Advanced AI-powered file assistant with support for multiple AI providers, error handling, memory system, and terminal command execution.
- Model: llama2-uncensored:7b
- URL: http://localhost:11434/api/chat
- Requirements: Ollama must be running locally
- Model: gemini-2.5-pro
- Requirements: Google AI API key
- Get your key: https://ai.google.dev/
- Model: mistral-tiny
- Requirements: Mistral API key
- Get your key: https://console.mistral.ai/
- Model: gpt-4
- Requirements: OpenAI API key
- Get your key: https://platform.openai.com/api-keys
- Model: claude-3-5-sonnet-20241022
- Requirements: Anthropic API key
- Get your key: https://console.anthropic.com/
✅ Fixed Directory Creation Logic - Files like .html
are now created as files, not directories
✅ API Key Configuration - Secure API key input for cloud providers
✅ Multiple AI Providers - Support for 5 different AI services
✅ Error Retry Logic - Automatic retry with different approaches
✅ Auto Error Checking - Automatic syntax validation after file operations
✅ Terminal Command Execution - AI-powered terminal command generation
✅ Memory System - Comprehensive logging and history tracking
✅ File Operations - Create, edit, rewrite, delete, review files
✅ AI Code Review - Intelligent code analysis and improvements
node index.js
- Select your AI provider
- Enter API key if required (masked input)
- Choose interaction mode:
- Agent Mode: AI plans and executes complex tasks
- Chat Mode: Conversational interaction
- Command Mode: Direct command execution
- API keys are entered securely (masked input)
- Keys are stored only in memory during session
- No API keys are saved to disk
- Each provider is configured independently
# Create files (now works correctly!)
"create calculator.html" # ✅ Creates HTML file (not directory)
"make style.css" # ✅ Creates CSS file
"create assets/" # ✅ Creates directory
# AI-powered operations
"review and improve calculator.html"
"check calculator.html for errors"
"run the project"
"build the application"