DOCUMENTATION
Everything the cockpit does.
Short pages, written next to the code they describe. If a page and the app disagree, the app is right and the page is a bug.
START HERE4 min
Install and first run
Install the app, point it at a repository, pick an execution environment and start your first agent. Nothing to configure before the first diff.
1 winget install helmry # or brew install --cask helmry
2 open Helmry → Add repository → ~/code/api
3 choose environment: local | wsl:ubuntu
4 ⌘K → New agent → paste the task
Install & updates
Installers per platform, unsigned-build warnings, how updates work, uninstall, where data lives.
First run
Account, Claude sign-in, first repository, first agent, first review.
The Cockpit
The screen explained: top bar, fleet rail, Orchestrate, Focus, console, keyboard shortcuts, status vocabulary.
Windows & WSL
Which side runs the server, per-agent WSL environments, environment overrides.
Working with agents
Models, reasoning effort, permission modes, worktrees, queueing, stopping, rewinding, background work, cost, saved prompts.
Reviewing & committing
The file-edit feed, Diff & review, the commit rail, artifacts.
Observed sessions
Bringing your own terminal claude sessions onto the board via hooks.
Git credentials
Why Helmry stores no git credentials, the one-time setup per platform, and the WSL credential-helper trap.
Settings
Every preference and every config.json key, with defaults.
Privacy & security
What stays local, what an account grants, the network rules.
Troubleshooting
Logs, health checks, and the failures people actually hit.
Something missing or wrong?
Docs live beside the code. Tell us at support@helmry.com and it gets fixed with the next version.