Claude Code has launched an innovative feature called agent view, designed to streamline the management of coding sessions. This new interface allows users to handle multiple agents more efficiently, eliminating the need for juggling numerous terminal tabs or complex setups.
Key Features:
- Start new agents and manage existing ones easily.
- Monitor the status of agents—whether they are waiting for input, still processing, or completed.
- Quickly access the last responses and interaction times for each session.
Users can access agent view by pressing the left arrow from any session or running the command claude agents in the terminal. Each session is displayed in a row, indicating its current state and allowing for quick interaction.
Interactivity: Selecting a session provides a preview of the last interaction. If a decision is required, users can respond inline, allowing the session to resume seamlessly. Direct attachment to sessions for full transcript exploration is also available with a simple press of the enter key.
Moreover, users can incorporate existing sessions into agent view using the /bg command or initiate a new session in the background with claude --bg [task].
Availability: This feature is currently in Research Preview for Pro, Max, Team, Enterprise, and Claude API plans. Users can opt-in by executing the claude agents command, with standard rate limits applying.
As early adopters explore agent view, it is expected to significantly enhance productivity and session management for developers using Claude Code.
