Vidvat Assistant — AI Code Assistant for Visual Studio 2022

Vidvat Assistant is an AI-powered sidebar for Visual Studio 2022 with:

• Chat interface
• Code generation
• Code explanation
• Model selection (Qwen3-Coder, Codestral, CodeGemma)
• Local LLM inference using Ollama
• FastAPI backend
• WebView2 UI embedded inside VS

---

Features

Seamless Visual Studio Integration

• A top-right corner button (like Copilot) opens a floating panel.
• Four tabs: Chat, Code, Explain, Settings.

Local LLMs via Ollama

Vidvat supports local, private, GPU-accelerated models:

• qwen3-coder
• codegemma
• codestral

Models are automatically downloaded when selected.

Backend Included (FastAPI)

A lightweight backend provides:

• /set_model
• /chat
• /available_models

Full source included inside backend/.

---

Repository Structure

VidvatAssistant/ ├── # VSIX project in the root ├── backend/ # FastAPI + Ollama server └── README.md

---

Setup Instructions

1. Install Dependencies

Requires:

• Visual Studio 2022
• Python 3.10+
• Ollama installed (https://ollama.com)

2. Start Backend

cd backend pip install -r requirements.txt python main.py

3. Install Extension

Download the .vsix from Releases and install it.

4. Open Vidvat Assistant Panel

Click the V button in the top-right corner of Visual Studio.

---

Model Selection

Go to Settings → Select Model.

Choosing a model will:

1. Download it via ollama pull
2. Save it inside ~/.vidvat/config.json
3. Use it for all chat sessions

---

Technologies Used

• Visual Studio Extensibility SDK
• WebView2
• WPF
• FastAPI
• Python
• Ollama
• Docker (optional)

---

Feedback & Issues

Please report bugs and feature requests in the Issues tab.

---

License

Public license