Skip to content

Dashboard

One running stack and you remember where everything is. Four running stacks started in parallel by four agents and you don't.

The dashboard at http://lich.localhost:3300/ is a single view of every running lich stack on the machine.

Open it

bash
lich dashboard

This auto-starts the daemon if needed (any directory works, you don't have to be in a worktree) and opens the dashboard in your default browser. Pass --no-browser (or set LICH_NO_BROWSER=1) to print the URL instead of opening it.

You can also navigate to it directly: http://lich.localhost:3300/. The *.localhost resolves to loopback on every OS, no DNS setup.

What you see

Per running stack:

  • Source worktree: which directory the stack belongs to (and the resolved ${worktree.name} / ${worktree.id}).
  • Services: every compose service and owned process, with status (starting / ready / failed / stopped) and uptime.
  • Friendly URLs: clickable <service>.<worktree>.lich.localhost:3300 links, one per service.
  • Logs: tail any service inline; full per-service log files are at ~/.lich/stacks/<id>/logs/<service>.log.
  • Restart / stop: per-service or whole-stack restart, stop the whole stack.

Why this matters for parallel agents

The dashboard is what makes N-agents-in-N-worktrees observable. Without it:

  • You forget which worktrees have running stacks.
  • A failed service in agent 3's stack looks identical to a failed service in agent 7's.
  • Tailing logs means remembering the full state-dir path of each stack.

With the dashboard, all of that goes through one URL. Agents can deep-link to specific services' logs (http://lich.localhost:3300/stacks/<id>/logs/<service>) so when an agent says "the api in worktree X is failing," you have the URL to verify.

Auto-launches on first stack

The daemon (and therefore the dashboard) autostarts the first time any lich command needs it lich up, lich dashboard, lich urls, etc.

See also

Released under the MIT License.