Anthropic Redesigns Claude Code Projects With Cloud AI Orchestration, Local Developers Wait

The capability is initially being offered to select Claude Pro and Max subscribers who use cloud sessions in Claude Code and do not have existing projects on the web and desktop, Anthropic said, according to ZDNet. Features will roll out to other users over time, but the company said Pro and Max subscribers already using projects will keep them as they are for now and be upgraded later. That initial scope excludes terminal and local-file workflows, ZDNet reported. Anthropic said threads run in the cloud today, while running on a user's machine alongside local tools and code and behind their network is coming very soon. The company also said it is not taking away the terminal and described the new projects capability as an additional feature, with no negative changes indicated for local or command-line functionality.

For cloud users, the redesigned Projects feature is meant to coordinate multi-agent work. Anthropic said Claude scopes the request, delegates the work, coordinates parallel threads, reviews the outputs and assembles the finished result. Users can steer progress throughout, and the system keeps working after the laptop is closed, ZDNet reported. Mobile check-ins are on the way, Anthropic added.

The orchestration layer goes beyond prompting. Each discussion becomes a full cloud session with its own branch and copy of the repository it is working on. A coordinator provides instructions that route sessions into existing threads as appropriate. Anthropic advised users to brief Claude on a project the way they would brief a chief of staff, and it routes work to new or pre-existing threads.

Threads are named, so users can intervene mid-flight, open a running thread and redirect it instead of waiting for completion. Threads also can be aware of each other. Anthropic said every thread adds to and draws from a shared memory, reducing the need for complex prompt engineering. This automated memory can let work span repositories, allowing a user to work across several repos at once, tie a thread to each and get feedback on which pull requests need to be merged in what order.

ZDNet reported that the architecture is not entirely new. Anthropic said the new projects use subagents internally, and Claude has been able to split work across subagents. The difference is that the split persists, and users can return to each subagent's discussion and chat about it. The feature does not appear to be strictly autonomous: threads begin when the user hits Start for each thread Claude recommends.

Existing Claude Code projects have been tied to a local directory as the root for source code, with separate directories isolated from one another. The redesign adds coordination for cloud sessions, but local workflows remain outside the new system for now. Anthropic said the features will reach more users over time.