Skip to main content
Glama
MCP Server for Google Cloud Healthcare API

MCP Server for Google Cloud Healthcare API

by Kartha-AI

An MCP Server for Google Cloud Healthcare API (FHIR)

A Model Context Protocol (MCP) server that provides healthcare tools for interacting with FHIR resources on Google Cloud Healthcare API and public medical research APIs like pubmed, using MCP clients like Claude and Goose.

This is a slightly modified version for AgentCare MCP Server for EHRs. https://github.com/Kartha-AI/agentcare-mcp

The maion differnce is that this repo talks to Google Cloud Healthcare FHIR APIs thru a SmartonFHIR gateway that's secured by Firebase Auth.

Architetcure

Related MCP server: Secure Remote MCP Server

Demo

  • Claude: demo/claude-demo.mp4

  • Goose: demo/goose-demo.mp4

Screenshots

Tools

FHIR Tools

  • find_patient - Search for a patient by name, DOB, or other identifiers

  • get_patient_observations - Retrieve patient observations/vital signs

  • get_patient_conditions - Get patient's active conditions

  • get_patient_medications - Get patient's current medications

  • get_patient_encounters - Get patient's clinical encounters

  • get_patient_allergies - Get patient's allergies and intolerances

  • get_patient_procedures - Get patient's procedures

  • get_patient_careteam - Get patient's care team members

  • get_patient_careplans - Get patient's active care plans

  • get_vital_signs - Get patient's vital signs

  • get_lab_results - Get patient's laboratory results

  • get_medications_history - Get patient's medication history

  • clinical_query - Execute custom FHIR queries

Medical Research Tools

  • search-pubmed - Search PubMed articles related to medical conditions

  • search-trials - Find relevant clinical trials

  • drug-interactions - Check drug-drug interactions

Usage

Each tool requires specific parameters:

Required Parameters

  • Most tools require patientId

  • Some tools have additional parameters:

    • lab_trend_analysis: requires labType

    • search-pubmed: requires query and optional maxResults

    • search-trials: requires condition and optional location

    • drug-interactions: requires drugs array

refer to: /src/server/constants/tools.ts for tools specirfication

Use with claude desktop

for claude desktop: 
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
(use the env variables as shown above)

{
  "mcpServers": {
    "google-cloud-healthcare-api-mcp": {
      "command": "node",
      "args": [
        "/Users/your-username/{google-cloud-healthcare-api-mcp dir}/build/index.js"
      ],
      "env": {
          "FIREBASE_API_KEY":"XXXXXXXXX",
          "FIREBASE_AUTH_DOMAIN":"XXXXXXXX",
          "FIREBASE_PROJECT_ID":"XXXXXXX",
          "FIREBASE_STORAGE_BUCKET":"XXXXXXXXX",
          "FIREBASE_MESSAGING_SENDER_ID":"XXXXXXX",
          "FIREBASE_APP_ID":"XXXXXXXXX",
          "FIREBASE_MEASUREMENT_ID":"XXXXXXXX",
          "FIREBASE_AUTH_CALLBACK_PORT":"3456",
          "FHIR_BASE_URL":"{gchapi-fhir-gateway-host}/fhir",
          "PUBMED_API_KEY":"your_pubmed_api_key",
          "CLINICAL_TRIALS_API_KEY":"your_trials_api_key",
          "FDA_API_KEY":"your_fda_api_key"
      }
    }
  }
}

Installing via Smithery

To install google-cloud-healthcare-api-mcp for Claude Desktop automatically via Smithery:

npx -y @smithery/cli install @Kartha-AI/google-cloud-healthcare-api-mcp --client claude

Start MCP Server Locally with MCP Inspector

git clone git@github.com:Kartha-AI/google-cloud-healthcare-api-mcp.git
cd google-cloud-healthcare-api-mcp
npm install
npm run build
npm install -g @modelcontextprotocol/inspector
mcp-inspector  build/index.js
http://localhost:5173
Set up the env vars on Inspector

Troubleshooting:

If Claude desktop is running it uses port 3456 for Auth. You need to terminate that process using the following command:

kill -9 $(lsof -t -i:3456)

Available Tools

15 tools
find_patientC

Search for a patient by demographics

Inputs: lastName (required), birthDate, firstName, gender.

get_appointmentsC

Get patient's Appointments

Inputs: patientId (required), dateFrom, dateTo.

get-drug-infoC

Get Drug details by a generic name

Inputs: genericName (required).

get_lab_resultsC

Get patient's lab results

Inputs: patientId (required), category, timeframe.

get_medications_historyC

Get patient's medication history including changes

Inputs: patientId (required), includeDiscontinued.

get_patient_allergiesC

Get allergies and intolerances for a patient

Inputs: patientId (required), category, status, type.

get_patient_careplansC

Get care plans for a patient

Inputs: patientId (required), category, dateFrom, dateTo, +1 more.

get_patient_conditionsC

Get medical conditions/diagnoses for a patient

Inputs: patientId (required), onsetDate, status.

get_patient_encountersC

Get healthcare encounters/visits for a patient

Inputs: patientId (required), dateFrom, dateTo, status.

get_patient_medicationsC

Get medication orders for a patient

Inputs: patientId (required), status.

get_patient_observationsC

Get observations (vitals, labs) for a patient

Inputs: patientId (required), code, dateFrom, dateTo, +1 more.

get_patient_proceduresC

Get procedures performed on a patient

Inputs: patientId (required), dateFrom, dateTo, status.

get_vital_signsC

Get patient's vital signs history

Inputs: patientId (required), timeframe.

search-pubmedC

Search PubMed for medical literature

Inputs: query (required), maxResults.

search-trialsC

Search ClinicalTrials.gov for relevant studies

Inputs: condition (required), location.

Tool Schema Changelog

Recent tool additions, removals, and schema changes observed during successful MCP inspections.

Loading tool schema changelog…

    TDQS

    B3.2/5.0

    Scored across 15 tools

    Disambiguation5/5

    Each tool has a clearly distinct purpose targeting specific healthcare data types (e.g., appointments, lab results, medications, allergies). There is no overlap in functionality, and an agent can easily differentiate between tools based on the resource they retrieve.

    Naming Consistency4/5

    Most tools follow a consistent verb_noun pattern (e.g., get_patient_allergies, get_patient_conditions), but there are minor deviations with hyphenated names (get-drug-info, search-pubmed, search-trials) and one tool using 'find' instead of 'get' (find_patient). Overall, the naming is readable and mostly predictable.

    Tool Count5/5

    With 15 tools, the server is well-scoped for a healthcare API, covering a comprehensive range of patient data and external medical resources. Each tool serves a specific and necessary function without redundancy, making the count appropriate for the domain.

    Completeness4/5

    The tool set provides extensive read-only coverage for patient data (e.g., conditions, medications, encounters) and includes useful external searches (PubMed, clinical trials). Minor gaps exist, such as lack of write operations (e.g., create or update patient records), but the surface supports most agent workflows for data retrieval and research.

    Related MCP Connectors

    Related MCP Servers