Anthropic on September 17 unveiled redesigned Claude Code Projects, a beta workspace intended to coordinate several AI coding sessions on a longer-running job. The product gives users a central project conversation, shared project context and parallel worker threads, rather than asking them to manually juggle separate Claude Code sessions.
According to Anthropic, a user starts by choosing a goal and supplying a repository or other context. A coordinator then assigns work to new or existing threads. Users can direct the overall effort in the main project chat or open an individual thread to inspect and steer that task. Anthropic says connected repositories can let threads run tests and open pull requests.
How do Claude Code Projects coordinate multiple agents?
Each worker thread is a Claude Code session running in Anthropic’s cloud, with its own branch and copy of the repository. Threads can also subdivide work through subagents, loops and workflows, Anthropic says. The coordinator is the layer that tracks and routes the overall assignment while those sessions run in parallel.
The arrangement does not make concurrent code changes magically harmonious. Anthropic says that if threads edit overlapping code, the result is a conventional merge conflict, resolved the same way it would be in an ordinary pull request. Separate branches and repository copies are part of the implementation, not a promise that parallel agents will avoid conflicting edits.
Projects also maintain shared memory across threads and a library for files supplied by users and artifacts Claude produces. Anthropic says that context can retain project details and prior decisions over time, so later work can draw on them. The company says users can set the project’s cloud environment, connectors, plugins, instructions and model, and adjust how often Claude checks in or starts new threads.
Who can use the redesigned Claude Code Projects beta?
At launch, the beta is limited to select Claude Pro and Max subscribers who use Claude Code cloud sessions and do not already have Projects on Claude’s web or desktop apps, Anthropic says. The company plans to extend access over the following week to more Claude Code users on those plans, followed by the redesigned experience across Claude and its Team and Enterprise plans. Existing Projects on Pro and Max are to keep operating as before during the rollout.
The launch also has a fairly plain constraint: threads run in the cloud. Anthropic says support for work on a user’s machine, including local tools, local code and operation behind a network, is coming soon but is not available at launch. That may be limiting for work tied to local development environments.
There is another operational cost. Because every parallel thread is a full Claude Code session, Anthropic says Projects can hit usage limits faster. Users can view project-specific usage and select model and effort settings for both the coordinator and worker threads. The product is therefore a coordination layer for multi-session development work, not a substitute for resolving the usual collisions created when several changes touch the same code.
This story draws on original reporting from The Verge.
