v0.1.17 is now live

The Autonomous AI Engineer
Inside VS Code.

Multi-provider AI support, multi-agent workflows, surgical code editing, and intelligent automation. 100% Free. Bring Your Own Key.

mydaus-code-demo.mp4
Mydaus Code in action - Multi-agent workflows, surgical code editing, and intelligent automation

Core Capabilities

Everything you need to build faster with AI assistance

🌐 Multi-Provider Support

  • OpenRouter Integration: Access Claude, GPT-4, Gemini, and more premium models
  • Local LLM with Ollama: Run powerful models locally with zero cloud costs
  • Intelligent Model Selection: Automatic hardware profiling and recommendations
  • Seamless Switching: Switch between cloud and local providers on the fly
  • Extended Thinking: Enhanced reasoning for Claude 3.7+ and 4.6+ models

🤖 Multi-Agent Workflow Engine

  • 7 Specialized Personas: Architect, Developer, QA, DevOps, PM, Tech Writer, UI Architect
  • Workflow Orchestration: Predefined multi-step workflows with automatic handoffs
  • Context-Aware Routing: Role-specific system prompts and tool access
  • TDD Workflow Support: Built-in Test-Driven Development workflow
  • Shared Memory System: Persistent state across workflow steps

Powerful Features

Advanced capabilities for modern development workflows

🧠 Intelligent Automation

  • • Context-aware assistance
  • • Self-healing retry logic
  • • Autonomous multi-tool orchestration
  • • @ Mention files in prompts
  • • Anti-looping safety
  • • Vocal status updates

📝 Surgical Code Editing

  • • Search-and-replace patches
  • • Multi-file refactoring
  • • Syntax-aware editing
  • • Rollback support
  • • Visible editing in real-time
  • • File restriction guards

⚡ Terminal & Commands

  • • Interactive terminal execution
  • • Process management
  • • Safe command whitelist
  • • Auto-approval settings
  • • Output streaming
  • • Exit code detection

🔄 Smart Git Integration

  • • AI-generated commit messages
  • • Premium Git dashboard
  • • Proactive Git telemetry
  • • Branch management
  • • Selective staging only
  • • Lock file protection

📊 Task Management

  • • Automatic task breakdown
  • • Dynamic complexity detection
  • • External memory system
  • • Session continuity
  • • Task duration tracking
  • • Cost tracking per task

🎨 Modern UI

  • • Svelte-powered webview
  • • 4-zone UI architecture
  • • Theme-aware sidebar icon
  • • Dedicated changes tab
  • • Rich omnibox input
  • • Tabbed settings UI

⚙️ Advanced Features

  • • Multimodal input support
  • • Parallel tool execution
  • • Prompt caching
  • • Real-time cost tracking
  • • @ Mention context menu
  • • Conversation history

🛡️ Safety & Control

  • • Granular auto-approval
  • • Docked action bar
  • • Human-in-the-loop safety
  • • Auto-approve internal tools
  • • Abort capability
  • • Input validation

🖥️ Local LLM Support

  • • Zero-friction Ollama onboarding
  • • Hardware detection & profiling
  • • One-click model setup
  • • Qwen 2.5 Coder variants
  • • Progress tracking
  • • macOS, Linux, Windows support

🔒 100% Private (BYOK)

We don't proxy your data. Your API keys stay encrypted on your local machine. Your code never leaves your device. Complete privacy and control.

File System Operations

Precision Editing

  • Search-and-Replace Patches: Targeted edits without rewriting entire files
  • Multi-File Refactoring: Safely refactor across multiple files with atomic operations
  • Syntax-Aware Editing: Understands code structure for intelligent modifications
  • Rollback Support: Undo changes if things go wrong

File Management

  • Read/Write/Create: Full file management capabilities
  • Smart File Discovery: Find files by name or pattern with glob support
  • Codebase Search: Regex-powered search across your entire project
  • Binary File Support: Handle images and other binary files

Local LLM with Ollama

Run powerful AI models locally with zero cloud costs. Mydaus Code automatically profiles your hardware and recommends the optimal model for your system.

32GB+ RAM/VRAM

qwen2.5-coder:32b

Maximum intelligence for high-end systems

16GB+ RAM/VRAM

qwen2.5-coder:14b

Best balance of performance and resources

8GB+ RAM/VRAM

qwen2.5-coder:7b

Optimized for resource-constrained systems