🚀 KnowNote v1.0.5

This release introduces local LLM support with Ollama, a complete provider system architecture overhaul, and various UI improvements.

✨ Major Features

🤖 Ollama Local LLM Provider

• Run AI models completely offline with Ollama integration
• Support for both chat and embedding capabilities
• OpenAI-compatible API implementation for seamless integration
• Default endpoint: http://localhost:11434/v1
• Flexible configuration with custom server address support

🏗️ Provider System Architecture Overhaul

• Capability-based design: Modular chat, embedding, rerank, and image generation capabilities
• Config-driven architecture: Simplified provider registration and management
• Unified OpenAI-compatible implementation: Reduced code duplication
• Registry pattern with ProviderDescriptor and ProviderRegistry
• Cleaner separation between configuration and runtime logic

🔧 Improvements

Provider Management

• ✅ Fix model name truncation issue - now properly preserves version tags
  • Examples: qwen3:0.6b, gpt-4o-mini:2024-07-18, llama3.2:latest
• ✅ Unified provider URL configuration naming across all providers
• ✅ Enhanced custom provider support with Base URL configuration
• ✅ Improved error handling and validation

UI/UX Enhancements

• 🎨 Custom scrollbar styling for dropdown menus
• 🔧 Improved provider settings panel layout
• 📋 Better model selection interface

🐛 Bug Fixes

• Fixed model ID parsing to support colon-separated version tags
• Improved provider configuration persistence
• Enhanced error messages for better debugging

📦 Technical Details

New Architecture Components

src/main/providers/
├── capabilities/          # Modular capability interfaces
│   ├── BaseProvider.ts
│   ├── ChatCapability.ts
│   ├── EmbeddingCapability.ts
│   ├── RerankCapability.ts
│   └── ImageGenerationCapability.ts
├── handlers/             # Protocol-specific implementations
│   ├── OpenAIChatHandler.ts
│   └── OpenAIEmbeddingHandler.ts
├── registry/            # Provider registration system
│   ├── ProviderDescriptor.ts
│   ├── ProviderRegistry.ts
│   └── builtinProviders.ts
└── base/
    └── OpenAICompatibleProvider.ts

Code Changes

• 30 files changed: +1,422 insertions, -723 deletions
• Removed individual provider classes (DeepSeekProvider, KimiProvider, etc.)
• Centralized logic in OpenAICompatibleProvider
• Created reusable handler pattern for API communication

🎯 Supported Providers

📚 Getting Started with Ollama

Installation

# macOS
brew install ollama

# Linux
curl -fsSL https://ollama.com/install.sh | sh

# Windows: Download from https://ollama.com

Usage

# Start Ollama service
ollama serve

# Download models
ollama pull qwen2.5:7b        # Chat model
ollama pull nomic-embed-text  # Embedding model

Configuration in KnowNote

1. Open Settings → Providers
2. Select Ollama
3. API Key: Enter any value (e.g., ollama)
4. Base URL: Keep default http://localhost:11434/v1
5. Click "Fetch Models" and select your downloaded models
6. Enable the provider

🔄 Migration Notes

For Existing Users

• No breaking changes for existing provider configurations
• Model selections will be automatically preserved
• The architecture changes are internal and backward-compatible

For Developers

• Old provider classes (OpenAIProvider, DeepSeekProvider, etc.) have been removed
• All providers now use OpenAICompatibleProvider as the base
• Custom providers should implement capability interfaces
• See src/main/providers/registry/builtinProviders.ts for registration examples

🙏 Acknowledgments

Thanks to the Ollama team for creating an excellent local LLM runtime!

📝 Full Changelog

Commits included in this release:

• fdf9062 feat: add Ollama local LLM provider support (#7)
• 9f26183 style: add custom scrollbar styling to dropdown menus (#6)
• 9e48198 refactor: unify provider URL configuration naming
• 6edf9d3 refactor: restructure provider system with capability separation and config-driven architecture

Full Changelog: v1.0.4...v1.0.5