Anthropic's redesigned Projects turns Claude Code into a coordinator running parallel AI agent teams
新智元 · wechat · 2026-09-18
Anthropic announced a redesigned Projects feature (Beta), landing first in Claude Code: you brief one project conversation, and Claude acts as a coordinator, splitting work into multiple cloud threads that run in parallel—continuing while your laptop is closed, with progress viewable from your phone.
- Two-layer architecture: the project conversation is the coordinator; each thread is an independent cloud Claude Code session with its own context, Git branch, and code copy. Conflicts merge like normal PRs, and threads can spawn subagents.
- Self-managed PRs: threads watch CI, push fixes, and address review comments before notifying you; an Overview panel shows what's done and what needs your review.
- Shared memory: project context, decisions, and your working habits are stated once and inherited by all new threads; artifacts (even non-code like contracts or support tickets) persist in a project Library.
- Costs and limits: worker threads run Opus at full reasoning intensity and share your regular quota—Pro users will hit limits sooner. No hard thread cap, but 200 new threads/day across all projects. Cloud threads can't touch local files, intranets, or VPN-protected services.
- Industry context: Cursor shipped an almost identical Projects beta a week earlier. The same day, Anthropic published a report saying 30,000 AI agents run concurrently on its internal R&D platform, making 1B+ decisions monthly, with 26% of R&D work at EpochAI's AL4 level (up from under 1% six months ago)—right after CEO Dario Amodei urged the industry to "slow down."
More from AGI Musings
- Dev says blog traffic fell from 700k to 7k reads as AI-generated job apps flood market — ericelliott_ · 2026-11-02
- Master foundational math to stay irreplaceable, argues mathematician — TivadarDanka · 2026-11-02
- Narayanan: labs racing toward Recursive Self-Improvement may hit Recursive Self-Extinction of their business model — random_walker · 2026-11-02
- Editor fully resigns from JMLR, arguing AI reviews beat rushed humans and journals need a rethink — roydanroy · 2026-11-02
- Alpha School investigation: 17 campuses on probation, Bill Ackman as shadow lead, Holmes involved — benjaminjriley · 2026-11-02
- Google's chief economist: AI will transform more jobs than it kills — gains come from redesigning workflows — danielrock · 2026-11-02