A clean, elegant command-line interface for chatting with local AI models via Ollama, featuring real-time Markdown rendering.
- 💬 Interactive chat interface with Ollama models
- 📝 Real-time Markdown rendering (code blocks, tables, lists, etc.)
- 🔄 Full conversation context preservation
- 🎯 Customizable system prompts
- 🌡️ Adjustable temperature settings
- 💾 Conversation saving to files
pip install haiku-ollama Make sure you have Ollama installed and running before using Haiku.
Start a conversation with the default model (llama3.1:8b):
haiku | Option | Description |
|---|---|
| --model | Specify which Ollama model to use (default: llama3.1:8b) |
| --keep-context, -k | Maintain full conversation history between prompts |
| --system, -s | Set a custom system prompt to guide the model's behavior |
| --temperature, -t | Set temperature (0.0-1.0) - lower values are more deterministic |
| --save | Save the conversation to a specified file |
Using a specific model:
haiku --model mistral:7b Preserving conversation context:
haiku --keep-context Setting a system prompt:
haiku --system "You are an expert programmer who explains code concisely" Adjusting temperature:
haiku --temperature 0.2 Saving your conversation:
haiku --save conversation.md Combining multiple options:
haiku --model codellama --keep-context --system "You write Python code" --temperature 0.3 --save coding_session.md To exit the program, simply type exit or bye, or press Ctrl+C.
- Python 3.8+
- Ollama installed and running
- Python packages: ollama, rich
Contributions are welcome! Feel free to submit issues or pull requests.
MIT License