A powerful fully autonomous AI system inspired by Goose, featuring multi-LLM support, self-improving capabilities, and a professional chat interface. The AI can handle any task, modify its own code, and continuously improve itself.
- Full autonomy: Handles any task from code generation to system improvements
- Self-improving: Analyzes performance and suggests optimizations
- Memory system: Learns from conversations and maintains context
- Task execution: Plans and executes complex multi-step workflows
- OpenAI GPT-4: Advanced reasoning and code generation
- Anthropic Claude: Safety-focused, high-quality responses
- Google Gemini: Fast, cost-effective default provider
- Smart routing: Automatically selects best provider for each task
- Cost optimization: Routes simple tasks to cheaper models
- Modern UI: Beautiful gradient design with smooth animations
- Real-time status: Shows agent learning/active states
- Provider indicators: Displays which LLM is being used
- Responsive design: Works perfectly on mobile and desktop
- Dark mode: Automatic system preference detection
- Code tools: Analyze, generate, and refactor code
- File operations: Read, write, and manage project files
- HTTP requests: Interact with external APIs
- Database integration: Supabase integration for data persistence
- Self-improvement tools: Performance analysis and optimization
- Frontend: React + Vite + Modern CSS with gradients and animations
- AI System: Custom autonomous agent with multi-provider support
- Database: Supabase (Postgres with RLS policies)
- Hosting: GitHub Pages + Cloudflare Workers
- LLM Providers: OpenAI, Anthropic, Google Gemini
git clone https://github.com/josolinap/josolinap.github.io
cd josolinap.github.io
npm installcp .env.example .env
# Edit .env with your API keysnpm run devVisit http://localhost:5173 and start chatting with your autonomous AI!
# Simple tasks
"Analyze the sentiment of this text: 'I love this amazing product!'"
# Complex tasks
"Create a React component for user authentication with form validation"
# Self-improving
"Analyze my recent conversations and suggest ways to improve your responses"
# File operations
"Create a new API endpoint file in the backend directory"
# System tasks
"Update the UI to use a darker color scheme"
# Creative tasks
"Write a short story about a robot learning to paint"The AI autonomously:
- Analyzes requests to understand intent and complexity
- Plans multi-step workflows for complex tasks
- Selects appropriate tools and LLM providers
- Executes actions safely and efficiently
- Learns from outcomes for continuous improvement
- Performance analysis: Reviews conversation quality and success rates
- Strategy optimization: Suggests better approaches for common tasks
- Code improvements: Can propose changes to its own codebase
- Memory management: Optimizes how information is stored and retrieved
- Prompt engineering: Refines its own system prompts automatically
- Task-specific routing: Uses GPT-4 for complex reasoning, Gemini for speed
- Cost optimization: Automatically selects cheapest viable model
- Fallback handling: Gracefully handles API failures and switches providers
- Quality monitoring: Tracks response quality per provider
- Provider switching: Manual provider selection via commands
# Primary LLM Providers (choose any combination)
VITE_OPENAI_API_KEY=sk-your-openai-key
VITE_ANTHROPIC_API_KEY=sk-ant-your-anthropic-key
VITE_GEMINI_API_KEY=your-gemini-key
# Model Preferences
VITE_OPENAI_MODEL=gpt-4-turbo-preview # or gpt-4, gpt-3.5-turbo
VITE_ANTHROPIC_MODEL=claude-3-opus-20240229 # or claude-3-sonnet, claude-3-haiku
VITE_GEMINI_MODEL=gemini-1.5-flash # or gemini-1.5-pro
# Production Workers
VITE_WORKER_URL=https://ai-notes-proxy.nosleep.workers.dev
# Supabase (for data persistence)
VITE_SUPABASE_URL=https://your-project.supabase.co
VITE_SUPABASE_ANON_KEY=your-anon-key├── 👤 Autonomous Agent
│ ├── 🧠 Memory System
│ ├── 🛠️ Tool Registry
│ └── 📝 Conversation Management
│
├── 🌐 Multi-LLM Providers
│ ├── 🤖 OpenAI (GPT-4)
│ ├── 👥 Anthropic (Claude)
│ └── 🔍 Google (Gemini)
│
├── 🎨 Professional UI
│ ├── 💬 Chat Interface
│ ├── 🎯 Status Indicators
│ └── 📊 Provider Metrics
│
└── 🛠️ Tool System
├── 📁 File Operations
├── 💻 Code Tools
├── 🌐 HTTP Requests
└── 🗄️ Database Access
npm run build
npm run deploy # Deploys to GitHub Pagesnpm run worker:create # Creates enhanced worker
# Follow deployment instructions in cloudflare-worker/README.md- Development: Direct API calls (CORS limited)
- Production: Cloudflare Workers for API proxying
- Performance: Workers provide better speed and reliability
This system draws inspiration from Goose but adds:
- Full autonomy: No external orchestration required
- Self-improvement: Automatic learning and optimization
- Professional UI: Polished, production-ready interface
- Multi-provider: Intelligent LLM selection and failover
- Tool ecosystem: Comprehensive utility functions
- ✅ API Key Protection: Keys never sent to client-side
- ✅ Row Level Security: Supabase policies prevent unauthorized access
- ✅ Input Validation: All agent inputs sanitized and validated
- ✅ Safe Execution: Tools designed to prevent harmful operations
- ✅ Memory Limits: Configurable memory and conversation size limits
Your autonomous AI assistant is ready to handle any task you throw at it! The system will continuously learn and improve from your interactions, becoming more capable over time.
Try asking it to:
- Build a new feature for your project
- Analyze and improve existing code
- Research topics or implement complex algorithms
- Set up development environments
- Debug issues or monitor performance
- Anything else you can imagine!
The future of AI-assisted development is here. 🚀