Anthropic has rebuilt Claude Code Projects around multiple cloud agents working in parallel. A project now holds shared goals, memory, files and artifacts, while a coordinator assigns pieces of work to separate threads and keeps their progress organized.
Each thread is an independent Claude Code cloud session with its own copy and branch of the repository. It can create subagents, loops or workflows for a delegated task. Users may talk directly to an individual thread or manage the whole effort from the main project chat. If two agents edit the same code, their work is reconciled through an ordinary merge conflict rather than silently combined.
The feature is available in beta to selected Claude Pro and Max subscribers. Anthropic plans to expand it to all Pro, Max, Team and Enterprise users, as well as Cowork and standard Claude chats. Threads initially run only in the cloud, with local tools and code support promised soon. Parallelism can shorten independent tasks, but it does not eliminate integration work or guarantee that combined changes are correct. Teams still need tests, review and clear task boundaries, particularly when several agents touch related code or share project memory.