c9ai/CHANGELOG.md

Universal AI assistant with vibe-based workflows, hybrid cloud+local AI, and comprehensive tool integration

# Changelog

All notable changes to C9 AI will be documented in this file.

## [2.2.3] - 2025-09-04

### Packaging
- Added `files` whitelist in `package.json` to include only runtime code (`src/`, `bin/`, `scripts/`, `server/`, `README.md`, `CHANGELOG.md`).
- Added metadata: `repository`, `bugs`, `homepage`, `keywords`, `engines`, and `license`.
- Verified package with `npm pack`; reduced published tarball size by excluding dev artifacts and tests.
- Published as `c9ai@2.2.3`. Previous versions remain available (no breaking changes).

### Notes
- No functional code changes in this release; safe upgrade for new installs.

## [2.2.4] - 2025-09-04

### Fixes
- JIT calculator: fixed method name collision causing `toLowerCase` error in `@calc` expressions.

### UI/UX
- RSS headlines: titles as H2 (no blue link), date + publisher line, pill‑style Read link, full‑width thumbnails, and tighter spacing.

### Docs
- Added `public/demo-lineup.html` printable demo handout.
- Added Gamma‑compatible deck at `docs/slides/c9ai-gamma-deck.md`.

### Packaging
- No change to package contents (uses `files` whitelist). Verified with `npm pack`.

## [2.2.5] - 2025-09-06

### UI/Slides
- Added Slides Library UI and API (`/api/slides`) to list and open generated decks.
- Added workflow template “Markdown → Web Slideshow” and generator script `scripts/md-to-slides.js`.

### Packaging
- Included `public/` in npm `files` so the updated UI and slides assets ship with the package.

## [2.0.0] - 2025-08-04

### 🚀 Major Features Added

#### 🧠 Local AI Integration
- **Local LLM Support**: Full integration with Microsoft Phi-3, TinyLLaMA, and LLaMA models
- **Privacy-First Processing**: All AI reasoning can run locally with zero external API calls
- **Model Management System**: Download, install, remove, and manage local AI models
- **Smart Model Loading**: Automatic model initialization and caching

#### 🗣️ Natural Language Interface
- **Conversational Commands**: Talk naturally to c9ai - "compile my research paper" 
- **System Command Understanding**: "list documents in directory" → `ls -la /path`
- **Intelligent Intent Recognition**: Natural language → structured @action format
- **Context-Aware Processing**: AI understands file paths, common tasks, and user patterns

#### ⚡ Enhanced Todo Management
- **Intelligent Todo Processing**: Plain English todos automatically converted to executable actions
- **Multi-Mode Support**: Manual @action, natural language, and hybrid approaches
- **Smart Action Generation**: AI suggests appropriate commands based on task description
- **Seamless Execution**: Natural language todos can be executed directly

#### 🔄 Advanced Model Switching
- **Local ↔ Cloud Routing**: Intelligent switching between local and cloud AI
- **Graceful Fallbacks**: Local → Cloud → Manual command progression
- **Real-time Switching**: Change models mid-conversation without losing context
- **Performance Optimization**: Local models for speed, cloud for complexity

### 🛠️ New Commands

#### Model Management
```bash
models list                    # Show available and installed models
models install <model>         # Download and install local AI models  
models remove <model>          # Remove installed models to free space
models status                  # Show disk usage and model information
```

#### Enhanced Todo Commands
```bash
todos add "natural language"   # AI converts to structured actions
todos execute                  # Interactive execution of actionable todos
todos actions                  # Show all actionable todos with intents
```

#### Interactive Natural Language
```bash
c9ai> compile my research paper           # Converts to: @action: compile research.tex
c9ai> list documents in /path/directory   # Executes: ls -la "/path/directory"  
c9ai> open my budget spreadsheet          # Converts to: @action: open budget.xlsx
```

### 🔧 Technical Improvements

#### Architecture Enhancements
- **Modular AI Backend**: Support for multiple local LLM providers
- **Async Model Loading**: Non-blocking model initialization
- **Configuration Persistence**: Model preferences saved across sessions
- **Enhanced Error Handling**: Graceful fallbacks for all failure scenarios

#### Performance Optimizations
- **Lazy Loading**: Models loaded only when needed
- **Caching System**: Intelligent model and response caching
- **Pattern Recognition**: Fast local pattern matching before AI processing
- **Resource Management**: Efficient memory usage for large models

#### Developer Experience
- **Enhanced Logging**: Detailed interaction logs in `~/.c9ai/logs/`
- **Debug Support**: Optional debug output for troubleshooting
- **Extension Points**: Easy integration of new AI models and actions
- **Configuration Flexibility**: Customizable model selection and routing

### 📊 Privacy & Security

#### Local Processing
- **Zero External Dependencies**: Complete local AI processing option
- **Data Privacy**: User data never leaves local machine in local mode
- **Transparent Processing**: Always shows which model (local/cloud) is being used
- **User Control**: Full control over when to use local vs cloud processing

#### Intelligent Routing
- **Privacy-First Defaults**: Attempts local processing first when available
- **Explicit Consent**: Clear indication when routing to cloud APIs
- **Fallback Transparency**: User always knows processing location
- **Opt-in Cloud**: Cloud processing only with explicit model selection

### 🎯 Use Case Expansions

#### Software Development
- Natural language compilation commands
- Intelligent project file navigation  
- Context-aware development task processing

#### Document Management
- Smart document operations via natural language
- Automatic file type detection and appropriate actions
- Intelligent search and organization commands

#### System Administration
- Natural language system commands
- Intelligent resource monitoring
- Context-aware file and process management

### 🔄 Breaking Changes
- **Version Bump**: Major version increase to 2.0.0
- **New Dependencies**: Added `node-llama-cpp` for local AI support
- **Configuration Changes**: New model configuration in `~/.c9ai/config.json`
- **Command Additions**: New `models` command namespace

### 🐛 Bug Fixes
- **Initialization Race Conditions**: Fixed async initialization in constructor
- **Configuration Loading**: Proper config loading before command execution
- **Error Handling**: Improved error messages and recovery
- **Path Resolution**: Better handling of file paths with spaces

### 📦 Dependencies
- **Added**: `node-llama-cpp@^3.11.0` - Local LLM integration
- **Updated**: Enhanced error handling across all existing dependencies
- **Maintained**: Full backward compatibility with existing tool integrations

---

## [1.0.1] - Previous Release

### Features
- Basic Claude and Gemini AI integration
- Todo management with GitHub Issues
- Simple @action execution system
- Interactive mode with AI switching

---

**Migration Guide**: Existing users can upgrade seamlessly. All existing functionality remains unchanged. New features are additive and optional.

**Next Release Preview**: Phase 2 will add learning system, analytics dashboard, and advanced context awareness.