My Code Assistant

🤖 An intelligent VS Code extension that combines powerful code exploration commands with AI-powered analysis, all in a unified chat interface.

Features

🎯 Unified Chat Interface

• Single conversation view for both commands and AI chat
• Smart routing - automatically detects commands vs questions
• Seamless workflow - run commands and get AI insights in one place

📋 File Exploration Commands

Explore your codebase with intuitive @ commands:

• @project - Display project structure and directory tree (3 levels deep)
• @directory <path> - Show directory tree for a specific path
• @file <path> - Display file content with metadata (size, lines, type)

🎯 Code Action Commands (AI-Powered)

Transform your code with intelligent AI analysis:

• @explain <path> - Get detailed, comprehensive code explanation
• @refactor <path> - Receive refactoring suggestions with improved code
• @test <path> - Generate complete unit test suites
• @document <path> - Create professional JSDoc/TSDoc documentation
• @debug <path> - Analyze code for bugs and get fixes
• @optimize <path> - Get performance optimization recommendations

🤖 AI-Powered Analysis

• Automatic AI analysis of command results when a model is selected
• Context-aware conversations using command output
• Intelligent explanations of your code structure and files
• Powered by Ollama for local, private AI processing

✨ Smart Features

• Autocomplete for file and directory paths (type @file or @directory )
• Keyboard shortcuts - Ctrl+Enter to send, arrow keys to navigate
• Real-time streaming responses from AI
• Markdown rendering for formatted AI responses
• State persistence - your conversation history is saved

Requirements

• Ollama must be installed and running locally
  • Download from: https://ollama.ai
  • Default URL: http://localhost:11434
  • Pull at least one model: ollama pull llama2 or ollama pull codellama

Installation

1. Install the extension from VS Code Marketplace (or install from VSIX)
2. Install and start Ollama
3. Pull an AI model: ollama pull llama2
4. Open the extension from the Activity Bar (🤖 icon)

Usage

Basic Workflow

1. Open the extension - Click the 🤖 icon in the Activity Bar
2. Select an AI model - Choose from the dropdown (or use "No AI model" for commands only)
3. Start exploring:
   • Type @project to see your project structure
   • Type @file package.json to view a file
   • Ask questions: "What does this project do?"

Example Conversations

Exploring a Project

You: @project

📋 System: [Shows project directory tree]

🤖 Assistant: This is a VS Code extension project with the following structure:
- src/ contains TypeScript source code
- Main files: extension.ts (entry point) and webview.ts (UI)
- Uses TypeScript 5.9.3 and targets VS Code 1.103.0
...

Analyzing a File

You: @file package.json

📋 System: [Shows file content and metadata]

🤖 Assistant: This is a Node.js package configuration for a VS Code extension.

Key dependencies:
• marked 17.0.1 - For markdown rendering
• TypeScript 5.9.3 - Development language
...

Using Code Actions

You: @explain src/extension.ts

📋 System: [Shows file with Action: EXPLAIN]

🤖 Assistant: 
**Overview**: This is the main extension file that activates the My Code Assistant.

**Key Components**:
1. PathValidator - Secures file system access
2. RateLimiter - Prevents API overload
3. HttpClient - Handles Ollama communication
4. MyCodeAssistantProvider - Manages the webview

**Logic Flow**: When activated, it registers the webview provider...

You: @test src/utils/validator.ts

📋 System: [Shows file with Action: TEST]

🤖 Assistant:
**Test Strategy**: Unit tests for PathValidator class

**Test Code**:
\`\`\`typescript
import { PathValidator } from './validator';

describe('PathValidator', () => {
  test('should validate paths within workspace', () => {
    const validator = new PathValidator('/workspace');
    const result = validator.validatePath('src/file.ts');
    expect(result.valid).toBe(true);
  });
  
  test('should reject path traversal', () => {
    const validator = new PathValidator('/workspace');
    const result = validator.validatePath('../../../etc/passwd');
    expect(result.valid).toBe(false);
  });
});
\`\`\`

Direct Questions

You: How does the extension work?

🤖 Assistant: Based on the code context, this extension works by...

Commands Reference

File Exploration

Command	Description	Example
@project	Show project structure	@project
@directory <path>	Show directory tree	@directory src
@file <path>	Show file content	@file package.json

Code Actions (AI-Powered)

Command	Description	Example
@explain <path>	Explain code in detail	@explain src/extension.ts
@refactor <path>	Get refactoring suggestions	@refactor src/utils.ts
@test <path>	Generate unit tests	@test src/api/client.ts
@document <path>	Generate documentation	@document src/helper.ts
@debug <path>	Find and fix bugs	@debug src/service.ts
@optimize <path>	Optimize performance	@optimize src/parser.ts

Tip: Use relative paths from workspace root. Code action commands require an AI model to be selected.

Keyboard Shortcuts

• Ctrl+Enter - Send message
• Arrow Up/Down - Navigate autocomplete suggestions
• Enter - Select autocomplete item
• Escape - Close autocomplete dropdown

Extension Settings

This extension contributes the following settings:

• myCodeAssistant.ollamaUrl - Ollama server URL (default: http://localhost:11434)
• myCodeAssistant.selectedModel - Last selected AI model

Access settings via:

• Command Palette: My Code Assistant: Configure Ollama URL
• Or click the gear icon in the extension panel

How It Works

Unified Interface

The extension features a single chat interface that intelligently handles both commands and AI conversations:

1. Type a command (@project, @file, @directory) → Executes and shows result
2. AI model selected? → Automatically analyzes the result
3. Type a question → Sends to AI with command context
4. No AI model? → Commands still work, helpful message for questions

AI Auto-Analysis

When you run a command with an AI model selected:

1. Command executes and shows result
2. Result is automatically sent to AI
3. AI analyzes and explains the output
4. All in one seamless conversation

Security

• Path validation prevents directory traversal attacks
• Workspace boundaries enforced for all file operations
• Rate limiting (10 requests/second) prevents API abuse
• Content Security Policy in webview
• Local AI processing - your code never leaves your machine

Architecture

Components

• Extension Host (extension.ts) - Main logic, command handling, Ollama integration
• Webview (webview.ts) - Chat UI, message routing, state management
• PathValidator - Security layer for file system access
• HttpClient - Ollama API communication with rate limiting
• RateLimiter - Request throttling

Message Flow

User Input → Smart Detection → Command or AI?
    ↓                              ↓
Command Execution            AI Processing
    ↓                              ↓
Result Display          ←    AI Response
    ↓
Auto AI Analysis (if model selected)

Troubleshooting

Ollama Connection Issues

• Error: "Connection failed"
  • Ensure Ollama is running: ollama serve
  • Check URL in settings (default: http://localhost:11434)
  • Click "🔄 Refresh" to retry connection

No Models Available

• Pull a model: ollama pull llama2
• Verify with: ollama list
• Restart Ollama if needed

Commands Not Working

• Ensure you're in a workspace (open a folder)
• Check file/directory paths are relative to workspace root
• Use autocomplete to avoid typos

Autocomplete Not Showing

• Type @file or @directory (with space after)
• Wait ~300ms for suggestions to appear
• Ensure you're typing in the input field

Development

Building from Source

# Install dependencies
npm install

# Compile TypeScript
npm run compile

# Watch mode for development
npm run watch

# Run tests
npm test

Project Structure

my-code-assistant/
├── src/
│   ├── extension.ts      # Main extension logic
│   ├── webview.ts        # Chat UI implementation
│   └── extention.test.ts # Test suite
├── out/                  # Compiled JavaScript
├── media/                # Static assets
├── package.json          # Extension manifest
└── tsconfig.json         # TypeScript config

Known Issues

• Large files (>1MB) are rejected for performance
• Directory trees limited to 3 levels deep
• Autocomplete limited to 50 results
• Requires Ollama to be running locally

Roadmap

• Multi-file context support
• Custom AI prompts
• Export conversations
• Code snippet extraction
• Syntax highlighting in results
• Remote Ollama support
• Command history with arrow keys
• Slash commands (/clear, /help)

Release Notes

0.0.1 (2026-01-25)

Initial Release 🎉

Features:

• ✅ Unified chat interface for commands and AI
• ✅ Three powerful commands: @project, @file, @directory
• ✅ AI auto-analysis of command results
• ✅ Autocomplete for file/directory paths
• ✅ Ollama integration for local AI
• ✅ Security features (path validation, rate limiting)
• ✅ State persistence
• ✅ Markdown rendering
• ✅ Real-time streaming responses

Contributing

Contributions are welcome! Please feel free to submit issues or pull requests.

License

MIT License

Credits

• Built with VS Code Extension API
• AI powered by Ollama
• Markdown rendering by marked

---

Enjoy exploring your code with AI assistance! 🚀

For questions or issues, please visit the GitHub repository.