THE POINT OF VIEW
ABOUT L.CODE
A window
around an agent
that already worked.
L.code does not reimplement Claude Code. It gives it a desktop the way an editor gives a compiler one: the same tool underneath, and everything it did visible while it does it.
Direct downloadNo account on this site, on any platform
You should be able to see
The agent is fast enough that the limiting factor is no longer the model — it is how much of what it did you can actually hold in your head. Everything in this app exists to make that part smaller.
Desktop platforms
macOS, Windows and Linux, built from the same source for every release.
Screens on this site
Every screenshot here is the current build, not a mockup of one.
Interface languages
And a separate setting for the language the agent replies in.
Ways to bring a model
An official provider sign-in, or your own endpoint and key.
WHERE IT RUNS
The same app,
on all three desktops.
macOS · Apple Silicon and IntelWindows · x64Linux · AppImage and .deb
HOW THE WORK IS DIVIDED
Small surface.
Deliberate edges.
Four areas, each with a rule it is not allowed to break. The rules are why the app stays predictable as it grows.
The desktop shell
Windows, tabs, and the session that lives in one
The agent bridge
Permissions, diffs, and approval before an action runs
Models and quota
Provider sign-in, your own keys, and what each turn costs
Release and updates
Signed builds, this site, and the feed installed copies read
THE SHORTEST WAY TO JUDGE IT