Claude Code Now Has AI Minions to Do Your Homework While You Sleep

Anthropic is pushing Claude Code beyond simple chat by transforming projects into autonomous command centers. Instead of manually pasting code, users define high-level objectives like migrating API services, and the system automatically decomposes the task into parallel execution streams.

Each stream operates in its own dedicated Git branch with a unique repository clone. These sub-agents handle the dirty work: reading code, executing unit tests, and opening pull requests, while a master Claude instance monitors progress and orchestrates the final merge logic. Since the shared memory preserves context across sessions, agents continue to crunch through backlogs in the cloud even after the developer shuts down their machine.

Currently, the feature is rolling out in beta for select Pro and Max subscribers. The entire workflow remains accessible from mobile devices, allowing users to watch their virtual workforce toil away from anywhere.

This marks a transition from AI as a fancy autocomplete tool to AI as a supervisor of its own hallucinations. By treating coding as a management problem rather than a syntax problem, the industry is betting that if you throw enough agents at a bug, eventually one of them will accidentally fix it while the others just create more technical debt to be cleaned up in the next sprint.