Skip to content
| Marketplace
Sign in
Visual Studio Code>Programming Languages>AutoDev Agent — Autonomous AI CoderNew to Visual Studio Code? Get it now.
AutoDev Agent — Autonomous AI Coder

AutoDev Agent — Autonomous AI Coder

Diwakar2005

|
1 install
| (0) | Free
Multi-agent AI system: intent → code → tests → deployment. Fix entire projects in one click.
Installation
Launch VS Code Quick Open (Ctrl+P), paste the following command, and press enter.
Copied to clipboard
More Info

AutoDev Agent - Autonomous AI Coder for VS Code

Intent -> Plan -> Execute -> Done. No manual coding for boilerplate, debugging, or scaffolding.

VS Code Marketplace License: MIT Open Source


What It Does

AutoDev Agent is a multi-agent AI system built inside VS Code. It replaces hours of developer work with a single command - no subscription, no cloud lock-in, fully open source.

Feature What happens
Fix Entire Project Scans all files, detects every bug, fixes everything, shows diff before applying
Generate from Intent Type "Build login system with JWT" and files are created, code is written, ready to run
Fix and Explain Selection Select broken code and get explanation plus working fix side by side
Refactor File Choose goals (performance, readability, security) and get production-ready code
Generate Tests Full test suite written automatically for any file
Run and Self-Heal Runs your code and automatically fixes any errors that appear

Quick Start

1. Install the extension

From source:

git clone https://github.com/Diwakarsrd/AutoDev.git
cd AutoDev
npm install
npm run compile

Press F5 in VS Code to launch the Extension Development Host.

2. Configure your LLM provider

Ctrl+Shift+P -> AutoDev: Configure LLM Provider

Free LLM Options

Ollama (Recommended - 100% local, 100% free)

# Linux/Mac
curl -fsSL https://ollama.com/install.sh | sh
# Windows: https://ollama.com/download

ollama pull qwen2.5-coder:7b
ollama serve

Google Gemini (Free tier - 1500 requests/day)

  1. Go to aistudio.google.com -> Get API Key -> Create key
  2. AutoDev: Configure LLM Provider -> Select Gemini -> Paste key

Groq (Free tier - fastest inference)

  1. Go to console.groq.com -> Create free account -> Generate API key
  2. AutoDev: Configure LLM Provider -> Select Groq -> Paste key

Keyboard Shortcuts

Shortcut Action
Ctrl+Shift+F Fix Entire Project
Ctrl+Shift+G Generate from Intent
Ctrl+Shift+A Open Agent Panel
Ctrl+Shift+H Run and Self-Heal

Usage Examples

Generate from Intent

Ctrl+Shift+G -> type your intent -> Execute Plan

Examples:

  • "Build a REST API with Express and JWT auth"
  • "Create a React dashboard with charts"
  • "Make a Python CLI tool that converts CSV to JSON"
  • "Build a FastAPI app with SQLAlchemy"

Fix Entire Project

Ctrl+Shift+F -> Confirm -> Review diff -> Apply

Scans your project files (excludes venv, node_modules, etc.) and fixes all bugs found.

Fix and Explain Selection

  1. Select any code lines
  2. Right-click -> AutoDev: Fix and Explain Selection
  3. Choose Apply to Selection (replaces only selected lines) or Show Diff

Run and Self-Heal

  1. Ctrl+Shift+H
  2. Enter command: npm test / python main.py / go build ./...
  3. AutoDev runs it, detects errors, fixes the code, runs again automatically

Architecture

User Input -> Planner Agent -> Validator Agent -> Executor Agent -> Debugger Agent -> Output

Planner:   Breaks intent into executable steps
Executor:  Creates files, installs packages, runs commands
Debugger:  Finds bugs, fixes code, generates tests
Validator: Checks quality (0-100 score) before applying changes

Supported LLM Providers

Provider Cost Notes
Ollama Free Local, private, no internet needed
Google Gemini Free tier 1500 requests/day free
Groq Free tier Fastest inference
Anthropic Claude Paid Best code quality
OpenAI Paid GPT-4o and variants
OpenAI-compatible Varies LM Studio, LocalAI, any endpoint

Supported Languages

TypeScript, JavaScript, Python, Java, Go, Rust, Ruby, C, C++, C#, PHP, Swift, Kotlin, Vue, Svelte, CSS, SCSS, HTML, Bash, SQL, JSON, YAML, Markdown


Configuration

{
  "autodev.provider": "ollama",
  "autodev.ollamaModel": "qwen2.5-coder:7b",
  "autodev.selfHealingEnabled": true,
  "autodev.selfHealingMaxRetries": 3,
  "autodev.conversationMemoryEnabled": true,
  "autodev.showDiffBeforeApply": true,
  "autodev.maxFilesPerScan": 50,
  "autodev.excludePatterns": [
    "node_modules", ".git", "dist", "build",
    "venv", ".venv", "__pycache__", "site-packages"
  ]
}

Privacy

  • Ollama mode: All code stays on your machine. Zero data sent anywhere.
  • Cloud providers: Code is sent to the respective API provider.
  • The extension never stores your code or API keys externally.

Contributing

PRs welcome. See CONTRIBUTING.md to get started.


License

MIT - free to use, modify, distribute, and sell.

  • Contact us
  • Jobs
  • Privacy
  • Manage cookies
  • Terms of use
  • Trademarks
© 2026 Microsoft