XNews

Claude Code Projects enables parallel cloud agents

Anthropic's Claude Code introduced Projects, a feature that lets one conversation spawn multiple cloud-based agent threads that work in parallel, even when the computer is off. Each thread runs as an independent Claude Code session with its own context, sandbox, and git branch, and can handle tasks like coding, testing, and PRs autonomously.

Key insights

Projects enables parallel, persistent cloud agents

Claude Code's new Projects feature allows a single conversation to act as a supervisor, spawning multiple cloud-based threads that work in parallel. Each thread is an independent Claude Code session with its own context window, sandbox, and git branch, and continues running even when the computer is off. Users can delegate tasks, and threads can open PRs, run tests, and request human input when needed.

Shared memory and background execution

Projects threads share a common project instruction, repository, and memory, so decisions made in one thread are known to others. The feature supports scheduled routines, allowing Claude to run tasks automatically. It also works beyond code, enabling document research and report generation, with results stored in a Library.

Source posts