IDE Assistant with Ollama

A VS Code extension that provides an AI-powered coding assistant, using a locally hosted language model via Ollama. This extension integrates directly into your VS Code environment, offering contextual code assistance, explanations, and suggestions.

Features

• AI Chat Interface: Interact with the AI assistant directly within VS Code
• Context-Aware Responses: Attach files to provide context for more accurate assistance
• Split View Integration: Automatically splits your editor, keeping code and AI assistance side by side
• File Context Management:
  • Add multiple files as context
  • View attached files in the UI
  • Remove individual files or clear all context

Requirements

• VS Code version 1.97.0 or higher
• Ollama installed and running locally
• DeepSeek model pulled in Ollama (ollama pull deepseek-r1:1.5b)

Installation

1. Install Ollama from ollama.ai
2. Pull the DeepSeek model:

   # Depending on your hardware:
   ollama pull deepseek-r1:1.5b
   

3. Install the extension from VS Code Marketplace (or load from source)
4. Ensure Ollama is running (ollama serve)

Usage

1. Open the AI Assistant panel:

   • Click the AI icon in the activity bar, or
   • Use the command palette (Ctrl+Shift+P) and search for "Open AI Assistant Panel"

2. Ask questions:

   • Type your question in the input box
   • Press Enter or click the send button

3. Add context:

   • Click the "Add context" button
   • Select one or more files
   • The AI will consider these files when answering questions

Extension Settings

This extension contributes the following settings:

• soft-assist.modelName: The Ollama model to use (default: 'deepseek-r1:1.5b')
• soft-assist.ollamaUrl: URL of the Ollama server (default: 'http://localhost:11434')

Known Issues

• Requires Ollama to be running locally
• Large files may take longer to process
• Currently only supports text-based files for context

Development

1. Clone the repository
2. Install dependencies:

   npm install
   

3. Build the extension:

   npm run compile
   

4. Press F5 to start debugging

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Release Notes

0.0.1

Initial release of Assistant:

• Basic chat interface
• File context support
• Split view integration
• DeepSeek model integration

Acknowledgments

• Built with Ollama and DeepSeek

Tips