Skip to content
DocsAgents and status

Agents and status

Every project has its own terminals next to its services. Start an agent in one, and lpm shows what it is doing without you watching the tab.
Lesson · 1:20 · Run any AI coding agent, with chapters and a transcript

Start an agent

  1. Open a project and click Claude or Codex in its header. A fresh install adds both buttons to every project.

  2. The agent starts in a new tab, in the project's folder, signed in as you. lpm doesn't install agent CLIs: you install them and sign in yourself.

  3. Write in the prompt box under the terminal and press Enter to send. Shift+Enter adds a line.

  4. For a second agent beside the first, click Split right, then another agent button.

New tab is ⌘T, close is ⌘W, and ⌘D splits the pane to the right.

Add any other agent

Any command can be a button. Click + Action in the header, give it a name, and type the command you would run in a terminal, such as agent for Cursor’s CLI. It joins Claude and Codex in the header. The Terminals page has the config behind those buttons.

Read the status

  • An agent's tab shows Working, Needs you, Done or Problem.
  • The sidebar lists every session under its project, with a running timer. Click a row to open its tab.
  • Claude Code, Codex, Gemini CLI, Cursor CLI, OpenCode and omp report on their own. Any other agent can post the same states with lpm set-status.
  • Claude Code and Codex also get session tools: resume a past conversation, fork one into a new tab, and switch model or effort from the menu next to Send.

Get told when one needs you

In Settings, pick a chime, a system sound or your own audio file for when an agent is done, wants approval or fails. When lpm is in the background you get a system notification too, and lpm Link sends a push to your phone.

On Linux

Sounds need a sound player, and banners need your desktop’s notification service.

Read more: lpm for Claude Code and Codex and status in the sidebar.