Google opens a CLI for Fitbit and Pixel Watch health data

Why it matters

Google is turning Fitbit and Pixel Watch data into an agent-readable developer surface, giving startups new primitives while deepening their dependency on Google Health API.

Google (@Google) has opened a command-line interface for Google Health, giving developers, health-data power users and AI agents a direct way to query Fitbit, Pixel Watch and other Google Health API data from the terminal, 9to5Google reported.

The tool, called ghealth, is less a consumer feature than a platform move. It turns health metrics that used to live behind app dashboards and mobile integrations into data that can be exported, scripted and handed to agents. Google describes the GitHub project as a CLI for Google Health API v4 "built for AI agents and developers," with support for 40 verified data types including steps, heart rate, exercise, sleep, weight, SpO2, HRV, ECG, blood glucose, nutrition and VO2 max.

That framing matters because Google is not merely adding a developer convenience. Google is rebuilding the Fitbit developer surface under the Google Health brand, while also pushing AI coaching and a new low-cost tracker deeper into the product line. Rishi Chandra, Google's GM for Health & Home, put the consumer side of that consolidation on the record in May with the launch of Google Fitbit Air and Google Health Premium. Taylor Helgren, Google's director of product management, separately said the Fitbit app would become the Google Health app and start rolling out automatically on May 19.

The CLI is the plumbing that makes that strategy usable outside Google's own app. Google's developer documentation says the Google Health API is the next generation of the Fitbit Web API, built to view and manage fitness and measurement data from Fitbit, Pixel Watch and third-party devices and apps on a unified OAuth 2.0 infrastructure. The same page says Google rebuilt the Fitbit Web API from the ground up and consolidated more than 100 legacy endpoints into data type bundles.

What Google actually shipped

The GitHub repository is public and Apache 2.0 licensed. At the time of publication, it showed 3 stars, 0 forks, 2 commits and no packaged releases. That is early for an open-source project from a major platform owner, and it suggests Google is putting a reference tool in developers' hands before a broader ecosystem has formed around the new API.

Developers install the CLI by cloning the repo and building a single Go binary. The README shows commands for one-time setup, weekly step totals, recent heart-rate readings and listing schema types. Setup walks users through a Google Cloud project ID, OAuth credentials, API enablement, scope selection and browser-based login. The repo also documents non-interactive setup for agents or CI, headless OAuth, token export and import, and authentication through environment variables or Google Cloud application default credentials.

The security tradeoff is visible in the README rather than buried in a policy page: the tool writes configuration under ~/.config/ghealth/, including client_secret.json, credentials.json and config.toml. The README says the OAuth client file is stored with mode 0600, but the credentials file contains access and refresh tokens as plaintext JSON. That is standard enough for local developer tooling, but it is also the exact implementation detail founders building on this surface will have to design around if they expect users to connect personal health data to automations or AI agents.

Google's support post, as summarized by 9to5Google, says the CLI can export data into JSON, terminal tables and CSV files for spreadsheets or custom dashboards. Google also says the tool ships with two starter skills for setup and authentication, so an agent can be pointed at the README and use the CLI as a tool rather than a human-only terminal utility.

The real platform shift is Fitbit becoming Google Health

The CLI arrives less than two months after Google renamed the Fitbit app into Google Health and paired it with Google Health Coach, an AI health product that Google says is part of Google Health Premium. Chandra's May 7 announcement priced Google Health Premium at $9.99 per month or $99 per year and said it is bundled with Google AI Pro and Ultra. The same announcement priced Fitbit Air at $99, with a Stephen Curry Special Edition at $129.

That packaging is the business context for the CLI. Google acquired Fitbit in 2021 and has since folded Fitbit hardware, app experiences and developer access into Google's account, cloud and AI stack. Google Health API is not positioned as a sidecar to Fitbit; Google calls it the successor to Fitbit Web API. For developers who built on Fitbit's old endpoints, the practical message is migration. For new teams building health dashboards, coaching tools, nutrition correlations or recovery alerts, the message is that Google wants Google Health API to be the default surface, not a retrofitted wearables feed.

Google's own examples make the intended automation layer explicit. 9to5Google lists use cases including daily syncs of sleep and recovery trends, calendar blocks when readiness is high, alerts when vitals deviate from a personal baseline, and workflows that correlate nutrition logs with blood glucose. Those examples are not full products. They are primitives for products. A founder building an AI coach, employer wellness dashboard, quantified-self tool or clinical-adjacent engagement app now has a Google-sanctioned way to pull structured data without starting from a mobile integration.

A new opening, and a new dependency

For startups, the opportunity is also a dependency. Google Health API gives third-party builders standardized access to Fitbit, Pixel Watch and compatible third-party health sources. It also puts those builders inside Google's OAuth, Google Cloud and Google Health policy regime at a time when the consumer experience is being tied to paid AI subscriptions.

That changes the competitive map. Apple HealthKit remains the natural anchor for iPhone-native health apps. Garmin Health API remains important for performance and endurance data. Samsung Health Data SDK and Health Connect matter in Android health aggregation. Unified data API vendors such as Terra sell abstraction across device ecosystems. Google's bet with ghealth is narrower and more strategic: make Google Health data agent-readable and scriptable at the developer edge, then let builders pull that data into dashboards, local automations and AI workflows.

The unanswered question is adoption. A GitHub repo with a few commits and no releases is not an ecosystem. Google has not disclosed CLI usage, developer signups, commercial API customers or the approval burden for broad-audience apps. Google also has to persuade health developers that its consolidation of Fitbit will produce a stable platform rather than another API migration cycle.

Still, the direction is clear. The Fitbit era was defined by a wearable, an app and a consumer dashboard. Google's next health stack is a device, a subscription coach, a unified API and now a terminal tool designed so software agents can operate on personal health data. That is a different kind of platform, and Google has started exposing it at the level where developers and founders actually build.