📅 November 13, 2026 • Anthropic
Anthropic has updated Claude Code Projects, and the headline change is simple: the project coordinator is now proactive. It answers questions directly in the project chat and posts status updates as parallel threads progress, instead of waiting to be polled (per Anthropic ClaudeDevs post on X, November 13, 2026, plus Anthropic docs). Anthropic describes the coordinator as now much more proactive — the September launch framed projects as briefing a chief of staff (per Anthropic).
One conversation, many hands
A project is one long-running conversation where Claude coordinates a stream of related work: Claude scopes the request, delegates work to threads and tracks every thread it started (per Anthropic docs, code.claude.com). Each thread is an independent Claude Code cloud session with its own copy of the repository and its own Git branch. With repositories connected, a thread opens pull requests and runs tests, and overlapping changes surface as standard Git merge conflicts (per Anthropic docs).
Where you can use it
Projects run on claude.ai/code, the desktop app and the Claude mobile app — not in the terminal CLI, the VS Code extension or the JetBrains plugin (per Anthropic docs). Availability: a public beta rolling out gradually on Pro, Max and Ultra plans, with a waitlist for access, and not offered on Team or Enterprise plans yet (per Anthropic docs). The public beta of the redesigned Projects experience started September 17, 2026 (per Anthropic announcement, claude.com).
Defaults and limits
A new project runs Opus everywhere: high effort for worker threads, low effort for the coordinator conversation (per Anthropic docs). Threads draw on the same plan limits as other Claude Code sessions, and several parallel workers can exhaust them faster than a single session — a thread that hits a usage limit pauses and resumes automatically (per Anthropic docs). Cloud threads keep working after the laptop closes, and when a task genuinely needs the local machine, the project can run that thread on the user's computer through Remote Control (per Anthropic docs).
What AlphaSignal reports on top
AlphaSignal reports expanded controls: read-only sharing links for individual threads, with threads that accessed a local computer excluded, scheduled thread reminders that hide a thread until a set time, a /btw command that opens a side thread for tangents, a /fork command that branches a new thread from the current one, a recents row and artifact thumbnails in the threads panel, and a limit of 200 new threads per day across all projects (per AlphaSignal). Anthropic's own post did not detail these items, so treat this list as reported by AlphaSignal only, not as Anthropic-confirmed.
IMPACT
For engineers on Pro, Max and Ultra plans, this moves Projects toward a real orchestrator pattern: one coordinator conversation, many disposable cloud sessions on isolated branches, with Git merge conflicts as the only synchronization point. Practically, that changes daily planning — you brief the coordinator, let workers fan out and review pull requests instead of babysitting sessions. Watch plan limits first: parallel workers burn them faster, and a thread that hits the ceiling pauses rather than failing loudly.
Teams on Team and Enterprise plans, and anyone who lives in the terminal or IDE, still wait for parity.
Sources: code.claude.com — claude.com — alphasignal.ai — news.sulat.com
Continue the conversation: https://t.me/AIAutomationArchitecture · Bluesky
Top comments (0)