azure-fhir-mcp-server
Azure AHDS FHIR MCP Server
Documentation
Azure AHDS FHIR MCP Server 🚀
A Model Context Protocol (MCP) server implementation for Azure Health Data Services FHIR (Fast Healthcare Interoperability Resources). This service provides a standardized interface for interacting with Azure FHIR servers, enabling healthcare data operations through MCP tools.
Setup 🛠️
Installation 📦
Requires Python 3.13 or higher and `uv`.
Install uv first.
Configuration ⚙️
See the FastMCP guidance on `mcp.json` here: https://gofastmcp.com/integrations/mcp-json-configuration
Client Credentials Flow (default):
- Used for service-to-service authentication
- Leave `USE_FAST_MCP_OAUTH_PROXY=false`
- Keep `HTTP_TRANSPORT=false` to use stdio transport
- Uses Azure AD client credentials flow
{
"mcpServers": {
"fhir": {
"type": "stdio",
"command": "uvx",
"args": [
"azure-fhir-mcp-server"
],
"env": {
"fhirUrl": "https://your-fhir-server.azurehealthcareapis.com/fhir",
"clientId": "your-client-id",
"clientSecret": "your-client-secret",
"tenantId": "your-tenant-id"
}
}
}
}OAuth On-Behalf-Of Flow:
Create the Azure App Registration
The OAuth on-behalf-of flow requires a confidential Azure AD application that represents the MCP server.
1. In the Azure portal, go to Microsoft Entra ID ➜ App registrations ➜ New registration. Give it a descriptive name such as `FHIR-MCP-Server`, set Supported account types to *Single tenant*, and leave the redirect URI unset for now.
2. After the app is created, capture the generated `Application (client) ID` and `Directory (tenant) ID` for later use.
3. Under Expose an API, select Set for the Application ID URI and accept the suggested value `api://{appId}`. Add a scope named `user_impersonation` with admin consent display/description also set to `user_impersonation`.
4. Under Certificates & secrets, create a New client secret (for example `FHIR-MCP-Secret-New`). Copy the secret value immediately; it is required for the MCP server `clientSecret` setting.
5. Under Authentication, add the following Web redirect URIs to support the FastMCP OAuth proxy:
Ensure Default client type remains *No* so the app stays confidential.
6. Under API permissions, choose Add a permission ➜ APIs my organization uses, search for your Azure Health Data Services FHIR server, and add the delegated scopes required for your scenario. Grant admin consent so the FastMCP proxy can request tokens without an interactive prompt.
- Environment variables:
- Start the MCP server with:
uv pip install -e .
uv run --env-file .env azure-fhir-mcp-server- Update mcp.json:
{
"mcpServers": {
"fhir": {
"type": "http",
"url": "http://localhost:9002/mcp"
}
}
}The following is a table of available environment configuration variables:
| Variable | Description | Default | Required |
|---|---|---|---|
| `fhirUrl` | Azure FHIR server base URL (include `/fhir`) | - | Yes |
| `clientId` | Azure App registration client ID | - | Yes |
| `clientSecret` | Azure App registration client secret | - | Yes |
| `tenantId` | Azure AD tenant ID | - | Yes |
| `USE_FAST_MCP_OAUTH_PROXY` | Enable FastMCP Azure OAuth proxy integration | `false` | No |
| `HTTP_TRANSPORT` | Run the MCP server over HTTP transport (required for OAuth proxy) | `false` | No |
| `FASTMCP_HTTP_PORT` | Port exposed when `HTTP_TRANSPORT=true` | `9002` | No |
| `FHIR_SCOPE` | Override FHIR audience scope for the OBO flow (space-separated) | `{fhirUrl}/.default` | No |
| `FASTMCP_SERVER_AUTH_AZURE_BASE_URL` | Public base URL of your FastMCP server | `http://localhost:9002` | No |
| `FASTMCP_SERVER_AUTH_AZURE_REDIRECT_PATH` | OAuth callback path appended to the base URL | `/auth/callback` | No |
| `FASTMCP_SERVER_AUTH_AZURE_IDENTIFIER_URI` | Azure App registration Application ID URI | `api://{clientId}` | No |
| `FASTMCP_SERVER_AUTH_AZURE_REQUIRED_SCOPES` | Space-separated scopes requested by the Azure provider | `user_impersonation` | No |
| `FASTMCP_SERVER_AUTH_AZURE_ADDITIONAL_AUTHORIZE_SCOPES` | Optional space-separated scopes added to the authorize request | - | No |
| `LOG_LEVEL` | Logging level | `INFO` | No |
Available Tools 🔧
FHIR Resource Operations
- `search_fhir` - Search for FHIR resources based on a dictionary of search parameters
- `get_user_info` - (OAuth only) Returns information about the authenticated Azure user
Resource Access
The server provides access to all standard FHIR resources through the MCP resource protocol:
- `fhir://Patient/` - Access all Patient resources
- `fhir://Patient/{id}` - Access a specific Patient resource
- `fhir://Observation/` - Access all Observation resources
- `fhir://Observation/{id}` - Access a specific Observation resource
- `fhir://Medication/` - Access all Medication resources
- `fhir://Medication/{id}` - Access a specific Medication resource
- And many more...
Development 💻
Local Development Setup
1 - Clone the repository:
git clone https://github.com/erikhoward/azure-fhir-mcp-server.git
cd azure-fhir-mcp-server2 - Create and activate virtual environment:
Linux/macOS:
python -m venv .venv
source .venv/bin/activateWindows:
python -m venv .venv
.venv\Scripts\activate3 - Install dependencies:
pip install -e ".[dev]"4 - Copy and configure environment variables:
cp .env.example .envEdit .env with your settings:
fhirUrl=https://your-fhir-server.azurehealthcareapis.com/fhir
clientId=your-client-id
clientSecret=your-client-secret
tenantId=your-tenant-id5 - Claude Desktop Configuration
Open `claude_desktop_config.json` and add the following configuration.
On MacOs, the file is located here: `~/Library/Application Support/Claude Desktop/claude_desktop_config.json`.
On Windows, the file is located here: `%APPDATA%\Claude Desktop\claude_desktop_config.json`.
{
"mcpServers": {
"fhir": {
"command": "uv",
"args": [
"--directory",
"/path/to/azure-fhir-mcp-server/repo",
"run",
"azure_fhir_mcp_server"
],
"env": {
"LOG_LEVEL": "DEBUG",
"fhirUrl": "https://your-fhir-server.azurehealthcareapis.com/fhir",
"clientId": "your-client-id",
"clientSecret": "your-client-secret",
"tenantId": "your-tenant-id"
}
}
}
}6 - Restart Claude Desktop.
Running Tests
# Run all tests
python -m pytest tests/ -v
# Run with coverage
pytest tests/ --cov=src/azure_fhir_mcp_server
# Run specific test
pytest tests/test_fastmcp_metadata.py::TestFastMCPMetadata::test_fastmcp_server_discovery -v
# Run with detailed output
pytest tests/test_fastmcp_metadata.py::TestFastMCPMetadata::test_output_detailed_metadata -v -sContributions 🤝
Contributions are welcome! Please feel free to submit a Pull Request.
1. Fork the repository
2. Create your feature branch (`git checkout -b feature/AmazingFeature`)
3. Commit your changes (`git commit -m '✨ Add some AmazingFeature'`)
4. Push to the branch (`git push origin feature/AmazingFeature`)
5. Open a Pull Request
License ⚖️
Licensed under MIT - see LICENSE.md file.
This is not an official Microsoft or Azure product.
Frequently asked questions
What is azure-fhir-mcp-server?
azure-fhir-mcp-server is Azure AHDS FHIR MCP Server
How do I install azure-fhir-mcp-server?
Open the GitHub repository and follow its README. Most MCP servers are added to your client's MCP config, then called by your agent.
Is azure-fhir-mcp-server open source?
Yes — it is hosted on GitHub at https://github.com/erikhoward/azure-fhir-mcp-server and has 15 stars.
Related MCP tools
Expose your FastAPI endpoints as Model Context Protocol (MCP) tools, with Auth! Python-based implementation. Trusted by 11000+ developers.
A text-based user interface (TUI) client for interacting with MCP servers using Ollama. Features include multi-server, dynamic model switching, streaming res...
基于大模型搭建的聊天机器人,同时支持 微信公众号、企业微信应用、飞书、钉钉 等接入,可选择ChatGPT/Claude/DeepSeek/文心一言/讯飞星火/通义千问/ Gemini/GLM-4/Kimi/LinkAI,能处理文本、语音和图片,访问操作系统和互联网,支持基于自有知识库进行定制企业智能客服。
An LLM agent that conducts deep research (local and web) on any given topic and generates a long report with citations. Built for the Model Context Protocol to
Build effective agents using Model Context Protocol and simple workflow patterns Python-based implementation. Trusted by 7600+ developers.
ACI.dev is the open source tool-calling platform that hooks up 600+ tools into any agentic IDE or custom AI agent through direct function calling or a unifie...
Run your own MCP server? See who uses it and what to fix.
Measure it with TrackMCP