Skip to main content
Glama
Apple Health AI Bridge MCP Server

Apple Health AI Bridge MCP Server


Apple Health AI Bridge connects Apple Health to MCP-compatible assistants. The iPhone app syncs the records you allow to your own computer, where your AI can query sleep, workouts, activity, and more.

Before we plan next week, what did my sleep and workouts actually look like?

Automatic sync brings new records into the same local history, without another export. Use it in the assistant you prefer or in your own agent workflows; your data isn't tied to a single AI provider.

Get started

You'll need an iPhone running iOS 18 or later, a macOS or Linux computer that can stay awake to receive records, and an AI assistant.

  1. Install the app. Download the iPhone app from the App Store. Use iOS Companion 1.2.0 (build 51) with the Receiver below.

  2. Set up the receiver. Install the published Receiver/CLI 1.2.0 using the setup guide, including a private HTTPS route for sync away from home.

  3. Pair your iPhone. Scan the pairing QR, review Apple's Health permissions, and enable Automatic Sync.

  4. Connect your AI. Choose your client, then ask your first question.

These versions use Batch Protocol health_bridge.batch.v1 (1.0.0). Already using the bridge? Follow the upgrade steps to keep your database and pairing. For the optional Mac-only iCloud Mailbox Beta, use the helper verification and installation guide, including its separate helper-replacement steps.

Related MCP server: MetricBridge (formerly Health Export AI)

Choose your AI

Find your assistant's setup guide below.

AI chat apps

Agents & harnesses

Coding agents & developer tools

Other local MCP clients can use the same read-only server when their configuration supports local stdio.

Supported data

Sleep, workouts, steps and energy, heart-rate measurements, weight, and body composition are supported. The full data catalog lists individual record types and their availability.

Privacy

Health access and MCP queries are read-only. The receiver stores your records on your computer, and you decide which clients can query them.

When you use a cloud model, the query results your client sends go to that model provider. Its privacy settings and policies apply.

Privacy policy · Security model · Remove local data

Join the community

Have a question or an idea you'd like to try? Join us on Discord for setup help, useful prompts, and conversations about using Apple Health with your AI.

You can also follow release news and see what we're working on next.

Documentation

License

Apple Health AI Bridge is an independent open-source project and is not affiliated with, endorsed by, or sponsored by Apple Inc.

Available Tools

9 tools
explain_sourcesB

Read-only provenance summary for local sources and sync cursors with missing-data caveats; no clinical interpretation.

Inputs: no named parameters.

get_bridge_context_markdownA

Read-only redacted Markdown bridge context for local agents/wiki; omits sample values, cursor values, token material, and clinical interpretation.

Inputs: no named parameters.

get_bridge_statusA

Read-only redacted bridge and receiver sync status for local agents; omits sample values, cursor values, and token material; no clinical interpretation.

Inputs: no named parameters.

get_daily_summaryB

Read-only daily observation summaries for a date range with source provenance, missing-data caveats, sum-safe sample_totals, and type-aware sample_statistics for non-summable metrics; no clinical interpretation.

Inputs: end_date (required), start_date (required).

get_sleep_summaryB

Read-only sleep-stage summary observations for a date range with source provenance and missing-data caveats; no clinical interpretation.

Inputs: end_date (required), start_date (required).

get_timeseriesA

Read-only time-series observations for selected metric types with source provenance and missing-data caveats; no clinical interpretation.

Inputs: end_time (required), start_time (required), type_codes (required).

get_workoutsA

Read-only workout observations for a date range with source provenance and missing-data caveats; no clinical interpretation.

Inputs: end_date (required), start_date (required).

list_supported_timeseries_typesB

Read-only supported timeseries type metadata; metadata only, no local health values, no cursor values, and no clinical interpretation.

Inputs: category.

list_synced_metricsA

Read-only catalog observation listing synced metric types, provenance, and missing-data caveats; no clinical interpretation.

Inputs: no named parameters.

Tool Schema Changelog

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

Loading tool schema changelog…

    TDQS

    A3.7/5.0

    Scored across 9 tools

    Disambiguation4/5

    Most tools target clearly different resources or metadata layers, but get_bridge_status and explain_sources overlap on sync/provenance, and get_bridge_context_markdown is a composite view that can blur with bridge status. Descriptions do help separate them.

    Naming Consistency5/5

    All tool names use consistent snake_case with a clear verb_noun pattern (get_, list_, explain_) throughout the set. No mixing of conventions or vague action names.

    Tool Count5/5

    Nine tools is well-scoped for a read-only health-data bridge; each covers a distinct resource or metadata layer without obvious bloat. The count sits comfortably in the ideal 3-15 range.

    Completeness4/5

    The surface covers bridge status/context, supported and synced metric metadata, timeseries, workouts, sleep, daily summaries, and source provenance, which is strong for a read-only bridge. Minor gaps like sync-control or deeper per-activity detail exist, but no critical dead ends for the stated read-only purpose.

    Maintenance

    ActivityActive
    ResponsivenessUsually handled

    Related MCP Connectors

    Related MCP Servers

    • A
      license
      A
      quality
      B
      maintenance
      Local-first MCP server that reads Apple Health export files (export.xml/zip) and exposes activity, sleep, HRV, and workout data to AI agents, keeping all data on your machine.
      18
      191 npm
      4
      MIT
    • A
      license
      A
      quality
      B
      maintenance
      Open-source MCP server for MetricBridge (formerly Health Export AI). 16 tools and 22 prompts let any MCP agent query 190 Apple Health metrics from files on your own machine. The server is read-only, has zero dependencies and makes zero network calls.
      21
      561 npm
      11
      MIT
    • A
      license
      Not graded
      quality
      D
      maintenance
      Read-only MCP server that exposes Garmin Connect health data from a local SQLite database to AI assistants, enabling queries for recent days, sleep, weekly trends, and a guarded SQL query surface.
      MIT