# NomadArch
Fork of CodeNomad by OpenCode
Features • AI Models • Installation • Usage • What's New • Credits
--- ## Overview NomadArch is an enhanced fork of CodeNomad by OpenCode, featuring significant UI/UX improvements, additional AI integrations, and a more robust architecture. This is a full-featured AI coding assistant with support for multiple AI providers including **GLM 4.7**, Anthropic, OpenAI, Google, Qwen, and local models via Ollama. ### Key Improvements Over CodeNomad - Fixed Qwen OAuth authentication flow - Enhanced MULTIX Mode with live token streaming - Improved UI/UX with detailed tooltips - Auto-build verification on launch - Comprehensive installer scripts for all platforms - Port conflict detection and resolution hints --- ## Supported AI Models & Providers NomadArch supports a wide range of AI models from multiple providers, giving you flexibility to choose the best model for your coding tasks. ### 🚀 Featured Model: GLM 4.7 (Z.AI) **GLM 4.7** is the latest state-of-the-art open model from Z.AI, now fully integrated into NomadArch. Released in December 2025, GLM 4.7 ranks **#1 for Web Development** and **#6 overall** on the LM Arena leaderboard. #### Key Features - 🔥 **128K Context Window** - Process entire codebases in a single session - 🧠 **Interleaved Thinking** - Advanced reasoning with multi-step analysis - 💭 **Preserved Thinking** - Maintains reasoning chain across long conversations - 🔄 **Turn-level Thinking** - Optimized per-response reasoning for efficiency #### Benchmark Performance | Benchmark | Score | Improvement | |-----------|-------|-------------| | SWE-bench | **+73.8%** | Over GLM-4.6 | | SWE-bench Multilingual | **+66.7%** | Over GLM-4.6 | | Terminal Bench 2.0 | **+41%** | Over GLM-4.6 | | LM Arena WebDev | **#1** | Open Model Ranking | | LM Arena Overall | **#6** | Open Model Ranking | GLM 4.7 beats GPT-5, Claude Sonnet, and Gemini on multiple coding benchmarks. #### Z.AI API Integration - ✅ Fully integrated via Z.AI Plan API - ✅ Compatible with Claude Code, Cline, Roo Code, Kilo Code - ✅ Get **10% discount** with code: [`R0K78RJKNW`](https://z.ai/subscribe?ic=R0K78RJKNW) - 🎯 [Subscribe to Z.AI with 10% off](https://z.ai/subscribe?ic=R0K78RJKNW) --- ### 🤖 All Supported Models #### Z.AI | Model | Context | Specialty | |-------|---------|-----------| | **GLM 4.7** | 128K | Web Development, Coding | | GLM 4.6 | 128K | General Coding | | GLM-4 | 128K | Versatile | #### Anthropic | Model | Context | Specialty | |-------|---------|-----------| | Claude 3.7 Sonnet | 200K | Complex Reasoning | | Claude 3.5 Sonnet | 200K | Balanced Performance | | Claude 3 Opus | 200K | Maximum Quality | #### OpenAI | Model | Context | Specialty | |-------|---------|-----------| | GPT-5 Preview | 200K | Latest Capabilities | | GPT-4.1 | 128K | Production Ready | | GPT-4 Turbo | 128K | Fast & Efficient | #### Google | Model | Context | Specialty | |-------|---------|-----------| | Gemini 2.0 Pro | 1M+ | Massive Context | | Gemini 2.0 Flash | 1M+ | Ultra Fast | #### Qwen | Model | Context | Specialty | |-------|---------|-----------| | Qwen 2.5 Coder | 32K | Code Specialized | | Qwen 2.5 | 32K | General Purpose | #### Local (Ollama) | Model | Size | Specialty | |-------|------|-----------| | DeepSeek Coder | Varies | Code | | Llama 3.1 | Varies | General | | CodeLlama | Varies | Code | | Mistral | Varies | General | --- ## Installation ### Quick Start (Recommended) The installers will automatically install **OpenCode CLI** (required for workspace functionality) using: 1. **Primary**: `npm install -g opencode-ai@latest` (fastest) 2. **Fallback**: Download from official GitHub releases if npm fails #### Windows ```batch # Double-click and run Install-Windows.bat # Then start app Launch-Windows.bat ``` #### Linux ```bash chmod +x Install-Linux.sh ./Install-Linux.sh # Then start app ./Launch-Unix.sh ``` #### macOS ```bash chmod +x Install-Mac.sh ./Install-Mac.sh # Then start app ./Launch-Unix.sh ``` ### Manual Installation ```bash # Clone the repository git clone https://github.com/roman-ryzenadvanced/NomadArch-v1.0.git cd NomadArch # Install dependencies npm install # Start the application npm run dev:electron ``` ### Building from Source ```bash # Build all packages npm run build # Or build individual packages npm run build:ui # Build UI npm run build:server # Build server npm run build:electron # Build Electron app ``` --- ## Features ### Core Features - 🤖 **Multi-Provider AI Support** - GLM 4.7, Anthropic, OpenAI, Google, Qwen, Ollama (local) - 🖥️ **Electron Desktop App** - Native feel with modern web technologies - 📁 **Workspace Management** - Organize your projects efficiently - 💬 **Real-time Streaming** - Live responses from AI models - 🔧 **Smart Fix** - AI-powered code error detection and fixes - 🏗️ **Build Integration** - One-click project builds - 🔌 **Ollama Integration** - Run local AI models for privacy ### UI/UX Highlights - ⚡ **MULTIX Mode** - Multi-task parallel AI conversations with live token counting - 🛡️ **SHIELD Mode** - Auto-approval for hands-free operation - 🚀 **APEX Mode** - Autonomous AI that chains tasks together - 📊 **Live Token Counter** - Real-time token usage during streaming - 💭 **Thinking Indicator** - Animated visual feedback when AI is processing - 🎨 **Modern Dark Theme** - Beautiful, eye-friendly dark interface - 🖱️ **Detailed Tooltips** - Hover over any button for explanations --- ## What's New in NomadArch ### Major Improvements Over Original CodeNomad #### 🎨 Branding & Identity - ✅ **New Branding**: "NomadArch" with proper attribution to OpenCode - ✅ **Updated Loading Screen**: New branding with fork attribution - ✅ **Updated Empty States**: All screens show NomadArch branding #### 🔐 Qwen OAuth Integration - ✅ **Fixed OAuth Flow**: Resolved "Body cannot be empty" error in Qwen authentication - ✅ **Proper API Bodies**: POST requests now include proper JSON bodies - ✅ **Fixed Device Poll Schema**: Corrected Fastify schema validation for OAuth polling #### 🚀 MULTIX Mode Enhancements - ✅ **Live Streaming Token Counter**: Visible in header during AI processing - ✅ **Thinking Roller Indicator**: Animated indicator with bouncing dots - ✅ **Token Stats Display**: Shows input/output tokens processed - ✅ **Auto-Scroll**: Intelligent scrolling during streaming #### 🖥️ UI/UX Improvements - ✅ **Detailed Button Tooltips**: Hover over any button for detailed explanations - AUTHED: Authentication status explanation - AI MODEL: Model selection help - SMART FIX: AI code analysis feature - BUILD: Project compilation - APEX: Autonomous mode description - SHIELD: Auto-approval mode - MULTIX MODE: Multi-task interface - ✅ **Bulletproof Layout**: Fixed layout issues with Editor/MultiX panels - ✅ **Overflow Handling**: Long code lines don't break layout - ✅ **Responsive Panels**: Editor and chat panels properly sized #### 📂 File Editor Improvements - ✅ **Proper File Loading**: Files load correctly when selected in explorer - ✅ **Line Numbers**: Clean line number display - ✅ **Word Wrap**: Long lines wrap instead of overflowing #### 🔧 Developer Experience - ✅ **Disabled Auto-Browser Open**: Dev server no longer opens browser automatically - ✅ **Unified Installers**: One-click installers for Windows, Linux, and macOS - ✅ **Enhanced Launchers**: Auto-fix capabilities, dependency checking, build verification - ✅ **Port Conflict Detection**: Warns if default ports are in use - ✅ **Error Recovery**: Provides actionable error messages with fixes #### 🐛 Bug Fixes - ✅ Fixed Qwen OAuth "empty body" errors - ✅ Fixed MultiX panel being pushed off screen when Editor is open - ✅ Fixed top menu/toolbar disappearing when file is selected - ✅ Fixed layout breaking when scrolling in Editor or Chat - ✅ Fixed auto-scroll interrupting manual scrolling - ✅ Fixed sessions not showing on workspace first entry --- ## Button Features Guide | Button | Description | |--------|-------------| | **AUTHED** | Shows authentication status. Green = connected, Red = not authenticated | | **AI MODEL** | Click to switch between AI models (GLM 4.7, Claude, GPT, etc.) | | **SMART FIX** | AI analyzes your code for errors and automatically applies fixes | | **BUILD** | Compiles and builds your project using detected build system | | **APEX** | Autonomous mode - AI chains actions without waiting for approval | | **SHIELD** | Auto-approval mode - AI makes changes without confirmation prompts | | **MULTIX MODE** | Opens multi-task pipeline for parallel AI conversations | --- ## Folder Structure ``` NomadArch/ ├── Install-Windows.bat # Windows installer with dependency checking ├── Install-Linux.sh # Linux installer with distro support ├── Install-Mac.sh # macOS installer with Apple Silicon support ├── Launch-Windows.bat # Windows launcher with auto-fix ├── Launch-Dev-Windows.bat # Windows developer mode launcher ├── Launch-Unix.sh # Linux/macOS launcher ├── packages/ │ ├── electron-app/ # Electron main process │ ├── server/ # Backend server (Fastify) │ ├── ui/ # Frontend (SolidJS + Vite) │ ├── tauri-app/ # Tauri alternative desktop app │ └── opencode-config/ # OpenCode configuration ├── README.md # This file └── package.json # Root package manifest ``` --- ## Requirements - **Node.js**: v18 or higher - **npm**: v9 or higher - **Git**: For version control features - **OS**: Windows 10+, macOS 11+ (Big Sur), or Linux (Ubuntu 20.04+, Fedora, Arch, OpenSUSE) ### Platform-Specific Requirements **Windows**: - Administrator privileges recommended for installation - 2GB free disk space **Linux**: - Build tools (gcc, g++, make) - Package manager (apt, dnf, pacman, or zypper) **macOS**: - Xcode Command Line Tools - Homebrew (recommended) - Rosetta 2 for Apple Silicon (for x86_64 compatibility) --- ## Troubleshooting ### "Dependencies not installed" Error Run the installer script first: - Windows: `Install-Windows.bat` - Linux: `./Install-Linux.sh` - macOS: `./Install-Mac.sh` ### "opencode not found" or Workspace Creation Fails The installer should automatically install OpenCode CLI. If it fails: **Option 1 - Manual npm install:** ```bash npm install -g opencode-ai@latest ``` **Option 2 - Manual download:** 1. Visit: https://github.com/sst/opencode/releases/latest 2. Download the appropriate ZIP for your platform: - Windows: `opencode-windows-x64.zip` - Linux x64: `opencode-linux-x64.zip` - Linux ARM64: `opencode-linux-arm64.zip` - macOS Intel: `opencode-darwin-x64.zip` - macOS Apple Silicon: `opencode-darwin-arm64.zip` 3. Extract and place `opencode` or `opencode.exe` in the `bin/` folder ### Port 3000 or 3001 Already in Use The launchers will detect port conflicts and warn you. To fix: 1. Close other applications using these ports 2. Check for running NomadArch instances 3. Kill the process: `taskkill /F /PIDMade with ❤️ by NeuralNomadsAI
Forked from CodeNomad by OpenCode