Sign in

It lives in your menu bar, beside your work.

The menu bar item opens today's session: the reading beside your editor, the starter file in your folder.

When you say the work is done, the check runs and the result goes on your record.

Download for Mac

For macOS 15 and later. The app asks for the code on your download page and nothing else.

lrnit builds you a path from where you are.

The route runs from what you already know to the goal you typed, one node a session.

A node you miss comes back in a later session, so nothing is passed once and lost.

  1. ModulesWhere you are
  2. net/http
  3. Goroutines
  4. Channels
  5. Errors
  6. Context and cancellationToday
  7. TimeoutsReturns in a later session
  8. Testing
  9. Graceful shutdownReturns in a later session
  10. Containers
  11. Observability
  12. Deploy
  13. Write a Go service that survives production

Every session ends with proof you can do it.

Choose the time you have, then work toward one goal from a short briefing.

lrnit runs the task's command in your folder, keeps the exit code and the output, and that is the evidence your record holds.

Task

Finish the handler so cancellation stops the work.

Cancel stops the workyou wrapped the fetch in the request's context

Stops the work when the request is cancelled
    <-ctx.Done()
    srv.Shutdown(context.Background())

exit 0, stopped 12 ms after Ctrl-C

Stop the work when the request is cancelledgo run main.go, exit 0Today

How lrnit teaches

Five things a session does, and the evidence for each.

Each one is a decision we took from what the studies found, not a rule we made up.

  1. At the end of the session the reading is closed and you answer from memory.

    In the studies reviewed, retrieving an idea from memory produced more learning than studying it again, so lrnit closes the reading before the check.

    The studies

    Across 222 classroom and laboratory studies involving 48,478 students, retrieving an idea from memory improved learning compared with studying it again. These studies cover many test formats, but they do not show that every short answer is judged accurately.

    Dunlosky and others, 2013. Yang and others, 2021.

  2. A missed idea returns at the start of a later session.

    For learners like the ones studied, spacing later retrieval helped knowledge last, so lrnit brings back a node you did not hold.

    The studies

    In an online study of 1,354 adults, spaced retrieval kept material available longer, with the useful gap changing with how long it needed to last. These studies do not validate lrnit's schedule or show that one interval fits every subject.

    Cepeda and others, 2008. Soderstrom and Bjork, 2015.

  3. You answer in your own words or make the code run.

    In the studies reviewed, generating an explanation and practising the target skill supported learning better than recognising an answer, so lrnit asks for a written response or runnable code.

    The studies

    Across studies of prompted self-explanation and observational data from a psychology course online, producing an explanation or doing the work was linked to stronger learning than passive study. These studies do not prove that runnable code caused the gain, or that explaining always beats equal-time rereading.

    Bisra and others, 2018. Koedinger and others, 2015.

  4. The session opens with its parts and one goal.

    For learners like the ones studied, seeing the structure before the work and pursuing a specific goal supported later performance, so lrnit begins each session with both.

    The studies

    In three experiments, learners who saw the parts and their states before a mechanism ran performed better on later transfer tasks. Goal-setting research also supports one specific, demanding goal, but neither source tests lrnit's exact briefing.

    Mayer, Mathias and Wetzell, 2002. Locke and Latham, 2002.

  5. When you are stuck, lrnit asks the next question.

    For students in the study, unrestricted AI help raised practice scores but lowered unaided performance, so lrnit prompts your next step instead of giving you the answer.

    The studies

    In a field trial with about a thousand high-school maths students in Turkey, unrestricted AI help raised practice scores but lowered later unaided exam results. The study does not test lrnit's exact question-only approach or show the same result in other subjects.

    Bastani and others, 2025.

What do you want to be able to do?

Type it and lrnit builds you a plan, sits beside you while you do it, and keeps proof of what stuck.

Start with the work

Write a Go service that survives production

It opens in your menu bar. Type the code from your download page once, choose a folder, and today's session is there.