rune/oai
1
0
Fork 0
Code Issues Pull Requests Releases 2 Activity
Files
93cd60b2717eb32aa1065d93723f9d79023d4992
oai/README.md

171 lines
12 KiB
Markdown
Raw Blame History

# oAI Chat App
A command-line interface (CLI) chat application for interacting with AI language models via OpenRouter. Seamlessly switch between 300+ models, attach files (images and text), enjoy real-time streaming responses, and manage settings persistently—all from your terminal. Perfect for developers, writers, and AI enthusiasts looking for a fast, feature-rich chat experience without leaving the command line.
[![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)](https://opensource.org/licenses/MIT)
## 🌟 Features
- **Model Flexibility**: Choose from hundreds of OpenRouter-supported models (e.g., GPT-4, Claude, Gemini) with substring-based search and selection—inspired by dynamic model retrieval in advanced CLIs [github.com](https://github.com/oop7/OrChat) for seamless user experience.
- **File Attachments**: Attach images (JPEG, PNG) or plain text files directly to messages using `@path` syntax, with automatic Base64 encoding for models that support modalities [github.com](https://github.com/rdhillbb/openRchat), enhancing multimodal interactions like in OrChat's smart picker.
- **Streaming Responses**: Enable real-time streaming for faster, live responses; toggle on/off via config commands (inspired by streaming in modern chat apps), with Ctrl+C cancellation support similar to advanced TUIs.
- **Persistent Configuration**: Securely store API keys, base URLs, and settings in a local SQLite database for reuse across sessions, leveraging cross-session persistence like in OrChat's conversation management [github.com](https://github.com/oop7/OrChat).
- **Rich UI in CLI**: Enhanced console output with Rich library panels, tables, and progress indicators for a polished terminal experience, drawing from Ink-based interfaces in React TUIs [github.com](https://github.com/rdhillbb/openRchat).
- **Arrow Key History Navigation**: Scroll through previous user inputs using up/down arrow keys for quick recall and editing, akin to command history in shells and CLIs like those using prompt_toolkit [prompt-toolkit.readthedocs.io](https://python-prompt-toolkit.readthedocs.io/en/stable/pages/tutorials/repl.html), for efficient iterative chatting.
- **Screen Clearing**: Instantly clear the terminal screen with the `/clear` command, maintaining session state while providing a fresh interface—similar to `/cls` or `/clear-screen` in advanced chat apps [github.com](https://github.com/oop7/OrChat), preventing visual clutter.
- **Clipboard Integration**: Copy AI responses to the clipboard with a single prompt.
- **Help Command**: Built-in `/help` with examples for all commands, making the app self-documenting and user-friendly [github.com](https://github.com/Piyush64-bit/ReadMe-Generator).
- **Error Handling**: Graceful prompts for missing configs, file issues, and model incompatibilities.
- **Cross-Platform**: Works on any OS with Python and internet access, no heavy dependencies required.
## 🖼️ Screenshots
[<img src="https://gitlab.pm/rune/oai/raw/branch/main/images/screenshot_01.png">](https://gitlab.pm/rune/oai/src/branch/main/README.md)
## 📋 Requirements
- **Python**: 3.9 or higher
- **OpenRouter API Key**: Free tier available at [openrouter.ai](https://openrouter.ai). Models like GPT-4 may require a paid plan.
- **Additional Dependencies**: Requires `prompt_toolkit` for interactive input features like arrow key history and multi-line support (install via `pip install prompt_toolkit`) [prompt-toolkit.readthedocs.io](https://python-prompt-toolkit.readthedocs.io/en/stable/pages/tutorials/repl.html).
- **Base Dependencies**: Listed in `requirements.txt` (generate automatically with tools like `pipreqs` if needed).
- **Internet Connection**: For API calls to OpenRouter.
## 🚀 Installation
1. **Clone the Repository**:
```bash
git clone https://github.com/yourusername/oai-chat-app.git # Replace with your actual repo URL
cd oai-chat-app
```
2. **Install Dependencies**:
Use `pip` for simplicity or `uv` for faster management:
```bash
# Using pip
pip install typer rich openrouter pyperclip requests prompt_toolkit
# Using uv
uv add typer rich openrouter pyperclip requests prompt_toolkit
# Using pip and requirements.txt (recommended)
pip3 install -r requirements.txt
```
3. **Set Up Configuration**:
- The app creates a SQLite database at `~/.config/oai/oai_config.db` on first run.
- Run the app once to set your API key: `python oai_chat.py chat`.
- For API key issues, visit [openrouter.ai/settings/keys](https://openrouter.ai/settings/keys).
4. **Make Executable (Optional)**:
```bash
chmod +x oai_chat.py # For Unix-based systems
```
Your app is now ready! Run `python oai_chat.py chat` to start chatting [github.com](https://github.com/waheed444/AI_Readme_Generator) inspired by simple setup in AI tools.
## 💻 Usage
Launch the app and interact via commands or direct chat:
```bash
python oai_chat.py chat
```
### Available Commands
| Command | Description | Example |
|---------------------|-----------------------------------------------------------------------------|:------------------------------------------------------------------------|
| `/help` | Display a table with examples for all commands and features, including new ones like `/clear`. | `/help` |
| `/model [search]` | Select or switch models; optional search for substring matching. | `/model gpt` (Filters to GPT models)<br>`/model` (Shows all models) |
| `/config api` | Update OpenRouter API key and reinitialize the client. | `/config api`<br>Enter key: sk-...<br>[Output: API key updated!] |
| `/config url` | Change the base URL for custom OpenRouter endpoints. | `/config url`<br>Enter URL: https://custom.openrouter.ai/api/v1 |
| /config model | "Set a default model that loads on startup; displays models for selection (persisted in DB, doesn't change current session model)." | `/config model gpt`<br />(Shows GPT models; user selects to set as default) |
| `/config stream on/off` | Toggle response streaming (default: on for real-time chat). | `/config stream off` (For full responses at once)<br>`/config stream on` |
| `/config` | View current configurations. | `/config` (Displays table: API Key, Base URL, Streaming, Database path) |
| `/clear` | Clear the terminal screen for a clean interface, preserving chat state. | `/clear`<br>[Output: Screen cleared. Ready for your next input!] |
| `exit`, `quit`, `bye` | Exit the app cleanly. | `exit` |
| Chatting with Files | Attach files using `@path` (e.g., images for vision models, text for context). | `Explain this @~/demo.png`<br>(Attaches image if model supports it) |
| Arrow Key Navigation | Use ↑/↓ arrow keys to scroll through and edit previous user inputs. | While at "You>" prompt, press ↑ for last message, ↓ for next. |
### Example Session
```
Welcome to oAI! Type your message, '/model [search]' to select/change model, '/config api/url/stream' to configure, '/help' for examples, or 'exit'/'quit' to end.
No model selected. Use '/model' to choose one.
You> /help
[Displays help table as above]
You> /model gpt
(Shows table of GPT models)
Enter model number: 1
Selected: GPT-4o (openai/gpt-4o)
You> Hello, how are you?
Streaming response... (Press Ctrl+C to cancel)
Hello! I'm doing great, thanks for asking. How can I help you today?
Type 'c' to copy response to clipboard, or press Enter to continue: c
Response copied to clipboard!
You> /clear
Screen cleared. Ready for your next input!
You> Analyze this text @~/notes.txt
Streaming response... (Press Ctrl+C to cancel)
Based on the attached file, the text discusses...
You> [↑ arrow key pressed - recalls previous message: Analyze this text @~/notes.txt]
[Edit as needed and press Enter]
You> exit
Goodbye!
```
- **Streaming**: Enabled by default for live responses; press Ctrl+C mid-stream to cancel (supported by most providers).
- **File Limits**: Max 10MB per file; only images for vision models, plain text otherwise. Supports `@~/relative/path` or absolute paths.
- **Arrow Key History**: Sessions retains a history per run (resets on exit), allowing quick re-use or modification of inputs for iterative conversations.
- **Error Recovery**: If no models fetch initially, check your API key. Prompts guide you through setup.
For more on OpenRouter models and features (e.g., tool calling), see [openrouter.ai/docs](https://openrouter.ai/docs).
## ⚙️ Configuration
All settings are stored securely in `~/.config/oai/oai_config.db` using SQLite:
- **API Key**: Required for authentication (set via `/config api`).
- **Base URL**: Defaults to `https://openrouter.ai/api/v1` (change for custom endpoints via `/config url`).
- **Streaming**: Toggles live responses (`on` for real-time, `off` for buffered). Defaults to `on`.
- **Database Path**: Automatically created; view full path with `/config`.
Edit any setting during a session without restarting. The database persists settings for future runs.
## 🛠️ Troubleshooting
- **No Models Available?** Ensure your API key is valid and has access to models [openrouter.ai](https://openrouter.ai/docs/api-keys). Check network connectivity.
- **Streaming Issues?** Some models/providers may fall back to non-streaming. Toggle with `/config stream off`.
- **File Attachment Errors?** Confirm paths are absolute (e.g., `@/home/user/file.png`) and files are under 10MB. Unsupported MIME types will be rejected.
- **Arrow Key Issues?** Ensure `prompt_toolkit` is installed; arrow keys work in supported terminals (TTY environments). Falls back gracefully in non-interactive modes.
- **Clear Screen Fails?** ANSI escape sequences are used cross-platform [usavps.com](https://usavps.com/blog/46203/), but may fall back to newlines in rare cases.
- **API Errors?** Re-set your key or check model compatibility (e.g., vision models for images).
- **Permission Denied?** Ensure the app has read access to attached files and the database directory.
- **Installation Fails?** Verify Python version and try `pip install --upgrade pip`. Use `uv` for dependency resolution [github.com](https://github.com/Herc-Ch/readme-generator).
- **Performance?** Large files or complex prompts may slow responses—trim inputs as needed.
If issues persist, open a GitHub issue with error logs.
## 🤝 Contributing
We welcome contributions to improve the oAI Chat App! Follow standard practices: fork, branch, test, and submit PRs. Please include updates to this README for new features.
## 📜 License
This project is licensed under the MIT License - see the [LICENSE](LICENSE) file for details.
## 🙏 Acknowledgments
- Built with OpenRouter for model access [openrouter.ai](https://openrouter.ai).
- Interactive input features inspired by prompt_toolkit [python-prompt-toolkit.readthedocs.io](https://python-prompt-toolkit.readthedocs.io/en/stable/pages/tutorials/repl.html) and advanced CLIs like OrChat [github.com](https://github.com/oop7/OrChat), openRChat [github.com](https://github.com/rdhillbb/openRchat), for history and clearing functionality.
- README structure inspired by AI generators for documentation polish [github.com](https://github.com/Herc-Ch/readme-generator), [github.com](https://github.com/owengretzinger/gitreadme), [github.com](https://github.com/Piyush64-bit/ReadMe-Generator), [github.com](https://github.com/waheed444/AI_Readme_Generator), [github.com](https://github.com/mrunankpawar/repo-readme-generator).
Star ⭐ if this app helps—it's open-source and free! Happy chatting!
Reference in New Issue View Git Blame Copy Permalink