Getting Started

Prerequisites

• MCP-compatible AI assistant (Claude, VS Code with GitHub Copilot, Cursor, etc.)
• A terminal with shell access
• Optionally: API credentials for authenticated services (UMLS, LOINC, ICD-11)

Download

Download pre-built binaries for your platform from the Downloads page.

Available platforms:

• macOS: Apple Silicon (arm64) and Intel (amd64)
• Linux: amd64 and arm64
• Windows: amd64

Source code is available on request.

Configure Your AI Client

Add servers to your AI client’s MCP configuration. See Configuration for client-specific setup.

Example for Claude Desktop:

{
  "mcpServers": {
    "snomed": {
      "command": "/path/to/bin/snomed-mcp"
    },
    "rxnorm": {
      "command": "/path/to/bin/rxnorm-mcp"
    },
    "icd10": {
      "command": "/path/to/bin/icd10-mcp"
    }
  }
}

Replace /path/to/bin/ with the actual path to your downloaded or built binaries.

Verify Installation

Restart your AI client after configuration. Ask:

“Search SNOMED CT for diabetes mellitus”

If configured correctly, your AI assistant will use the snomed_search tool to find matching concepts.

Available Servers

See Configuration for details on setting up authenticated services.

Next Steps

• Configuration Reference — Environment variables and options
• Use Cases — Practical examples and workflows
• Integrations — Detailed documentation for each terminology