Claude Code: AI Agents Get Superpowers

Anthropic has significantly redesigned its Claude Code projects feature, introducing advanced capabilities aimed at managing multi-agent systems more effectively. The revamped projects offer persistent, addressable threads that can communicate and share memory, providing a sophisticated orchestration layer for AI coding agents.

This new functionality allows complex workflows to run in the cloud, with agents coordinating tasks and continuing work even after a user closes their laptop. While initially available to select Claude Pro and Max subscribers using cloud sessions, Anthropic plans a phased rollout to other users over time. Local development workflows are not yet integrated.

Key enhancements include the ability for threads to learn from each other and contribute to a shared memory, simplifying development across multiple repositories. This persistent memory reduces the need for extensive prompt engineering and allows for finer-grained control over ongoing tasks. However, users must still manually resolve merge conflicts that may arise from overlapping threads.