Lab Killer AI - Local Ollama Assistant

A powerful VS Code extension that brings automated build planning and code generation to your local development environment using Ollama.

Features

• 🤖 Automated Build Plans: AI-generated phase-by-phase development plans
• ⚡ Local Ollama Integration: Run completely offline with your own models
• 🎯 Acceptance Criteria Validation: Automatic verification of generated code
• 🔄 Multi-Model Support: Use different models for planning and coding
• 📊 Visual Progress Tracking: Beautiful glassmorphism UI with real-time updates
• 🏗️ Project Memory: Context-aware development with importance scoring

Requirements

• Ollama: Must be installed and running locally
  • Download from: https://ollama.ai
  • Default URL: http://localhost:11434
• Recommended Models:
  • qwen2.5-coder:7b - Main coding model
  • llama3.1:8b - Build plan orchestrator

Installation

1. Install Ollama from https://ollama.ai
2. Pull recommended models:

   ollama pull qwen2.5-coder:7b
   ollama pull llama3.1:8b
   

3. Install this extension from the VS Code marketplace
4. Open the Lab Killer AI sidebar (icon in activity bar)

Usage

Quick Start

1. Click the Lab Killer AI icon in the sidebar
2. Select your models from the configuration panel
3. Enter your task in the chat box (e.g., "Create a Python web scraper")
4. Press Enter or click GENERATE BUILD PLAN
5. Watch as AI creates a detailed build plan with phases and tasks

Configuration

Access settings via File > Preferences > Settings and search for "Lab Killer":

• Main Model: Model used for code generation
• Orchestrator Model: Model used for build planning
• Ollama URL: Custom Ollama server URL
• Auto Execute: Automatically run build plans after generation
• Audit Timeout: Maximum time for validation checks

Keyboard Shortcuts

• Enter - Send message / Generate plan
• Shift+Enter - New line in chat input
• Ctrl+Enter / Cmd+Enter - Force generate plan

Architecture

lab-killer-ai/
├── extension.js           # Main extension entry point
├── core/                  # Core logic
│   ├── orchestrator.js    # Build plan execution
│   ├── ollama-client.js   # Ollama API integration
│   ├── agent-registry.js  # AI agent personas
│   └── memory-vault.js    # Project context management
├── schema/                # Data structures
│   ├── build-report.js    # Build plan schema
│   └── tool-spec.js       # Tool definitions
├── tools/                 # Execution tools
│   ├── terminal.js        # Command execution
│   ├── filesystem.js      # File operations
│   └── validator.js       # Acceptance criteria checker
├── metadata/              # Intelligence layer
│   ├── tags.js            # Auto-tagging system
│   └── prompts.js         # System prompts
└── media/                 # UI components
    ├── shell.html         # Main HTML frame
    ├── config-view.html   # Model configuration
    ├── chat-view.html     # Chat interface
    ├── build-view.html    # Build plan view
    ├── styles.css         # Glassmorphism styles
    └── ui-logic.js        # Frontend logic

Build Plan Structure

Build plans consist of Phases → Tasks → Acceptance Criteria:

{
  "projectName": "My Project",
  "phases": [
    {
      "name": "Setup & Infrastructure",
      "tasks": [
        {
          "taskId": "task-1",
          "description": "Create main.py file",
          "workflow": "python",
          "files": ["main.py"],
          "acceptanceCriteria": {
            "type": "file_check",
            "description": "File exists and contains imports"
          }
        }
      ]
    }
  ]
}

Troubleshooting

Ollama Not Connecting

1. Verify Ollama is running: ollama list
2. Check URL in settings (default: http://localhost:11434)
3. Restart VS Code

No Models Available

1. Pull models: ollama pull qwen2.5-coder:7b
2. Click refresh icon in footer
3. Verify models: ollama list

Build Plan Not Generating

1. Ensure a model is selected in configuration
2. Check Ollama server logs
3. Try a simpler task first
4. Verify model supports JSON mode

Supported Workflows

• ✅ Python
• ✅ JavaScript/Node.js
• ✅ TypeScript
• ✅ PineScript (TradingView)
• ✅ C++
• ✅ HTML/CSS
• ✅ React/Vue

Contributing

Contributions welcome! Please submit issues and PRs to the GitHub repository.

License

MIT License - see LICENSE file for details

Credits

Built with:

• VS Code Extension API
• Ollama Local AI Platform
• VS Code Webview UI Toolkit

Note: This extension requires a working Ollama installation. All AI processing happens locally on your machine.