Claude Code Desktop Version Revamps Session Management: AI Automatic Task Allocation, Stalled Tasks Centrally Displayed

Beating AI Newsflash, Anthropic has updated the Claude Code desktop version, focusing on improving the management experience when multiple agents work simultaneously. Now, when Claude handles several tasks in parallel, it can automatically create nested sessions, modify code, run tests, and report back to the main session upon completion. For example, while fixing webpage layouts, Claude may identify other issues like image loading or shopping cart problems and proactively suggest follow-up tasks. Users can confirm with one click, enabling multiple sessions to start simultaneously without the need to create them one by one, and avoiding cluttering the main conversation with execution logs. A more practical change is task status management. The new sidebar categorizes sessions into three groups: "Needs Intervention," "In Progress," and "Completed." Users can easily see which tasks are still running and which are awaiting confirmation without repeatedly switching windows. Each session also includes a real-time to-do list and phase summary, making it convenient to take over at any time. Claude can also query the progress of other desktop sessions, read results, and send messages. The process for creating new sessions has been simplified as well. Users can open recently used folders with one click, saving multiple code repositories, SSH hosts, or cloud configurations as commonly used environments. Different sessions can also use independent Git Worktrees to avoid overwriting each other's files during parallel modifications. However, the core of this update is not the first implementation of multi-agent collaboration. Claude Code previously supported parallel sessions, Worktrees, and bypass queries, and similar automatic task allocation and task boards have already appeared in other AI programming tools. This update primarily tightens the integration of task splitting, session initiation, progress tracking, and human intervention, making it easier for developers to manage multiple tasks simultaneously. The new interface has also sparked controversy, with some users complaining that the original project grouping has been disrupted, making it more cumbersome to check context usage. [Original Link]