HeadsUpAI

Claude Launches Agent View to Manage Fleets of Parallel Coding Agents

Claude launched agent view in research preview for Claude Code, a centralized dashboard for managing multiple agentic coding sessions within the terminal. Accessible via the claude agents command, the list displays every active task, its current status, and whether the agent is waiting for human input.

Managing parallel agents previously required juggling multiple terminal tabs, creating significant mental overhead. This update complements Claude Code's session recaps by providing a dedicated control tower for long-running tasks. It allows developers to scale output by dispatching several ideas at once and monitoring them from one interface.

You can now background active sessions using the /bg command or launch fresh tasks directly into the background. A "peek" feature lets you view the last turn and reply inline to unblock an agent. Agent view is available in research preview for all paid Claude plans.

Claude
Claude
@claudeai
X

New in Claude Code: agent view. One list of all your sessions, available today as a research preview. https://t.co/NnbsAQjSPW

2.2kretweets29klikes
View on X

Still wondering? A few quick answers below.

Agent view is a centralized management interface within the Claude Code terminal that allows developers to track and interact with multiple concurrent AI sessions. It provides a row-by-row status of every session, showing whether an agent is currently working, finished, or waiting for human input to proceed with a task.

You can access agent view by running the claude agents command directly in your terminal. Alternatively, if you are already inside an active Claude Code session, you can press the left arrow key to open the view. This allows you to switch between different tasks without losing your place in any specific session.

Agent view is currently available as a research preview for users on several paid Anthropic plans. This includes subscribers on the Pro, Max, Team, and Enterprise plans, as well as those accessing Claude through the API. Standard rate limits for your specific plan still apply when running multiple concurrent sessions in the background.

To move an active session to the background, you can use the /bg command within the session. If you want to start a completely new task in the background without entering the foreground first, you can run claude --bg followed by your specific task description from your terminal command line.

Yes, agent view includes a peek feature that allows you to see the last turn of any session. If an agent is blocked and waiting for a decision, you can reply to the prompt inline from the list to unblock it. This removes the need to fully attach to the session transcript.

Share this update