As of , checked against the official docs.
Running several Claude Code sessions is the easy part: open more terminal tabs, or give each session its own working directory with claude --worktree <name>. The hard part is not losing track of which session is working and which is waiting for you. Name your sessions, then watch them at a glance with Claude Code’s agent view (claude agents) or an app that keeps every session’s state on screen.
How do you run several Claude Code sessions at once?
There are three ways, and they combine (official worktrees docs, as of November 2026).
- Open several terminal tabs or windows. Run
claudein each. Two sessions editing the same files in the same folder can overwrite each other, though. - Give each session its own worktree.
claude --worktree feature-auth(or-w) creates a separate working directory under.claude/worktrees/feature-auth/in your repository, on a new branch namedworktree-feature-auth, and starts the session there. Run it again with another name in another terminal for a second session. One session can build a feature while another fixes a bug, and their files don't collide. - Add tmux.
claude -w feature-auth --tmuxcreates a tmux session for that worktree;--tmuxrequires--worktree.
claude --worktree feature-auth
claude --worktree fix-login
A worktree is a fresh checkout, so files git ignores, such as .env, aren't there. List them in a .worktreeinclude file at the project root and they are copied into each new worktree. Add .claude/worktrees/ to your .gitignore.
What is really hard about several sessions?
Keeping track, not starting them. A session waiting for approval sits there silently, so with three or four of them you find out which tab needs you, and which has already finished, only by opening the tabs one by one. A session you notice ten minutes late has done nothing for ten minutes.
Naming sessions helps first. Start with claude -n auth-refactor, or type /rename auth-refactor in a session, and that name shows in the terminal title and in the /resume list.
What are the ways to watch them all at a glance?
| Way | What it shows | Agents | Worth knowing |
|---|---|---|---|
| Tab titles and session names | Each tab's session name | Claude Code | Hard to take in at a glance as tabs multiply |
Agent view (claude agents) |
For each session: Working, Needs input, Idle, Completed, Failed or Stopped. Peek, reply, attach | Claude Code background sessions | A research preview. An interactive session in another terminal appears only after you send it to the background with /bg |
| Notification hooks | An alert when a session waits for approval or input | Claude Code | The example commands don't say which session |
| Devalley (a desktop app) | One resident per session: working, waiting for you, done, with its project's name | Claude Code, Codex, Gemini CLI and more | An app for Windows and macOS. Not visible from a phone or another computer |
Agent view is Claude Code's own screen for this: one terminal that shows your background sessions' states, where Space peeks at a session and Enter attaches to it. A permission request is answered after attaching.
Devalley shows every session as a pixel resident, wherever you started it: a terminal, an editor or an app. While it works the resident is at a workplace in the village; when it has something to ask it hurries to the middle of your screen with a speech bubble. The conversation window lists the residents along its left side, with those waiting for you at the top under "Waiting for you", the one who has waited longest first. Ctrl+Alt+↑ (⌥⌘↑ on a Mac) goes to that resident from any app, and a resident's card brings its session's terminal forward.
What Devalley doesn't do: it creates no worktrees and hands out no work among sessions. Do that with the Claude Code features above; Devalley is where you watch the sessions that result and answer them.
What changes when you run Claude Code and Codex together?
Agent view manages Claude Code sessions only. Codex is watched in its own terminal or app, so mixing the two gives you two places to check. Bringing several agents onto one screen takes a tool that doesn't care which agent it is, and Devalley is one such app: Claude Code and Codex sessions are residents of the same village.
How do you find a past session and carry on?
claude --continue(-c) reopens the most recent session in the current directory.claude --resumeopens the session picker; inside a session it is/resume. In the picker,Ctrl+Awidens the list to every project on this machine andCtrl+Wto every worktree of the repository.- With a name, resume directly:
claude --resume auth-refactor. - In Codex,
codex resumecontinues the most recent chat or picks a session by ID.
Resuming the same session in two terminals as it is interleaves both sides' messages into one transcript. To split it, take a new session ID with claude --resume <session> --fork-session. The official sessions docs have the details.
Frequently asked questions
Can several Claude Code sessions run in the same folder?
They can, but sessions editing the same files may overwrite each other. claude --worktree <name> gives each session its own working directory and branch, so they don’t collide.
Where do you see the list of Claude Code sessions?
Past sessions are in the picker that claude --resume, or /resume inside a session, opens. Background sessions running now are in agent view, opened with claude agents.
Can you manage multiple Claude Code sessions without tmux?
Yes. Tabs and worktrees are enough to run several sessions, and agent view or an app such as Devalley shows their states. tmux is optional.
Does running several sessions use up your limit faster?
Yes. Every session draws from the same 5-hour and weekly limits. Check what is left with /usage inside a session.
What is the fastest way to know which session is waiting for you?
With Claude Code alone, look for Needs input in agent view. To see it across terminals, an app such as Devalley brings the waiting session to your screen.