LUNAR LABS docs
$LUNE · Documentation v1

Everything you need before your first night off.

How Lunar Labs works, how to give it work, and how to stay in control while it runs.

Quickstart →Your first shift in four steps Inside a shift →The six stages, start to proof Writing goals →Get exactly what you wanted

Meet the engineer who works while you sleep

Lunar Labs is your own unattended software engineer. You give it a goal on your codebase, walk away, and come back to finished work: planned, built, reviewed, verified and opened as a pull request.

It's built for builders who have more ideas than hours. You stay in charge of what it builds, what it can touch and what "done" means. Lunar Labs handles the shift.

What one shift hands back

  • A map of your codebase before any code is touched
  • A plan split into small units of work
  • Code written, reviewed independently and repaired when something breaks
  • Proof that the result works, attached to the pull request
  • A memory of everything it learned, ready for the next shift

Your first shift starts in four steps

01

Open Lunar Labs

Open the Lunar Labs app and sign in.

02

Connect your codebase

Point Lunar Labs at the repository you want it to work on.

03

Set the rules

Choose what it may change, which tools it may use and what counts as done.

04

Give it work, leave it running

Describe your goal in plain words, start the shift, and check back when the pull request opens.

Inside a shift: what happens after you log off

Every shift runs the same six stages, and nothing is called done until it's proven.

  1. 01

    Understand

    It studies your repo and builds a feature map of where everything lives.

  2. 02

    Plan

    Your goal becomes a step-by-step plan, broken into small units of work.

  3. 03

    Execute

    Units run one after another, or side by side in isolated environments when it's safe.

  4. 04

    Review

    Each unit gets an independent review. Anything that breaks is repaired and checked again.

  5. 05

    Prove

    It verifies the result and attaches the evidence, then opens a pull request.

  6. 06

    Watch

    The shift doesn't end at the PR. It keeps watching and responds to what happens next.

It never forgets what happened last night

Lunar Labs keeps a living memory of your project, so every shift starts smarter than the last.

It remembersWhy it matters
What it learnedYour repo's patterns, quirks and conventions carry over
What changedEvery edit it made, and why, stays on record
What workedGood approaches get reused
What failedYesterday's dead end isn't tonight's
What happens nextThe next shift picks up a clear queue

Over time, Lunar Labs uses that memory to improve how it works, tune its own workflows and suggest improvements to the project itself.

Unattended doesn't mean unleashed

You set the rules of the shift before it starts. Lunar Labs works only inside them.

What it builds

It only takes on the goals, features and fixes you hand it.

What it can touch

You choose which files, folders, tools and environments are in bounds.

What "done" means

You define the finish line. It doesn't stop until it can prove it got there.

Everything lands as a pull request, so nothing reaches your main branch until you approve it.

The secret to a great shift is a great goal

Say what you want, where it lives and how you'll know it's done.

Instead of

"Fix the login"

Try

"Users get logged out after 5 minutes on mobile. Find the cause and fix it. Done = they stay signed in for 7 days."

Instead of

"Add billing"

Try

"Add team billing to the dashboard settings page. Done = an admin can add a card and see the monthly total."

Instead of

"Clean up the code"

Try

"Remove unused files in the /utils folder. Done = all tests still pass."

  • One goal per shift works better than a long wish list.
  • Name the place. Mention the page, feature or folder it should work on.
  • Define done. Lunar Labs proves its work against your definition, so make it checkable.

Still wondering? Start here

Do I need to watch it while it works?

No. That’s the point. Give it work, leave it running, and come back to the result.

Can it break my project?

It only touches what you allow, and its work arrives as a pull request you review before anything is merged.

Does it work on more than one thing at once?

Yes, when it’s safe. Independent units of work can run side by side in isolated environments.

What happens after the pull request opens?

Lunar Labs keeps watching it and responds to what happens next.

Does it get better over time?

Yes. Its project memory helps it improve its workflows and its understanding of your codebase with every shift.

Where can I get help?

DM us on X at @lunarlabs_rh.

Ready to clock out?

Give it work. Leave it running.

Open the app