Anthropic overhauls Claude Code's Projects to coordinate multiple cloud AI agents

Anthropic has revamped the Projects feature within its Claude Code tool, turning it into a hub where users can manage multiple AI agents working together. The updated version lets a single project run several "threads" in parallel, each tackling a different task while sharing memory, goals, and a library of files and artifacts. A "coordinator" oversees the overall work.

The approach echoes other tools built to manage groups of AI agents, such as Grok Bot.

How it works

Each thread is actually a separate Claude Code cloud session, operating on its own branch and copy of the code repository. The coordinator keeps everything organized, but if two threads happen to edit the same code, the resulting overlap is resolved as a merge conflict, just as it would be in any ordinary pull request. When a task is large, a thread can break it down further into subagents, loops, and workflows so the work finishes faster.

Users can interact with individual threads directly, or oversee and steer the whole project through a main chat window.

Availability

For now, all threads run in the cloud, though Anthropic says support for local tools and code is coming "very soon." The refreshed Projects feature launched today in beta for "select" Claude Pro and Max subscribers, with plans to later expand access to all Pro, Max, Team, and Enterprise users, as well as to Cowork and standard Claude chats.