AI Playground

The AI Playground is your interactive development environment for experimenting with cutting-edge AI models and building intelligent agents. It's a creative sandbox where you can blend powerful language models with real-world tools, creating AI experiences that go far beyond simple chat.

Whether you're a developer, researcher, or AI enthusiast, the playground provides everything you need to transform ideas into working prototypes quickly and efficiently.

Core Capabilities

The playground combines three essential components: advanced AI models from OpenAI (GPT-4, ChatGPT) and Anthropic (Claude), MCP servers that give AI real-world capabilities like checking weather or executing code, and a local development toolbox that securely connects everything on your machine.

MCP servers are specialized toolkits that transform AI from conversation partners into capable assistants. Instead of just talking about tasks, your AI can actually perform them—checking live data, running programs, or interacting with external services.

Getting Started

Access the playground from the main navigation (desktop browsers only for optimal performance with complex AI interactions).

Configure Your AI Model: Choose between OpenAI (GPT-4, ChatGPT) or Anthropic (Claude) providers. Enter your API key, select a model, and customize settings like temperature (creativity level), token limits (response length), and system prompts (AI personality and behavior).

Install Local Toolbox (recommended): Click install for guided setup that handles everything automatically. The system will help install Docker if needed and configure all components with clear status indicators.

Add MCP Servers: Browse the server directory to find tools for your use cases. Each server description explains its capabilities. Toggle servers on/off with simple switches, then test integration by chatting with your AI to see tools in action.

Key Features

Model Configuration: Switch between OpenAI and Anthropic providers with models ranging from GPT-3.5 to GPT-4 and Claude variants. Temperature controls adjust creativity (0-2 scale), while token limits control response length. System prompts shape AI behavior, and multi-step reasoning enables complex problem-solving workflows.

MCP Server Management: Intuitive server discovery with visual health indicators and real-time connection monitoring. Automatic reconnections and error recovery keep tools running smoothly, with clear feedback about available capabilities.

Local Toolbox: Sophisticated monitoring system with continuous health checks, automatic updates, and intelligent port management. The status dashboard provides system overview, while built-in diagnostics resolve issues quickly.

Chat Experience: Streaming responses show AI thinking in real-time. Code appears with syntax highlighting, tool execution is transparent, and clear error messages help troubleshoot issues. Natural conversation flow maintains context throughout sessions.

Advanced Usage

Custom System Prompts transform general AI models into specialized assistants. Create a "senior software engineer helping debug code" for development work, a "creative writing assistant focused on storytelling" for content creation, or a "data analyst who explains insights clearly" for analytical tasks.

Temperature Settings control AI personality: use 0.0-0.3 for consistent, focused responses ideal for coding and analysis, 0.7-1.0 for balanced general conversation, or 1.5-2.0 for highly creative brainstorming and creative writing.

Multi-step Workflows enable sophisticated problem-solving where AI breaks down complex tasks, uses multiple tools sequentially, iterates on solutions, and learns from previous steps to tackle challenges systematically.

Troubleshooting

Connection Issues

• API Key Problems: Verify your API key is valid and has sufficient credits
• Network Errors: Check internet connection and firewall settings
• MCP Server Failures: Try disabling and re-enabling problematic servers

Local Toolbox Issues

• Docker Not Running: Ensure Docker Desktop is installed and running
• Port Conflicts: The system uses port 11990 - ensure it's available
• Proxy Mismatches: Follow on-screen prompts to resolve ID conflicts

Performance Tips

• Model Selection: Faster models (GPT-3.5, Claude Instant) for quick tasks
• Token Limits: Lower limits for faster responses
• Server Selection: Enable only needed MCP servers to reduce overhead

Requirements

• Desktop Browser: Chrome, Firefox, Safari, Edge (mobile not supported)
• API Keys: Valid OpenAI or Anthropic API key with credits
• Docker (for local toolbox): Docker Desktop installed and running
• Internet Connection: Stable connection for AI model access

Start with basic chat, then gradually add MCP servers and advanced configurations to unlock the playground's full potential as your AI development laboratory.