What matters in AI.

Subscribe

Claude Code Projects tells developers when a thread is blocked

Each worker uses the Claude Code allowance of the subscriber, and one account can start 200 new threads each day.

Anthropic changed Claude Code Projects, a beta tool in which one coordinator gives work to some Claude Code sessions in the cloud. At this time the coordinator writes status updates in the project chat. A developer can send one thread by a read-only URL. The beta opened on 17 September for Pro and Max subscribers on the waitlist. One account can start 200 new threads each day.

How it works

Each thread is a cloud session with a copy of the code and a Git branch. The coordinator reads the reports of threads and controls the project. Changes to the same code show as merge conflicts when developers put branches together.

The numbers

Projects have no more cost. Each worker uses the Claude Code allowance of the subscriber. New projects use Opus for the coordinator and for workers. The coordinator uses low effort and workers use high effort. Anthropic gave no test results.

What is not known

Team and Enterprise plans cannot use Projects. Projects operate only in the cloud and only with Claude Code. Developers must put the pull requests of connected tasks in a sequence and must remove merge conflicts.

Insight: Parallel agents move the limit to the merge step

Projects send parallel work back through Git branches and merge conflicts. Anthropic tells that it merges more than 30,000 code changes each day, up from about 2,500 in May 2026. It also moved its code from Git to Scion. This reads as a sign that the merge step is the new limit, and the tool that controls merges gets control. Monitor if Projects adds Scion.

Sources

  1. Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-Freealphasignal.ai
AI MATTER · NEWS · AI MATTER · NEWS ·14 NOV2026

Posted

Tags

More in Agents

All Agents news