inbody-api-mcp

Hosted version for Claude.ai, ChatGPT, and Grok coming soon. Join the waitlist →

An MCP (Model Context Protocol) server for InBody body-composition data, built on the reverse-engineered mobile REST API used by the InBody Android app.

InBody has no public API and no web UI for personal scan data. This server talks to the same JSON REST endpoints the mobile app uses, exposing your body composition history (body fat, muscle mass, body water, segmental impedance) to any MCP client.

Features

• Profile -- identity and baseline metrics (height, weight, age, gender)

• Scan history -- chronological summaries (weight, BMI, % body fat, muscle mass)

• Full scan metrics -- complete body composition (BCA), BMI/%fat/muscle with normal ranges (MFA), and segmental/multi-frequency impedance (IMP)

• Automatic region routing -- resolves the correct regional API host from your country code

• Automatic re-authentication -- caches the 24h JWT and re-logs in on expiry

Related MCP server: withings-mcp

Quick Start

1. Install uv

curl -LsSf https://astral.sh/uv/install.sh | sh

2. Set credentials

export INBODY_LOGIN_ID="3026323617"   # registration phone number, digits only
export INBODY_LOGIN_PW="your-password"
export INBODY_COUNTRY_CODE="US"       # ISO country code (default US)

INBODY_LOGIN_ID is the phone number used at registration, not your email -- digits only, no country code or + (e.g. 3026323617). InBody keys login on the phone number; the email is only returned as profile data. An email value will fail login with EmptyData.

3. Configure your MCP client

uvx downloads and runs the server on demand -- no separate install step.

OpenCode (opencode.json)

{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "inbody": {
      "type": "local",
      "command": ["uvx", "inbody-api-mcp"],
      "environment": {
        "INBODY_LOGIN_ID": "{env:INBODY_LOGIN_ID}",
        "INBODY_LOGIN_PW": "{env:INBODY_LOGIN_PW}",
        "INBODY_COUNTRY_CODE": "US"
      },
      "enabled": true
    }
  }
}

Claude Desktop (claude_desktop_config.json)

{
  "mcpServers": {
    "inbody": {
      "command": "uvx",
      "args": ["inbody-api-mcp"],
      "env": {
        "INBODY_LOGIN_ID": "3026323617",
        "INBODY_LOGIN_PW": "your-password",
        "INBODY_COUNTRY_CODE": "US"
      }
    }
  }
}

Available Tools

This server is read-only: no write or delete endpoints are exposed.

How It Works

This server communicates with the regional *.lookinbody.com REST API -- the same backend used by the InBody Android app (v2.8.31). The API was reverse-engineered by capturing app traffic with mitmproxy and confirming payload shapes against the live API.

The auth flow:

1. POST /CommonAPI/GetCountryInfoV2 (on appapicommon.lookinbody.com) returns a per-country host table. The Type == "API" row for your ISO country code gives the regional API base (US -> appapiusav2.lookinbody.com) and the numeric phone code used in request bodies.

2. POST /V2/Main/GetLoginWithSyncDataPartV2 exchanges the login ID + password for a 24-hour JWT, a refresh token, and the account UID.

3. Subsequent calls send Authorization: Bearer <JWT>. The client re-authenticates automatically when the token expires.

Each scan record nests three blocks: BCA (body composition analysis -- body water, protein, mineral, fat, segmental water), MFA (BMI, % body fat, skeletal muscle mass, WHR with normal ranges), and IMP (raw impedance per frequency and body segment).

Python API

You can use the client directly:

from inbody_api_mcp.client import InBodyClient

client = InBodyClient()

# Total number of scans
count = client.get_scan_count()

# Recent scans (newest first), paginated
scans = client.get_scans(number=20, index=0)

# User profile
profile = client.get_user_info()

Transport

stdio only. MCP clients (OpenCode, Claude Desktop) spawn the stdio process directly via uvx/uv.

License

MIT