Ollama Code Agent - VS Code Extension

A free, local AI-powered code generation agent for Visual Studio Code using Ollama.

Features

• Inline Code Completion - Get AI-powered code suggestions as you type
• Chat Panel with Copilot-Style Modes - Switch between ⚡ Agent (auto-creates files on prompt/response) and 💬 Ask (Q&A mode)
• Model Selector Dropdown in Chat - Switch between installed Ollama models directly inside the chat header UI
• Autonomous File Creation - Automatically creates, writes, and opens files in your workspace root when Agent mode is active or when you request file creation
• Image & Screenshot Support - Paste (Ctrl+V), drag & drop, or attach images into chat for multimodal AI vision models (e.g. llava, llama3.2-vision)
• Code from Comments - Generate code implementations from natural language comments (Ctrl+Shift+G)
• Fully Local & Free - All processing happens on your machine, no API keys needed

Prerequisites

1. Install Ollama: Download from ollama.ai
2. Pull a coding model:

   # Recommended models for code generation:
   ollama pull deepseek-coder:6.7b    # Best balance of speed/quality
   ollama pull codellama:7b            # Meta's code model
   ollama pull starcoder2:7b           # BigCode's model
   

3. Ensure Ollama is running: The Ollama service must be running (default: http://localhost:11434)

Installation

From Source

cd C:\Arul\Projects\VCode\VSCodeAgentPlugin
npm install
npm run compile

Then press F5 in VS Code to launch the extension in debug mode.

Run the extension-host checks with:

npm test

Package as VSIX

npm run package

Then install the .vsix file via VS Code: Extensions → ⋯ → Install from VSIX.

Usage

Inline Completion

Simply start typing code — the extension will suggest completions after a brief pause. Press Tab to accept.

Chat Panel

1. Click the Ollama icon in the Activity Bar (left sidebar)
2. Type your coding question and press Enter
3. Click "Insert" on any code block to paste it into your editor

Generate Code from Comments

1. Write a comment describing what you want:

   # Function to calculate fibonacci numbers recursively with memoization
   

2. Place your cursor on or below the comment
3. Press Ctrl+Shift+G (or run command "Ollama: Generate Code from Comment")

Select Model

Run command "Ollama: Select Model" to pick from your installed Ollama models.

Configuration

Agent responses are always confirmed before files are written, and generated paths must remain inside the open workspace.

Recommended Models

Troubleshooting

• "Cannot connect to Ollama": Ensure Ollama is running (ollama serve)
• Slow completions: Try a smaller model like deepseek-coder:1.3b
• No suggestions appearing: Check that inline completion is enabled in settings

License

MIT