ollama - Local LLM Management & Interaction

The ollama module provides an enhanced CLI interface for Ollama, enabling users to easily manage local models, perform translations, and interact with various AI architectures.

When to Activate

• When the user wants to install or manage Ollama services.
• When pulling models from the Ollama registry (e.g., Llama 3, Mistral).
• When chatting with local models or using them for file-based translation.
• When using an interactive UI to browse and download models.

Core Principles & Rules

• Registry Interaction: Use pull and push to sync models with the Ollama registry.
• Convenience Aliases: Support for @o alias for rapid local LLM chat.
• File Context: Leverage the --file flag to provide local context to models.

Additional Scenarios

• Interactive Browsing: Use la for an interactive UI to explore available models.
• Service Control: Use serve to manually start the Ollama backend.

Patterns & Examples

Pull and Run

# Download and start a chat session with Mistral
x ollama pull mistral
@o "How does a vector database work?"

Translate with File Context

# Translate local documents using a local model
@o --file ./abstract.en.md "Translate to Chinese"

Interactive UI

# Browse the Ollama registry interactively
x ollama la

Checklist

• Ensure the Ollama service is installed and running.
• Verify if the desired model has been pulled to local storage.
• Check compatibility of the --file input with the model's context window.