Claude Code projects run parallel cloud agents, but Anthropic says they'll hit usage limits faster
A project is now something you brief rather than a folder you fill, and each thread inside it is a cloud session on its own branch.
On September 17, 2026, Anthropic rebuilt projects in Claude Code. A project used to be a folder you dropped files into. Now you describe a goal, and Claude splits the work across threads and runs them while you do something else.
Each thread is a full Claude Code cloud session on its own branch and its own copy of the repo, with a coordinator on top directing them. Anthropic also says, in the same post, that projects can reach usage limits faster. A launch that warns you about your own quota surprised me.
- What
- projects in Claude Code, redesigned
- Who gets it today
- select Claude Pro and Max subscribers who use cloud sessions in Claude Code
- Over the coming week
- more Claude Code users on those plans
- After that
- all of Claude, and Team and Enterprise plans
- Where threads run
- in the cloud today, on your machine coming very soon
How the work gets split
You pick a goal and a repo, and Claude opens with work it can pick up right away. The example in the post is a migration (Anthropic's own, not a hypothetical of mine): connect your API, web and mobile repos, set a goal to retire a deprecated v1 endpoint, and Claude creates a thread per repo to migrate the callers, run the tests and open PRs.
Threads can split their own work further using subagents, loops and workflows. Two threads touching the same code get no special handling.
The coordinator keeps work organized, but if any threads work on the same code, the overlap is resolved as a merge conflict just like any other PR.
Turns out the job title Anthropic reaches for is chief of staff. You brief the project, and it routes work to new or pre-existing threads, checks in, and follows through. You can steer it from your phone, and it keeps working after you step away from your computer.
What it remembers
Every thread adds to and draws from a shared memory, which Anthropic says cuts the need for complex prompt engineering. Its examples are the things you'd otherwise retype every session: that "the release moved to Friday", or "who to check in with before touching the billing service".
It also remembers how you like to be handled. You can ask it to change how often it checks in, how frequently it starts new threads, or how detailed to make each update (that last one is the setting I'd reach for first). Alongside memory there's a library collecting the files you add and the artifacts Claude produces.
What it costs in usage
Several threads at once, each a full Claude Code session, means a project burns quota faster than a chat does. Granted, Anthropic gives you the dials: you can check usage for a project on its own, and pick the model and effort levels for the coordinator and the worker threads separately.
Threads run in the cloud today, and Anthropic says running them on your machine, alongside your local tools and behind your network, is coming very soon. Existing projects on Pro and Max keep working as they do now. They get upgraded when the rollout reaches chat and Cowork.