Claude Code adds teams, worktrees, voice, 1M context

Hot take: Anthropic is trying to make Claude Code the control plane for serious coding work, not just a smarter shell. The big shift is less about novelty features and more about orchestration: parallel agents, persistent memory, and fewer context resets.

• –Agent teams plus shared memory make the tool feel more like a staffed queue than a solo chatbot.
• –Git worktrees are the right isolation boundary for parallel agents, because they keep experiments from clobbering the main branch.
• –Voice, remote control, and recurring commands make Claude Code feel usable outside the classic terminal typing loop.
• –1M context is the standout upgrade for sprawling refactors, but it will punish sloppy prompts and token-hungry workflows.