Ollama Code Pilot

⚠️ THIS IS CURRENTLY EXPERIMENTAL ONLY

Local AI-powered code completion for Visual Studio Code using Ollama.

Features

• Inline Code Completions: Get AI-powered code suggestions as you type
• Multiple Model Support: Use any Ollama model for code completion
• Local Inference: All processing happens on your machine - no data leaves your device
• Customizable Settings: Configure trigger characters, temperature, and max tokens
• Status Monitoring: Built-in sidebar view to monitor Ollama service status
• Model Management: Easy model selection and switching through the UI

Requirements

• Visual Studio Code v1.74.0 or higher
• Ollama installed and running locally
• At least one code completion model installed (e.g., codellama)

Installation

From Source

1. Clone this repository
2. Install dependencies:

   npm install
   

3. Compile the extension:

   npm run compile
   

4. Open in VS Code and press F5 to run

From VSIX

1. Build the extension:

   npm run package
   

2. Install the generated .vsix file in VS Code

Prerequisites

1. Install Ollama from ollama.ai
2. Pull a code completion model:

   ollama pull codellama
   

3. Start the Ollama service:

   ollama serve
   

Usage

Inline Completions

Simply start typing in any file. The extension will automatically suggest completions based on your code context.

Command Palette

• Ollama: Complete with Model - Generate completion with a specific model
• Ollama: Select Model - Change the default model
• Ollama: Show Status - Check Ollama service status

Status Bar

The status bar shows the current model and connection status. Click it to see detailed information.

Sidebar

The Ollama Code Pilot sidebar provides:

• Real-time service status
• List of available models
• Quick access to settings

Development

Prerequisites

• Node.js v20.x or higher
• npm v7.x or higher
• Visual Studio Code
• Ollama installed locally

Setting Up Local Development Environment

1. Clone the repository:

   git clone https://github.com/il-dat/open-code-assistant.git
   cd open-code-assistant
   

2. Install dependencies:

   npm install
   

3. Install and configure Ollama (if not already done):

   # Install Ollama (macOS)
   brew install ollama
   
   # Or download from https://ollama.ai/
   
   # Start Ollama service
   ollama serve
   
   # Pull a code model
   ollama pull codellama
   

4. Compile TypeScript:

   npm run compile
   # Or watch mode for development
   npm run watch
   

Available Scripts

Code Quality Tools

ESLint Configuration

The project uses ESLint with TypeScript support. Configuration is in .eslintrc.json.

Run linting:

npm run lint

ESLint rules enforced:

• TypeScript naming conventions
• Semicolon usage
• Curly braces for all control structures
• Strict equality checks
• No literal throws

To auto-fix some issues:

npx eslint src --ext ts --fix

Running in Development

1. Open in VS Code:

   code .
   

2. Start debugging (choose one):

   • Press F5 to run the extension
   • Use Run → Start Debugging from menu
   • Use the Run view and select "Run Extension"

3. Test the extension:

   • A new VS Code window will open with the extension loaded
   • Open any code file and start typing to see completions
   • Check the Ollama status in the status bar
   • Use Command Palette (Cmd+Shift+P / Ctrl+Shift+P) for Ollama commands

Testing

Run the test suite:

npm test

Run tests with coverage report:

npm run test:coverage

The coverage report will show:

• Text summary in the terminal
• HTML report in coverage/index.html

Current coverage thresholds:

• Lines: 30%
• Functions: 30%
• Branches: 30%
• Statements: 30%

Note: VS Code extensions typically have lower coverage due to the heavy integration with VS Code APIs. The thresholds are set accordingly.

The test suite includes:

• Unit tests for the Ollama client
• Integration tests for the completion provider
• Extension activation tests
• Core logic unit tests

Debugging Tips

1. View extension logs:

   • Open Output panel (View → Output)
   • Select "Ollama Code Pilot" from dropdown

2. Debug completion provider:

   • Set breakpoints in src/completion_provider.ts
   • Use Debug Console to inspect variables

3. Test with different models:

   # Pull additional models
   ollama pull deepseek-coder
   ollama pull starcoder
   

Building for Production

1. Update version in package.json

2. Build and package:

   npm run compile
   npm run package
   

3. Test the VSIX:

   • Install: code --install-extension ollama-code-pilot-0.1.0.vsix
   • Or drag the .vsix file into VS Code Extensions view

Publishing to VS Code Marketplace

The extension is automatically published when a version tag is pushed:

# Update version in package.json
npm version patch  # or minor/major

# Push changes and tag
git push origin main
git push origin v0.1.0  # Use the new version number

The GitHub Actions workflow will:

• Run tests on multiple platforms
• Build and package the extension
• Publish to VS Code Marketplace

Prerequisites:

• VS Code Marketplace publisher account
• VSCE_PAT secret configured in GitHub repository

Project Structure

open-code-assistant/
├── src/                    # Source code
│   ├── extension.ts        # Extension entry point
│   ├── ollama_client.ts    # Ollama API client
│   ├── completion_provider.ts # Inline completion logic
│   ├── commands.ts         # Command implementations
│   ├── models_provider.ts  # Model tree view provider
│   └── status_view_provider.ts # Status webview provider
├── test/                   # Test files
│   └── suite/             # Test suites
├── media/                  # Webview assets
├── resources/              # Extension resources
├── .vscode/               # VS Code configuration
├── .eslintrc.json         # ESLint configuration
├── tsconfig.json          # TypeScript configuration
└── package.json           # Extension manifest

Configuration

Access settings through VS Code's settings UI or settings.json:

Supported Models

This extension works with any Ollama model, but these are recommended for code completion:

• codellama - Meta's Code Llama model optimized for code
• deepseek-coder - DeepSeek's coding model
• starcoder - StarCoder model for code generation
• phind-codellama - Phind's fine-tuned Code Llama

To install a model:

ollama pull <model-name>

Troubleshooting

Ollama Service Not Running

If you see "Ollama: Offline" in the status bar:

1. Ensure Ollama is installed
2. Start the service with ollama serve
3. Check that the service URL in settings matches your Ollama instance

No Completions Appearing

1. Check that a model is installed (ollama list)
2. Verify the model name in settings matches an installed model
3. Try increasing maxTokens in settings
4. Check the Output panel for error messages

Performance Issues

• Use smaller, quantized models for faster inference
• Reduce maxTokens for quicker responses
• Consider using GPU acceleration if available

Attribution

This extension is built with:

• Ollama - Local LLM inference
• llama.cpp - High-performance inference engine
• Meta Llama 3 - When using LLaMA-based models

Built with Meta Llama 3 (when using LLaMA-based models)

License

This extension is licensed under the Apache License 2.0. See LICENSE for details.

About Infinite Lambda

Infinite Lambda is a cloud and data consultancy. We build strategies, help organisations implement them and pass on the expertise to look after the infrastructure.

We are an Elite Snowflake Partner, a Platinum dbt Partner and two-times Fivetran Innovation Partner of the Year for EMEA.

Naturally, we love exploring innovative solutions and sharing knowledge, so go ahead and:

🔧 Take a look around our Git
✏️ Browse our tech blog

We are also chatty, so:
#️⃣ Follow us on LinkedIn
👋🏼 Or just get in touch