A powerful terminal-based chat interface for OpenRouter AI models built with React and Ink. OpenRChat provides a clean, interactive command-line experience for conversing with various AI models through the OpenRouter API.
OpenRChat is a React-based Terminal User Interface (TUI) application that brings AI conversations directly to your command line. Built with Ink and TypeScript, it offers a seamless chat experience with multiple AI models available through OpenRouter's unified API.
- Terminal-First AI Interactions: Chat with AI models without leaving your terminal environment
- Multi-Model Support: Switch between different AI models (GPT, Claude, DeepSeek, etc.) in real-time
- Developer-Friendly: XML-based configuration system and clean architecture for easy customization
- Conversation Management: Full conversation history with save/restore capabilities
- Interactive Terminal UI: Clean, responsive chat interface built with React/Ink
- Real-Time Model Switching: Switch between AI models mid-conversation with
/swmodel - Smart Input Handling: Automatic truncation display for long text (shows
[text #]while preserving full content) - Conversation Persistence: Save conversations with timestamps using
/save - XML Configuration: Easily manage models and system prompts through XML files
- Full Message History: Complete conversation context sent to AI models
- Slash Commands: Built-in commands for session management
- Error Handling: Robust API error handling with user-friendly messages
- Long Text Placeholder System: Cleanly handles large text inputs without cluttering the interface
- Dynamic Model Addition: Add new models through the UI that persist to configuration
- Status Bar: Always-visible current model and command information
- Keyboard Navigation: Arrow key navigation for model selection
- Scrollable History: Navigate through long conversations with smooth scrolling
- Node.js 18+ (specified in package.json engines)
- OpenRouter API Key (get one at openrouter.ai)
-
Clone the repository
git clone <repository-url> cd openrouter
-
Install dependencies
npm install
-
Set up your OpenRouter API key
# Option 1: Export for current session export OPENROUTER_API_KEY="your_api_key_here" # Option 2: Add to your shell profile (.bashrc, .zshrc, etc.) echo 'export OPENROUTER_API_KEY="your_api_key_here"' >> ~/.zshrc source ~/.zshrc # Option 3: Create a .env file (if supported) echo "OPENROUTER_API_KEY=your_api_key_here" > .env
-
Build the application
npm run build
-
Run OpenRChat
npm start
For development with hot reloading:
npm run dev| Variable | Required | Description | Default |
|---|---|---|---|
OPENROUTER_API_KEY |
✅ | Your OpenRouter API key | None |
OpenRChat uses two XML configuration files in the root directory:
<?xml version="1.0" encoding="UTF-8"?>
<models>
<model id="deepseek/deepseek-chat-v3.1" name="DeepSeek Chat v3.1" />
<model id="openai/gpt-5-chat" name="GPT-5 Chat" />
<model id="anthropic/claude-opus-4.1" name="Claude Opus 4.1" />
<model id="x-ai/grok-4" name="Grok 4" />
</models><?xml version="1.0" encoding="UTF-8"?>
<config>
<system>You are a helpful AI assistant. Please provide clear, concise, and accurate responses to user queries.</system>
</config>- Automatic Fallbacks: Missing XML files won't crash the application
- Runtime Updates: New models added via UI are automatically saved to
models.xml - Custom System Prompts: Modify
system.xmlto change the default AI behavior
-
Start the application
npm start
-
Begin chatting
OpenRChat v0.1.0 - Connected to OpenRouter Current Model: DeepSeek Chat v3.1 You: Hello! Can you help me with Python? AI: Hello! I'd be happy to help you with Python. What specific Python topic or problem would you like assistance with? -
Use commands as needed
You: /swmodel [Model selector appears with arrow key navigation] You: /save my-python-conversation Conversation saved to: my-python-conversation.json
OpenRChat automatically handles long text inputs:
# When pasting or typing long text (>200 chars or >2 lines)
You: [text 1] # Shows placeholder in UI
AI: [Full original text is sent to AI model]- Switch models mid-conversation: Use
/swmodelto select different AI models for different tasks - Reset conversation: Use
/resetto start fresh while keeping your session active - Save conversations: Use
/saveto preserve important chat sessions with timestamps - Clean exit: Use
/exitto close the application gracefully
| Command | Description | Usage |
|---|---|---|
/help |
Show help message with all commands | /help |
/swmodel |
Switch AI model | /swmodel (opens interactive selector) |
/save |
Save conversation with AI-generated filename | /save |
/listchats |
List all saved conversations with dates | /listchats |
/newchat |
Save current conversation and start fresh | /newchat |
/exit |
Exit the application | /exit |
You: /swmodel
┌─ Select Model ─────────────────────┐
│ ↑/↓: Navigate Enter: Select Esc: Cancel │
├────────────────────────────────────┤
│ > DeepSeek Chat v3.1 │
│ GLM 4.5V │
│ GPT-5 Chat │
│ Claude Opus 4.1 │
│ Grok 4 │
│ │
│ + Add new model │
└────────────────────────────────────┘-
Fork and clone the repository
git fork <repository-url> git clone <your-fork-url> cd openrouter
-
Install dependencies
npm install
-
Set up your development environment
export OPENROUTER_API_KEY="your_test_api_key" npm run dev
- New Model Support: Add support for additional OpenRouter models
- UI Enhancements: Improve the terminal interface experience
- Configuration Options: Extend XML configuration capabilities
- Performance Optimization: Improve response times and memory usage
- Documentation: Enhance user documentation and examples
- Error Handling: Improve error messages and user experience
For technical details and code integration examples, see codeintegration.md.
ISC License - See LICENSE file for details.
OpenRChat v0.1.0 - Bringing AI conversations to your terminal.
For support or questions, please open an issue on the project repository.
For developers interested in code integration and technical details, see codeintegration.md.