Anthropic relaunches Claude Code Projects to coordinate multi-agent coding in the cloud
The AMW Read
Meaningful Claude Code baseline update—cloud multi-thread agents with coordinator and PR-style merges—raising segment-level pressure on AI-native coding tools without a new structural capital or compute signal.
Anthropic relaunches Claude Code Projects to coordinate multi-agent coding in the cloud
Anthropic has relaunched Projects inside Claude Code so users can run multiple coding agents under one shared workspace. Each project keeps shared memory, goals, and a library of files and artifacts; parallel “threads” handle separate tasks while a coordinator steers the work. Under the hood, every thread is a Claude Code cloud session on its own branch and repo copy, with overlapping edits resolved as ordinary merge conflicts. Threads can further fan out via subagents, loops, and workflows. At launch the feature is cloud-only, with local-tool support promised soon, and beta access starts for select Claude Pro and Max subscribers before widening to Pro, Max, Team, Enterprise, Cowork, and regular Claude chats.
For the AI coding market, this is a frontier lab pushing past single-session autocomplete into orchestrated multi-agent software work—the same competitive lane as Cursor-class IDEs and other agent-team managers such as Grok Bot. Shared project state plus git-native conflict handling treats agent output as reviewable engineering work, not disposable chat. Coming a day after Anthropic’s Chat and Cowork unification, it extends the same product push from general assistants into the developer inner loop.
Builders should evaluate whether cloud-first multi-thread sessions fit their repo workflows and security posture before local support arrives; investors should watch how quickly Anthropic converts Pro/Max beta usage into Team and Enterprise seats against AI-native IDE incumbents.
