Synapse

Synapse is a powerful, cross-platform desktop application for running AI models locally. Built with Electron, React, and TypeScript, it provides a user-friendly interface for chatting with local AI models while offering advanced features like profile management, hardware detection, and tool calling capabilities.

Features

Core Functionality

• Local AI Model Execution: Run AI models directly on your machine for privacy and low-latency responses
• Chat Interface: Intuitive chat UI with support for tool calls and extended conversation contexts
• Model Management: Browse, download, and manage AI models with automatic profile creation
• Profile System: Create and customize chat profiles with different model configurations, system prompts, and temperature settings

Advanced Features

• Hardware Detection: Automatic detection of CPU, RAM, and GPU/VRAM to recommend optimal model configurations
• Tool Calling: Built-in filesystem and Git tools for AI agents to interact with your local environment
• Multi-Model Support: Support for various model formats including GGUF
• Settings Management: Comprehensive settings for model parameters (temperature, top-k, top-p, min-p, seed, etc.)

Technical Highlights

• Electron + React + TypeScript: Modern tech stack for cross-platform desktop development
• WebGPU Support: Leverages modern GPU acceleration for AI inference
• IPC Communication: Efficient main/renderer process communication via Electron IPC
• State Management: React Context and hooks for managing application state

Getting Started

Prerequisites

• Node.js 18+ and npm or yarn
• A modern system with at least 8GB RAM recommended
• GPU with WebGPU support (optional, for accelerated inference)

Installation

1. Clone the repository:

git clone https://github.com/your-repo/synapse.git
cd synapse

2. Install dependencies:

npm install

3. Start the development server:

npm start

Building for Production

To build the application for your platform:

npm run package

Usage

Chat

• Start a conversation with your selected AI model
• View tool calls with expandable details
• Copy responses and manage chat history

Models

• Browse available AI models
• Download and install models automatically
• Create default profiles for new models

Profiles

• Create custom chat profiles with specific model configurations
• Set system prompts for each profile
• Adjust model parameters (temperature, top-k, top-p, etc.)
• Import/export profiles

Settings

• Configure model preferences
• Set hardware limits (RAM/VRAM allocation)
• Manage API keys (e.g., Hugging Face token)
• Configure auto-detection settings

Architecture

src/
├── main/                 # Electron main process
│   ├── main.ts          # Main entry point
│   └── preload.ts       # Preload script for IPC
├── renderer/            # React renderer process
│   ├── pages/           # Application pages
│   │   ├── ChatPage.tsx
│   │   ├── ModelsPage.tsx
│   │   ├── ProfilesPage.tsx
│   │   └── SettingsPage.tsx
│   ├── components/      # Reusable React components
│   ├── styles/          # CSS modules
│   └── types/           # TypeScript type definitions
└── data/                # Static data files
    ├── defaultTools.ts  # Available AI tools
    └── languages.ts     # Language support

Available Tools

Synapse includes built-in tools that AI models can use to interact with your system:

Filesystem Tools

• read_text_file: Read text files with optional line ranges
• read_media_file: Read media files (images, videos, PDFs) as base64
• read_multiple_files: Read multiple text files simultaneously
• write_file: Write content to files
• edit_file: Edit files with text replacements
• create_directory: Create directories recursively
• list_directory: List directory contents
• list_directory_with_sizes: List with file sizes
• move_file: Move or rename files
• search_files: Search for files by glob pattern
• directory_tree: Generate directory tree view
• get_file_info: Get detailed file metadata
• list_allowed_directories: List configured allowed directories

Git Tools

• git_status: Get repository working tree status
• git_diff_unstaged: Show unstaged changes
• git_diff_staged: Show staged changes
• git_diff: Compare against branches/commits
• git_commit: Commit staged changes
• git_add: Stage files for commit
• git_reset: Unstage all changes
• git_log: View commit history
• git_create_branch: Create new branch
• git_checkout: Switch branches
• git_show: Show commit contents
• git_branch: List branches

Configuration

Model Parameters

• Temperature: Controls randomness in responses (0.0-2.0)
• Top-K: Limits sampling to top K tokens (1-100)
• Top-P: Nucleus sampling threshold (0.0-1.0)
• Min-P: Minimum probability threshold
• Seed: Random seed for reproducibility

Hardware Settings

• RAM Allocation: Recommended based on model size
• VRAM Allocation: For GPU-accelerated inference
• GPU Selection: Choose specific GPU for multi-GPU systems

Development

Scripts

# Development
npm start              # Start development server
npm run lint           # Run ESLint
npm run lint:fix       # Fix linting issues

# Build
npm run build          # Build for production
npm run build:main     # Build main process
npm run build:renderer # Build renderer process

# Package
npm run package        # Package application

Project Structure

• .erb/: Electron React Boilerplate configuration
• node_modules/: Dependencies
• public/: Public assets
• release/: Build artifacts

Roadmap

• Auto-detect and recommend optimal model settings on first boot
• Cache hardware configuration for faster startup
• Add support for more AI model formats
• Implement cloud model integration
• Add export/import for chat history
• Enhance tool calling with custom tool definitions

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

1. Fork the repository
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a Pull Request

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• Electron for the desktop framework
• React for the UI library
• TypeScript for type safety
• WebGPU for GPU acceleration

---

Built with ❤️ for the AI community

For more information and documentation, please visit our documentation (placeholder).