azure-ai-vision-face-api

This MCP server provides tools to perform Azure AI Vision Face API tasks within Agentic AI workflows. It exposes capabilities for face attribute detection, face recognition and enrollment, and image management integration via an MCP bridge. The included HTTP bridge (in bridge/) can be used to expose a REST API for listing available MCP tools and calling them from external clients, such as OpenAI responses or frontend apps. Typical use cases include detecting facial attributes (age, gender, presence of glasses or masks), enrolling faces into a person group, listing or identifying faces against large person groups, and performing image-based face comparisons within an automated workflow. Start the MCP server using UV, ensure the required environment variables are set, and then interact with the available MCP tools either directly via the UV runtime or through the HTTP bridge endpoints (GET /health, GET /mcp/tools, POST /mcp/call).

Prerequisites:

• A working install of UV (the Universal Verifier/Orchestrator) per the project's installation guide.
• Access to Azure Face API (resource endpoint URL and key).
• Optional: Azure OpenAI (for open-set attribute detection) and Azure Storage (for image management).

Install steps:

1. Install UV following the official guide:
   • Refer to the UV installation instructions: https://docs.astral.sh/uv/getting-started/installation/
2. Clone the MCP server repository:

   git clone https://github.com/Azure-Samples/azure-ai-vision-face-api-mcp-server.git
   cd azure-ai-vision-face-api-mcp-server
   

3. Prepare environment variables (fill in with your real credentials):
   • Azure AI Face API Endpoint
   • Azure AI Face API Key
   • (Optional) Azure OpenAI Endpoint
   • (Optional) Azure OpenAI API Key
   • (Optional) AZURE STORAGE ACCOUNT
   • (Optional) AZURE STORAGE CONTAINER
   • (Optional) AZURE STORAGE SAS TOKEN
4. Run the MCP server with UV:

   uvx --from azure-ai-vision-face-api-mcp-server
   

5. If you use the HTTP bridge, install and run the bridge locally:

   cd bridge
   npm install
   cp .env.example .env   # fill in your environment values (see step 3 above)
   npm run dev
   

6. Validate the server is running by hitting bridge endpoints, for example:
   • GET http://127.0.0.1:8787/mcp/tools
   • POST http://127.0.0.1:8787/mcp/call with a tool name and arguments in JSON

Prerequisites note: If you plan to test locally, copy .env.example to .env and populate keys that are shared across MCP tools.

Tips and notes:

• The MCP server relies on Azure Face API credentials; ensure your endpoint and key are correct and have the required permissions.
• When enabling open-set attribute detection via Azure OpenAI, keep in mind this requires an active Azure OpenAI deployment and associated credits.
• For image management, you’ll need a configured Azure Storage account with a container and a valid SAS token; renew tokens as needed.
• The HTTP bridge provided with this repo enables easier integration with web frontends or the OpenAI API responses; use the /mcp/tools endpoint to discover available tools and /mcp/call to execute them.
• If you encounter environment variable prompts on first run, ensure UV is started with the appropriate env settings loaded (e.g., via a .env file or shell exports).
• The examples folder contains sample prompts and data to help you validate behavior before integrating into production workflows.