Paita is textual assistant for your terminal that supports multiple AI Services and models.
- Supports Multiple AI Services: Paita integrates with a variety of AI services through the LangChain library. If AI service is compatible with LangChain then it can be used also with Paita.
- Textual User Interface on your terminal: Paita is based on Textual and provides a sophisticated user interface right within your terminal, combining the complexity of a GUI with console simplicity.
- Cross-Platform Compatibility: Paita is compatible with Windows, macOS, and Linux systems across most terminals; if Python runs in your environment and Textual supports it, then Paita will work.
- OpenAI
- AWS Bedrock
- Ollama (local models)
- Python 3.8.1+
- Access to AI Service and configured in terminal
Install using pip (or pipx)
pip install paita
Run and enjoy!
paita
Paita is textual ui application so using keyboard shortcuts is recommended:
- Use
tabandshift+tabto navigate between input field, send-button and question/answer boxes - While question/answer box is focus use
enterto "focus-in" andescto "focus-out" - Use
cto copy content from question/answer box - Contextual keyboard shortcuts are shown at the bottom of the UI
OpenAI usage requires valid api key in environment variable.
export OPENAI_API_KEY=<OpenAI API Key>
Enable AI model access in AWS Bedrock. Configure aws credential access accordingly.
Ollama enables running chat models locally.
Install ollama for operating system or use official (docker image)[https://hub.docker.com/r/ollama/ollama]
Once ollama installed pull desired model from a registry e.g.
ollama pull llama3
By default paita connects to local Ollama endpoint. Optionally you can configure endpoint url with env variable:
export OLLAMA_ENDPOINT=<protocol>://<ollama-host-address>:<ollama-host-port>