To try out this new exciting features I just updated the labs on the AI Gateway repo:
1๏ธโฃ New lab ๐งช mcp-from-api.
The lab uses the Basic v2 SKU and BICEP ๐ฆพ to create MCP servers from existing REST API's and MCP pass-trough:
- Weather MCP server based on an existing API (the hello world for MCP)
- Microsoft Learn MCP pass-trough for the Learn MCP server (https://learn.microsoft.com/api/mcp)
- Product Catalog MCP based on an existing API (mock) - protected with OAuth Authorizations
- Place Order MCP based on a Logic Apps workflow - protected with OAuth Authorizations
๐ To test MCPs, the lab features Python snippets with the MCP SDK and showcases Agent implementations using both the OpenAI Agents and Azure AI Foundry Agents SDKs! The inference APIs, the backend APIs and MCP servers are all served trough the AI Gateway, enabling the
๐ The lab includes an Azure Dashboard and an Azure Monitor Workbook with specific MCP Insights.
โจ The lab also uses Azure API Center as the enterprise MCP registry, loading all APIs MCP's and third-party MCPs for easy discovery and seamless onboarding to MCP clients like VS Code.
-------
2๏ธโฃ To explore the new MCP Dynamic Client Registration using OAuth, I have updated the lab ๐งช mcp-client-authorization.
๐ This is an experimental feature (not ready for production) that could be easily tested with the MCP inspector tool.
------
3๏ธโฃ I have also updated the lab ๐งช realtime-mcp-agents to try out the APIM credential manager for authorizing backend APIs and MCP servers!
๐ This lab uses the real-time API and Spotify MCP tool to create exciting audio conversations, allowing you to easily interact with your playlists and control playback!
------
๐ We're thrilled to announce that our backlog now includes even more labs and samples! Share your thoughts and let us know what features or scenarios you'd like to see next!