Claude Code Agent View: One Screen for Every Background Session

Canonical version: Claude Code Agent View: One Screen for Every Background Session.

Claude Code v2.1.139 ships agent view, opened with claude agents. It's a single inbox-style screen for every background Claude Code session you've dispatched, across every project. Rows show what's running, what needs your input, what's done. You dispatch new sessions from the same screen, peek at any of them in a side panel, and attach to one when you want the full conversation.

This is a research preview, so the UI and shortcuts may still change.

Why it matters

Running several Claude Code sessions in parallel used to mean several terminals, several tmux panes, and the constant mental tax of remembering what each one was doing. Agent view replaces that with one inbox. Dispatch a bug fix, a pull-request review, and a flaky-test investigation as three rows. Keep working in another window. Step in only when a row says it needs you or shows a green PR dot.

Each session is its own Claude Code process, hosted by a per-user supervisor. They keep running with no terminal attached. You can close agent view, close the shell, or start a fresh interactive session; the dispatched work continues. Sessions only stop when your machine sleeps or shuts down, and claude respawn --all brings them back from where they left off.

How it works

Open the view:

claude agents

Sessions appear grouped by state, with Needs input and Ready for review pinned above Working and Completed. Each row shows an icon (animated when working, yellow when blocked, green when done, etc.), a name, a one-line summary of the current activity, and how long ago it last changed. When a session opens a pull request, a coloured dot appears on the right and links straight to the PR.

The core loop:

• Type a prompt and press Enter to dispatch a new background session.
• Select a row and press Space to peek (last output or pending question), then type a reply and press Enter to send it without leaving the screen.
• Press Enter or → to attach to a session in full.
• Press ← on an empty prompt to detach back to the inbox (or to background any session you're currently in).

Dispatch patterns worth knowing

Several prefixes change how a dispatched session starts:

• @<subagent-name> runs a custom subagent as the session's main agent.
• @<repo> runs the session in a sibling repository under the directory agent view was opened from.
• /<skill> suggests skills to dispatch as the prompt; recurring tasks belong in a skill so you can start them by name.
• #<number> or a PR URL selects the existing session for that PR if there is one.
• Shift+Enter dispatches and immediately attaches.

From the shell, claude --bg "<prompt>" does the same thing without opening the view, and claude --agent <name> --bg runs a specific subagent as the main agent. Sessions get short IDs you can use with claude attach, claude logs, claude stop, claude respawn, and claude rm.

Worktree isolation by default

Before editing files, a background session moves itself into an isolated Git Worktree under .claude/worktrees/. Parallel sessions can read the same checkout but each writes to its own, so two agents working on the same repo at the same time don't trample each other. The worktree is removed when you delete the session, so push or merge what you want to keep first.

How it compares to the other parallel options

Agent view is another way to run Claude Code in parallel, alongside the existing options:

• vs AI Subagents: subagents run inside one session and report back to it. Agent view sessions are independent processes, each with its own conversation and quota.
• vs Claude Code Agent Teams: teams coordinate via a shared mailbox and task list and can message each other. Agent view sessions are siblings, not collaborators.
• vs Claude Code Web: Web runs sessions in Anthropic's cloud and survives your machine sleeping. Agent view runs locally.

If you've never used any of these, agent view is the easiest entry point. You can use claude agents as your primary entry point instead of claude, dispatch everything from there, and attach when you want the full conversation.

Limitations

• Background sessions run on your machine and stop on sleep or shutdown.
• Each session consumes subscription quota independently; ten in parallel burns quota roughly ten times as fast as one.
• Deleting a session removes its worktree, so push or merge first.
• Subagents and teammates a session spawns don't show up as separate rows.
• You can disable the feature entirely with the disableAgentView setting or the CLAUDE_CODE_DISABLE_AGENT_VIEW environment variable.

References

• Agent view documentation: https://code.claude.com/docs/en/agent-view
• Run agents in parallel (comparison page): https://code.claude.com/docs/en/agents

Related

• Claude Code
• Claude Code Agent View
• Claude Code Agent Teams
• Claude Code Web
• Claude Code Tasks
• AI Subagents

About Sébastien

I'm Sébastien Dubois, and I'm on a mission to help knowledge workers escape information overload. After 20+ years in IT and seeing too many brilliant minds drowning in digital chaos, I've decided to help people build systems that actually work. Through the Knowii Community, my courses, products & services and my Website/Newsletter, I share practical and battle-tested systems.

I write about Knowledge Work, Personal Knowledge Management, Note-taking, Lifelong Learning, Personal Organization, Productivity, and more. I also craft lovely digital products and tools.

If you want to follow my work, then become a member and join our community.

Ready to get to the next level?

If you're tired of information overwhelm and ready to build a reliable knowledge system:

• 📚 KM for Beginners — 10+ hours of structured video lessons
• 🚀 Obsidian Starter Kit — Ready-made vault with 40+ templates
• 💼 Knowledge Worker Kit — Complete guides + lifetime community
• 🦉 1-on-1 Coaching — Personalized guidance
• 🎯 Join Knowii — Community + ALL courses & tools