Overview
What is 🧪 Foundry MCP Playground?
🧪 Foundry MCP Playground is a streamlined development environment that provides an MCP server for interacting with Azure AI Foundry. It enables exploration and use of AI models from the Azure AI Foundry Catalog and Azure AI Foundry Labs, management and querying of knowledge via Azure AI Search, and evaluation of text and agent responses using observability features—all without the need to manually set up these resources. It is designed for developers who want to rapidly prototype AI-based solutions inside GitHub Copilot with a one-click setup.
How to use 🧪 Foundry MCP Playground?
Create your own repository from the template and open it in GitHub Codespaces or VS Code with devcontainer support (Docker required locally). In VS Code, enable Agent Mode in settings, then open the Chat view, switch to Agent mode, and Copilot automatically discovers the MCP server. Click “New tools available” to load the tools, then start asking natural-language prompts about Azure AI Foundry models, knowledge, or evaluations.
Key features of 🧪 Foundry MCP Playground
- One-click setup via GitHub template and Codespaces
- Automatically starts MCP server for Azure AI Foundry
- Exposes tools to explore, compare, and use AI models
- Manage and query knowledge with Azure AI Search
- Evaluate text and agent responses (observability)
- No extra configuration required inside Copilot
Use cases of 🧪 Foundry MCP Playground
- Explore and compare AI models from Azure AI Foundry Catalog
- Build prototype applications that use models with a GitHub token
- Deploy and manage OpenAI models on Azure AI Foundry
- Create a model comparison app with evaluation capabilities
- Use Azure AI Foundry Labs features like Magentic One or Omniparser
FAQ from 🧪 Foundry MCP Playground
How do I start using the playground?
Create a repository from the template, open it in Codespaces or VS Code with devcontainer support, enable Agent Mode in VS Code settings, then switch the Chat view to Agent mode. Copilot discovers the MCP server automatically; click “New tools available” to load the tools.
What tools are available in GitHub Copilot?
Copilot gains tools to interact with Azure AI Foundry: explore models, manage and query knowledge using Azure AI Search, and evaluate text and agent responses—all with no extra configuration.
Do I need to configure anything?
No. The MCP server starts automatically inside the devcontainer, and Copilot discovers it via stdio. You only need to enable Agent Mode in VS Code settings once.
What are the system requirements?
For local VS Code devcontainer usage you need a Docker engine running. Using GitHub Codespaces requires no local setup beyond a GitHub account.
Is the MCP server production-ready?
The MCP Server for Azure AI Foundry is marked
