshooting
• Replace Tags: Overwrite existing tags or append to them

📖 Usage Examples

• Research Notes: Automatically categorize research papers and findings
• Project Management: Tag project notes for better organization
• Knowledge Base: Discover connections between concepts
• Content Creation: Generate relevant tags for blog posts or articles
• Personal Journal: Track themes and topics in your journal entries

🌐 Language Support

Tag Generation

Generate tags in multiple languages including English, Chinese, Japanese, German, French, Spanish, Russian, and many more.

Interface Localization (NEW!)

• Full Chinese Interface: Complete Chinese language support for the plugin interface
• Bilingual Support: Seamlessly switch between English and Chinese interfaces
• Localized Settings: All configuration panels and options available in Chinese
• Translated Commands: Command palette and ribbon actions fully localized
• Multilingual Messages: All notifications, prompts, and feedback in your preferred language

To change the interface language:

1. Go to AI Tagger Universe Settings
2. Navigate to the "Interface" section
3. Select your preferred language (English/中文)
4. Restart Obsidian for the change to take effect

🔄 Fork Improvements

This fork includes several enhancements over the original plugin:

Bug Fixes

• Fixed malformed tag prefixes: Resolved issue where some LLMs would generate tags like tag:matchedExistingTags-medical-research instead of clean tags like medical-research
  • Added robust tag sanitization that strips malformed prefixes (tag:, matchedExistingTags-, suggestedTags-, etc.)
  • Enhanced prompts with explicit examples of correct vs. incorrect tag formats

Prompt Engineering Improvements

• Claude-optimized prompts: Restructured all prompts using XML-style tags (<task>, <requirements>, <output_format>) for better LLM comprehension
• Enforced kebab-case formatting: All tagging modes now consistently generate tags in kebab-case format (e.g., machine-learning, data-science)
• Improved tag quality guidelines: Added explicit requirements for concise (1-3 words), specific, and descriptive tags
• Real-world examples: Replaced placeholder examples with actual domain-appropriate tag examples
• Consistent structure: Unified prompt structure across all tagging modes (GenerateNew, PredefinedTags, Hybrid, Custom)

Code Quality

• Enhanced error handling: Better validation and sanitization of LLM responses
• Comprehensive documentation: Improved inline code comments and type definitions

Testing

• Included test script (test-sanitization.js) for verifying tag generation with your actual LLM endpoint
• See TEST_INSTRUCTIONS.md for testing guidance

These improvements result in more reliable tag generation, better formatting consistency, and improved compatibility with various LLM providers including Claude, GPT-4, and local models.

☕ Support

If you find this plugin helpful, consider buying me a coffee!

📝 License

MIT License - see the LICENSE file for details

🤝 Contributing

Contributions are welcome! Please feel free to submit Issues or Pull Requests.

📮 Support

If you encounter any issues or have feature requests, please submit them on GitHub Issues.

🙏 Acknowledgments

Thanks to all the developers and users who have contributed to this project!