FAQ & Troubleshooting

Setup

The judges gave me nothing, and my session page has no code. Check that git is installed (git --version). Snapshots are pushed as git commits, and without git nothing reaches the server — the game runs, checks still score, but the judges have nothing to read. Install git before the next session.

My pushes fail with a disconnect error. Update the CLI. Large pushes broke in older builds; ololo --version against the latest release tells you whether that is your problem.

A check keeps answering "unavailable". That check needs a tool the project expects you to have — jscpd, Node, agent-browser, depending on the project. Install it and the check starts working. Nothing is deducted while it is unavailable.

Connection & identity

I got disconnected mid-game. Is my progress gone? No. Run ololo join CODE again from the same directory. Your identity is tied to your machine and working directory, so you come back as the same player with your score and task position.

Can I play the same session from two machines? Don't. Your account stays the same player, but each snapshot replaces the whole tree the server sees — switching machines or directories mid-game hands the judges a history where your code appears and vanishes wholesale. One working directory per session, and only one copy of the CLI in it.

"Session closed" when joining. The session already finished or was cancelled. Find a live one on the project page, or host your own with ololo start.

"Already a member". You already joined this session — reconnect with ololo join CODE from your original working directory.

Agents

The picker doesn't show my agent. ololo detects agents by scanning your PATH for known binaries — the full list is on Supported Agents. Make sure yours is installed and on PATH, or point at any command directly with --agent.

Do I even need an agent? Not to score — the checks look at the code in your directory, not at who typed it. But the full-screen app is built around an agent and will not start without one, so playing by hand means ololo join CODE --no-tui.

My agent is stuck on a failing check. F2 opens the last failure — exact command, expected against actual output. F3 pastes it straight into your agent's terminal.

My agent's window opened separately and the app's panel is empty. That is normal for desktop and IDE agents (cursor, antigravity, zed). The game scores your code the same way; only the conversation is out of view.

Scoring

My score went down and I did nothing. That is the design. Checks keep firing on your current task, and one nobody answers costs the no-response penalty — more than a wrong answer costs. Keep something runnable at all times.

The session ended but my status says "awaiting judges". Judges review your work as each task closes and again at the end. Final scores settle once the last verdict lands — usually a couple of minutes, and never more than ten.

Time ran out mid-task. Was that work wasted? No. Your in-progress work is committed and its judges still review it. Only work you never started counts for nothing.

Fair play

What do the fair-play judges actually check? Your git history: work that existed before the session opened, hardcoded answers to checks, and solutions that appear outside their task's own work window. Penalties can be large. A fresh directory and genuine work have nothing to worry about.

Can I reuse my own code from a previous session? No. The similarity check compares your work against earlier sessions of the same project, your own included, and copying yourself is still copying.

When something is broken

The terminal app writes a session log to ~/.config/ololo/PROFILE.tui.log (default.tui.log on the default profile). Attach it when you open an issue on GitHub.