LocalPilot

The Privacy-First AI Pair Programmer for Visual Studio.
Bringing the power of local LLMs directly into your IDE with Ollama.

🌟 Overview

LocalPilot is a powerful Visual Studio extension that integrates local Large Language Models (LLMs) via Ollama. It provides a seamless, high-performance coding experience without the need for cloud-based subscriptions or data privacy concerns.

🚀 Key Features

🛡️ Why LocalPilot?

• 🔒 Absolute Privacy: Your source code stays on your machine. No telemetry, no cloud hooks, no data leakage. Perfect for enterprise and sensitive projects.
• ⚡ Zero Latency: No waiting for cloud API responses. Local inference provides near-instantaneous completions.
• 💰 One-time Setup, Zero Cost: No recurring subscriptions. Use the power of your own hardware to fuel your development.
• 🎨 Native Experience: Designed to feel like a built-in Visual Studio feature, supporting both Light and Dark themes natively.

🛠️ Getting Started

1️⃣ Prerequisites

You must have Ollama installed and running on your machine.

• Download: ollama.com
• Launch a Model: We recommend code-centric models like llama3, codellama, or phi3.

  ollama run llama3
  

2️⃣ Installation

1. Visit the Visual Studio Marketplace.
2. Click Download, or search for "LocalPilot" within the Visual Studio Extension Manager:
   • Extensions > Manage Extensions > Online
3. Restart Visual Studio to complete the installation.

3️⃣ Configuration

Navigate to Tools > Options > LocalPilot > Settings.

1. Ollama Base URL: Usually http://localhost:11434. Click "Test Connection" to verify.
2. Model Assignments: Assign preferred models for Chat and Inline Completions.

[!TIP] For optimal performance, use a lightweight model like phi3 or starcoder2:3b for Inline Completions, and a larger model like llama3:8b or deepseek-coder for the Chat Assistant.

📖 Usage Guide

💡 Inline Completion

Simply start typing in any supported file. LocalPilot will provide translucent "ghost-text" suggestions.

• Tab: Accept the suggestion.
• Esc: Dismiss the suggestion.

⚡ Contextual Actions

Right-click on any code selection to access the LocalPilot menu:

• Explain Code: Breakdown complex logic.
• Generate Docs: Auto-generate XML/docstring comments.
• Refactor: Suggest improvements for readability and performance.

🤝 Contributing

We welcome community contributions! Whether it's bugs, features, or documentation, your help is appreciated.

🛠️ How to Help

1. Check Issues: See the Existing Issues to avoid duplicates.
2. Clear Reports: For bugs, include your VS version, Ollama model, and reproduction steps.
3. Pull Requests: Create a branch from main, ensure the project builds, and submit your PR with a clear description.

💻 Hardware Requirements

Since LocalPilot runs Large Language Models (LLMs) entirely on your local machine via Ollama, your hardware performance directly impacts the speed and responsiveness of AI suggestions.

🏁 Minimum Requirements

• CPU: Recent Multi-core processor (Intel i5/AMD Ryzen 5 or equivalent).
• RAM: 8GB (16GB+ strongly recommended for a smooth experience).
• GPU: 4GB VRAM (Dedicated NVIDIA or Apple Silicon GPU preferred for faster inference).
• Storage: 5GB+ for model storage (SSD/NVMe highly recommended).

🚀 Recommended for "Pro" Experience

• RAM: 32GB+ for handling larger models (13B+) alongside Visual Studio.
• GPU: NVIDIA RTX 3060/4060 or higher with 12GB+ VRAM.
• NVIDIA CUDA: Ensure latest drivers are installed for GPU acceleration.

[!IMPORTANT] LocalPilot is designed for efficiency, but because it performs all AI processing locally, it requires capable hardware. If suggestions feel slow, consider using a smaller, quantized model (e.g., phi3:mini or starcoder2:3b) in the settings.