Projects now coordinate multiple cloud sessions
Anthropic has rebuilt Claude Code Projects around a coordinator that can scope a goal, delegate work and route it across parallel threads. Each thread is a full Claude Code cloud session working on its own branch and copy of the repository, while the main project chat provides a place to monitor or steer the overall job.
With a repository connected, individual threads can run tests and open pull requests. A thread can also split delegated work further through subagents, loops and workflows. Anthropic says the coordinator reviews outputs and assembles the result, but overlapping edits still meet ordinary Git merge conflicts rather than receiving a separate conflict-resolution mechanism.
Shared memory and a project library carry context forward
Every thread now contributes to and reads from shared project memory. Anthropic presents this as a way to retain project decisions, changing release dates, service ownership and preferred working style across work that spans more than one conversation.
Projects also include a library for uploaded files and artifacts produced by Claude. Users can configure the cloud environment, connectors, plugins, instructions and model at project level, then inspect or redirect individual threads from the main project chat, including from a phone.
The beta starts with selected Pro and Max users
The redesigned experience is initially available to selected Claude Pro and Max subscribers who use cloud sessions in Claude Code and do not already have projects on the web or desktop. Anthropic says access will expand to more Pro and Max Claude Code users over the coming week, with the updated project model across the rest of Claude and Team and Enterprise plans later.
Existing Pro and Max projects continue to work in their current form during the rollout. Anthropic says local threads that can run on a user machine, alongside local tools and code and behind a private network, are coming soon; the threads announced on 17 September run in Anthropic's cloud.
Parallelism trades convenience for more usage and coordination cost
The product change moves Claude Code closer to a persistent multi-agent workspace: a developer can give one project a longer-running goal and let separate sessions work concurrently while sharing accumulated context. That can reduce manual handoffs for work spread across repositories or independent tasks.
Anthropic also warns that projects can reach usage limits faster because several threads may run at once and every thread is a full Claude Code session. Users can inspect project-specific usage and choose the model and effort level separately for the coordinator and worker threads. The beta announcement does not provide a separate project price, concurrency guarantee or independent measure of how reliably the coordinator resolves long-running multi-thread work.