Claude Code adds agent view for managing parallel sessions

Anthropic said on May 11, 2026, that it is introducing agent view in Claude Code, a new way to manage multiple Claude Code sessions from one place. The feature is aimed at developers running agents in parallel and trying to keep track of which sessions need input, which are still running, and which have finished.

Before agent view, users often had to juggle multiple terminal tabs, tmux panes, and a mental checklist of what each agent was doing. Anthropic said agent view is meant to reduce that overhead by giving users a single list of active sessions and their status.

According to the company, users can start new agents, send them to the background, and only step in when Claude needs help. Each row in the view shows the session, whether it is waiting on user input, the last response, and the last time the user interacted with it.

The feature is accessed from the command line interface, or CLI, by pressing the left arrow from any session or by running claude agents in the terminal. Anthropic said users can then see all sessions at once and choose whether to peek at the last turn, reply inline, or attach directly to a session for the full transcript.

If a session is waiting for a decision, the user can answer inline and the agent continues from there. If the user wants the full context, pressing enter opens the session directly. Anthropic said users can also move an existing session into agent view with /bg, or launch a new background session with claude --bg [task].

The company said early users have been using the feature to run several ideas at once, each optionally paired with a skill, and later review a list of pull requests ready for review. Anthropic also pointed to long-running jobs such as PR babysitters and dashboard updaters, where the next run time is visible in the list.

Another use case Anthropic mentioned is moving between related tasks without losing context. Users can jump out of one session, start a quick codebase question or related task, then jump back to where they were. The peek view is designed to show the answer when it arrives.

Anthropic said status indicators and the peek title also make it easier to scan which sessions produced a pull request. That makes it simpler to sort through work when many agents are running at the same time.

Agent view is available today as a Research Preview on Pro, Max, Team, Enterprise, and Claude API plans. Anthropic said users can opt in by running claude agents, and that usual rate limits apply.