I built an alarm for the moment your coding break quietly becomes an hour

Session: scanning Codex and Claude Code usage screens, planning a return, and what ADHD research can actually tell us.

You hit your coding agent's limit. You step away while it resets. You check your phone. The tool becomes available again, but you are somewhere else entirely.

The break was reasonable. Losing the rest of the afternoon wasn't the plan.

I'm Kaan, an indie developer at DijitalGözlük. I built Session: Focus & App Lock around that return to the desk. I want it to be useful to people with ADHD who work at a computer, especially developers. It is also for anyone who recognizes this pattern without having ADHD.

Session's App Store presentation, captured November 2026. The examples below use sample reminder data.

Remembering to return is a design problem

“Just remember to come back” leaves quite a lot to the person taking the break: decide when to return, keep that intention alive, notice the time, and restart the work.

While preparing this article, I looked at research that helps put those steps in context.

A 2013 experiment on prospective memory in adults with ADHD compared 45 adults with ADHD with 45 controls. The clearest difficulty was planning the future action. The other measured components did not all show the same impairment. That matters: ADHD is not a simple story in which everyone forgets everything.

A separate task-interruption experiment by Altmann and Trafton studied the delay involved in resuming a task and the role of cues. It suggests that having context available can help people find their place again. This was laboratory research, not an ADHD study or a test of a reminder app.

There is a developer connection, too. Abad and colleagues studied interruptions in software development, using recorded tasks and a survey. The disruption depended on context and the kind of interruption. “Taking a break” is not one uniform event.

For Session, the practical design question is: what can the app make explicit before you leave?

The part I wanted to make easier

Imagine Claude Code says your usage will reset later today. Instead of remembering that time or copying it into another app:

1. Open the supported Claude Code /usage or Codex /status screen on your computer.
2. Point Session's camera at the usage rows.
3. Review the return time Session reads from the screen.
4. Choose how much help you want, then take your break.

The live camera reads automatically after matching valid frames. There is no shutter to press. You still confirm the proposed time. If the screen can't be read, you can retry or set the time manually.

This is screen reading, not an account integration. Session doesn't sign into your Anthropic or OpenAI account or keep polling it in the background. The supported screen and the time you confirm are what the reminder is based on.

The guided first-reminder entry point. Manual setup is available too.

The scan is deliberately narrow. It is for supported usage screens, not a promise to understand any dashboard, screenshot or terminal layout. Camera images and recognized text are processed on device. That is separate from product analytics, which has its own opt-out.

A gentle reminder, or a stronger return cue

Reminder Only rings once at the time you chose. It blocks nothing.

Focus Lock adds a stronger commitment. With Screen Time permission, you choose which distracting apps to restrict. Returning to a supported work screen and completing the scan ends the reminder and removes Session's restrictions. Recovery Unlock is available if the scan cannot be completed.

It doesn't lock the whole phone. And a successful scan is not proof that you spent the next hour doing useful work. It is a concrete return action, which is the smaller problem I am trying to help with.

Focus Lock is optional. Reminder Only remains the gentler path.

A visible return time on Today. Sample data, not a real user's activity.

Why another focus app?

There are already good tools here. The distinction I care about is where each one intervenes in the day.

Alarmy is centered on waking up. Its official mission features include math, photo and QR/barcode tasks. Its photo missions are an established example of asking for an action before dismissal.

Opal offers app blocking, Focus Rules and timed focus sessions. Its reminders and broader controls make it useful across the day.

Opal screenshots shown in its App Store listing. © Opal OS. This is a feature comparison, not an endorsement or a hands-on effectiveness test.

one sec puts an intervention in front of opening a distracting app. Its official FAQ also describes complete blocking, so it is more than a breathing prompt.

Official one sec press-kit page. © one sec. The two app screens illustrate the interruption before opening an app.

Session's specific workflow is the coding break: read a supported usage reset screen, choose a return time, and optionally use a work-screen scan to complete Focus Lock. You can use it alongside a focus routine you already have.

I am aiming at that specific transition from waiting for a coding tool to returning to it. Which approach fits depends on the part of your day that needs help.

What I'd like to learn from you

Session is free to download, with a limited number of reminders each week. Pro removes the weekly limit and unlocks History. You don't have to buy it to try the basic loop.

If you work at a computer and have ADHD, what happens between “my break is over” and actually reopening the work? Would an agreed return time help, or would you dismiss it? Does scanning the work screen feel useful or like another hurdle?

I'd rather hear where the idea breaks down than pretend one alarm fixes attention.

Try Session on the App Store · Website

Disclosure: I make Session. It is a productivity tool, not an ADHD treatment, and has not been clinically evaluated for ADHD. Session is independent of OpenAI and Anthropic. Competitor features were checked against their official pages in November 2026. This draft was prepared with AI assistance; I am responsible for the final text.