Best Wear OS apps for developers, and what is still missing

Developers buy smartwatches for the same reasons as everyone else and then discover the app store has almost nothing for the work they do. Here is what a Wear OS watch can do for a developer, which categories are worth the install, and the category that was empty until recently. No app is named below that we have not used; the categories are what matter, and they change less than the apps.

What can a developer actually do from a Wear OS watch?

Less than the marketing suggests and more than most people set up. The watch is good at three things: delivering a small number of important signals, answering a yes/no question, and taking a short voice input. Everything that fits that shape works; everything that needs a keyboard or a large screen does not.

Two things shape every row in the table. A watch has no keyboard worth using, so anything that needs typed input stays on the phone or the laptop. And it is on you all day, so anything that reaches it will reach you in a meeting, on a run and at dinner. The best developer apps are the ones that respect both constraints: little input, little noise.

Notifications, filtered hard

The single most useful thing a watch does for a developer is deliver the three notifications that matter without the phone. Spend the time on the filters: allow only the apps whose alerts you would answer in a meeting, which for most people is calls, one messaging app, the pager and CI. Everything else stays on the phone. Setting up a watch as a work device is mostly this step done properly.

The reason to be strict is the same reason to be strict with a phone: every buzz is an interruption, and a watch that buzzes for a newsletter has cost you more than it saved.

Two rules of thumb. If you would not interrupt a conversation to answer it, it does not go to the wrist. And if it needs more than a glance and a tap to deal with, it does not either; a watch is for triage, and the phone is where the reading happens.

PeakWatch does this from your own data. Your coding agent, on your wrist. Join the free beta →

Tiles: the glanceable layer

Wear OS tiles are the screens you reach by swiping from the watch face. Google's developer documentation describes them as "a way for you to show a small amount of key information, which users can read through after they glance at a tile for a few seconds", with quick actions such as starting a timer or sending a message. Tiles cannot scroll; a tap opens the app for anything more.

This is the right shape for most developer information: a 2FA code, a build status, an on-call indicator, a session that needs you. If an app you rely on offers a tile, put it one swipe from the face and you will stop reaching for the phone.

Complications do the same job on the watch face itself: a single number or icon from an app, next to the time. A build-status complication or a count of blocked sessions is the most compact dashboard a developer can carry, and it costs nothing in attention until it changes.

Two-factor, SSH and CI status

• Two-factor codes. Authenticator apps with Wear OS tiles turn a phone hunt into a glance. The code is on your wrist in the server room.
• CI and uptime status. Several dashboard apps put build and incident state on a tile. Most are thin wrappers over a web API; all of them beat unlocking the phone.
• SSH and terminal clients. They exist. Typing a command on a 1.5-inch screen is an emergency measure, not a workflow, but restarting one service from a car park has saved more than one evening.
• Timers, voice notes and reminders. The unglamorous productivity layer, and it is good. Dictating "check the deploy at four" is faster on the wrist than anywhere else.

For the health side of the same device, how the recovery scores compare is a separate question with its own answer.

What was missing: agent supervision

Anything to do with the agents that now do a large share of the work. A coding agent runs for minutes, stops to ask for permission, waits for a prompt, and finishes. Every one of those moments is a natural watch interaction: read a sentence, tap, dictate a line. No watch app did it, because a watch cannot run the agent and there was no protocol for driving someone else's session.

Anthropic's Remote Control, which arrived in 2026, closed part of the gap for the phone: a Claude Code session on your machine can be driven from the Claude app, with permission prompts forwarded and push notifications when it needs you. It is worth setting up whatever watch you wear. It does not reach the wrist.

What the wrist needs is narrower than what the phone offers, and that is why it fits. You do not want the transcript on your wrist. You want to know which session is blocked, what it wants to do, and a way to say yes, no or why in under five seconds. Everything else can wait for the phone or the desk.

How PeakWatch fills it

PeakWatch pairs the watch with a small relay on your machine that owns the sessions through the Claude Agent SDK. The watch lists sessions ranked by what needs you, approves or denies tool requests with the arguments in view, takes prompts by voice, streams replies, summarises a session in one screen and shows your plan usage. It also mirrors sessions you started in a terminal. Standalone over WiFi or LTE; phone optional.

It is in private beta on Wear OS watches, including Galaxy Watch, so for now it is a waitlist rather than a Play Store listing, and that is said plainly.

What to want next

Diff previews on the wrist for small changes. Voice-driven code review for pull requests. A richer set of quick replies learned from your own habits. A tile that shows which session is blocked without opening the app. And the same supervision surface for agents beyond Claude Code.

Hen Eytan · Founder, UPeak

Builds the UPeak apps and writes up the research behind them.