The best terminal for Claude Code and Codex, with every agent in view.

Live status, sounds, and banners cover Claude Code, Codex, Gemini CLI, Cursor CLI, OpenCode and omp. Cursor CLI asks for approval without telling lpm, so it never shows Needs you. Any other CLI runs fine in lpm terminals and can post its own status with lpm set-status.

1. 01

   A second agent starts on a copy that already runs

   Duplicate clones the project folder, so the new copy arrives with node_modules, .venv, .env files, and your uncommitted work in place. Only rebuildable caches like .next and dist are left out. Want a branch instead? New worktree makes a real Git checkout of your current commit. It starts clean, so uncommitted changes and ignored files like node_modules and .env stay behind in the original.

2. 02

   One prompt, several answers to choose from

   Write the prompt once and choose Run in duplicates: it starts in this tab and in new copies of the repo, one agent per folder. Pick 2 to 10 runs; the Duplicate dialog it opens lets you change the count before anything starts, and the same prompt is queued in each. You keep the version you like instead of rerolling one session and hoping. The full walkthrough lives on running Claude Code in parallel.

3. 03

   Any conversation picks up where you left it

   The resume picker lists every Claude Code and Codex conversation for the project, including ones started outside lpm and tabs you closed. Fork a conversation into a new tab to try another direction, or copy the last answer as clean text with its tables intact. After a relaunch, agent tabs reopen on the conversation they had.

4. 04

   You see how much of your plan the agents have burned

   Claude and Codex each get a usage bar in the sidebar with its reset time, for the window you pick: weekly out of the box, the 5-hour window, or whichever of the two is higher. A pace marker shows whether you are burning faster than the clock, and each Claude account you add gets its own row. Codex needs no setup; Claude tracking is one click in the Usage window.

• Does lpm work with Claude Code and Codex out of the box?

  Yes. Every project gets a Claude and a Codex button on first launch, and lpm connects to both agents automatically, so the sidebar shows when an agent is working, needs an answer, hit a problem, or finished. Your services run in panes beside them. lpm doesn't install the agents themselves: you install the Claude Code and Codex CLIs and sign in as usual.
• Will lpm tell me when Claude Code or Codex needs me?

  Yes. The agent's tab and its row in the sidebar change state, a chime plays when an agent finishes, asks for approval, or errors, and a system notification names the tab and project when no lpm window is in front. With the lpm Link iPhone app paired, the same events arrive as encrypted push notifications.
• Can I see my Claude Code and Codex usage limits?

  Yes. The bottom of the sidebar shows how much of each plan window is used and when it resets, with a pace marker, and the Usage window shows the 5-hour and weekly windows side by side. Codex meters appear on their own once Codex runs; for Claude, press Enable once in the Usage window. Meters apply to subscription logins, since API-billed accounts have no 5-hour or weekly limit.
• Can I resume an old Claude Code or Codex session?

  Yes. Resume session lists every Claude Code and Codex conversation for the project, including ones started outside lpm and tabs you closed, with search and a filter by agent. Picking one reopens it in a new tab, and you can fork a conversation to try a different direction without losing the original.
• Can I run multiple agents in parallel on the same repo?

  Yes. Each agent works in its own copy of the repo, so they never edit the same files. Duplicate clones the project folder (dependencies, .env files, and uncommitted work included), and New worktree checks out a real Git worktree on its own branch. Either way you can create up to 50 at once and queue the same prompt in each. The guide to running Claude Code in parallel walks through it, and the worktree guide covers which of the two to reach for.
• Does lpm support Gemini CLI, Cursor CLI, and OpenCode?

  They run in lpm terminals like any other command, and lpm suggests launch buttons for Gemini CLI and OpenCode when their CLIs are installed. Gemini CLI, Cursor CLI, OpenCode and omp also get live status and alerts, though Cursor CLI asks for approval without telling lpm, so it never shows Needs you. Resume, fork, and the model switcher are built for Claude Code and Codex. Any agent can still drive lpm through the lpm CLI and agent skills, and post its own status badge with lpm set-status.
• Do I need Docker to use lpm?

  No. lpm runs anything that runs in a terminal. Docker Compose stacks work, and plain Rails, Next.js, Go, Django, or Flask projects work without containers.
• Is lpm open source?

  Yes. The source, issue tracker, and releases live on GitHub. Contributions and bug reports are welcome.
• Which frameworks does lpm work with?

  Rails, Next.js, Django, FastAPI, Laravel, Phoenix, Go, Docker Compose, and anything else that runs in a terminal. When you add a folder, lpm reads its manifests and sets up the services for you. To go further, have your AI agent draft services and actions from the config editor.