AI Assistant

Overview

The AI Assistant helps with coding, debugging, and project questions. It can read your project files, generate code, and answer questions.

Two Types of Assistant

In-Project Assistant

Available when working inside a project:

Capabilities

• Read project files
• Generate code
• Modify files
• Run terminal commands
• Answer project-specific questions
• Debug issues
• Refactor code
• Explain code

How to Use

1. Open the Assistant panel (right side)
2. Type your question or request
3. The AI analyzes your project and responds
4. Review suggestions before applying

Example Requests

• "Add a login form to the auth page"
• "Fix the error in utils.ts"
• "Explain how the authentication works"
• "Refactor this component to use hooks"
• "Add error handling to the API call"

General Assistant

Available outside projects:

Capabilities

• General coding help
• Technology questions
• Best practices
• Learning resources
• Cannot access project files

Use Cases

• "How do I use React hooks?"
• "What's the difference between REST and GraphQL?"
• "Best practices for error handling in Node.js"

AI Features Explained

Code Generation

• Create functions, components, and features
• Follows your project's patterns
• Includes comments and documentation
• Generates tests when appropriate

Code Review

• Analyze code for issues
• Suggest improvements
• Identify bugs
• Performance recommendations
• Security concerns

Debugging

• Analyze error messages
• Trace issues through code
• Suggest fixes
• Explain why errors occur

Code Explanations

• Explain how code works
• Break down complex logic
• Document functions
• Clarify algorithms

Refactoring

• Improve code structure
• Extract reusable components
• Optimize performance
• Modernize code patterns

Model Selection

Choose from different AI models:

Available Models

• GPT-4 (OpenAI): Versatile, good general performance
• GPT-3.5 (OpenAI): Faster, good for simple tasks
• Claude (Anthropic): Strong reasoning and long context
• Gemini (Google): Good for certain use cases

Switching Models

1. Click the model selector in the Assistant panel
2. Choose a different model
3. Conversations continue with the new model

Model Recommendations

• Complex logic: GPT-4
• Quick questions: GPT-3.5
• Code generation: GPT-4
• Simple explanations: GPT-3.5

Conversation Management

Conversation History

• All conversations are saved
• Access previous conversations
• Continue where you left off
• Conversations organized by project

Starting New Conversations

• Click "New Conversation" in the Assistant panel
• Previous conversations remain accessible
• Each conversation is independent

Managing Conversations

• View all conversations in the sidebar
• Search conversations
• Delete conversations you no longer need
• Conversations are stored per project