GIT & VERSION CONTROL / A VISUAL LEARNING LAB
Stop memorizing commands in the dark. See what moves, what stays, and how to recover—one small mission at a time.
Your files have a story.
Learn to read—and shape—it.
Explore the sandbox, then check your result.
Reveal the steps, then play them one at a time. Starting a walkthrough reloads this mission’s fixture. You can inspect every intermediate snapshot.
The editor is a draft until you apply it.
One command at a time. No shell, network, untracked files or remote operations. Invalid commands do not change the model.
Every retained practice commit is shown, including commits not currently reachable from a branch. Lines connect commits to their parents. Select a commit to inspect its saved file.
This independent experiment starts at c1 with the launch plan committed, plus an additional uncommitted accessibility line. Choose a reset mode targeting c0 and compare the resulting snapshots. It does not affect your mission.
Real hard resets can discard work. This fixed model has one tracked file and no untracked files or submodules.
No remote, push, pull, merge commits, conflicts, rebase, detached HEAD, binary files, untracked files, filesystem permissions or authentication. Switch requires a clean tree, and stash pop requires a clean matching base. These restrictions are stricter than real Git in some situations. Revert supports only the current single-parent tip. The graph uses retained synthetic IDs rather than hashes, and the reflog is simplified.
Unsupported commands are rejected rather than simulated incorrectly. Mission goals are learning checks on this model, not certification for real repositories.
git init and configure the name and email you want associated with practice commits.git branch --show-current to check the actual initial branch name; do not assume it is main.The browser’s Undo lab step button is a teaching feature, not a Git command. Reflog may help with committed history, but it does not guarantee recovery of arbitrary uncommitted file edits.
Current mission actions, reflections, predictions and earned missions save locally when storage is available. Export a backup to move devices. Mission replay reconstructs its current sandbox on reload.