Anthropic has introduced a proactive coordinator to its Claude Code Projects, a feature designed to manage multiple parallel AI workers without requiring constant manual oversight from developers. This update allows a central coordinator to assign tasks to independent threads, each running as a separate cloud session with its own repository copy and Git branch. By automating status updates and decision-making, the system aims to simplify the management of large-scale engineering workflows like performance profiling and cross-repository migrations.
Automated Coordination and Thread Management
The primary change in the update is the shift toward a proactive coordinator. Instead of developers manually checking a list of sessions to see which are waiting for input, the coordinator now pushes pending decisions and completed task reports directly into the main project conversation.
Each thread in a project operates as an independent Claude Code cloud session. When these threads complete work, overlapping code changes are surfaced as standard Git merge conflicts, allowing developers to reconcile different branches of work systematically.
“The coordinator directs the project from thread reports, while overlapping code changes surface as standard Git merge conflicts when branches are combined.”
— Anthropic Upgrades Claude Code Projects to Run Parallel AI Workers Hands-Free | AlphaSignal
New Workflow Controls and Navigation
Anthropic added specific commands to handle complex project contexts. The /btw command opens a side thread for tangents, while /fork creates a new thread from the current state to test alternative implementations or debugging paths.
To improve navigation across large projects, the threads panel now features a 'recents' row and thumbnails for generated artifacts like diagrams and documents. Reminders also allow users to 'park' a thread and have it resurface in the project chat at a scheduled time.
This screenshot demonstrates the specific navigation and management features mentioned in the article, such as parking threads and branching.
Usage Limits and Availability
While Projects do not carry a separate fee, every worker thread consumes the subscriber's existing Claude Code allowance. Anthropic limits accounts to 200 new threads per day across all projects, and threads that hit their usage limit will pause and resume automatically when capacity is available.
The update is currently in public beta for Pro and Max subscribers across web, desktop, and mobile platforms. Notably, Team and Enterprise plans are excluded from this specific rollout at this time.
What Hacker News readers are saying
- Technical note: A representative from the project team explained that the update was motivated by users running multiple parallel agents, which creates a difficult user experience for managing context and complexity. comment ↗
- Technical note: One commenter who has built similar systems described their own approach of using 'keepers' for each microservice to manage domain-specific tasks and self-compacting context. comment ↗
- Skeptical: Several commenters expressed concern that these features might primarily serve to increase API usage and costs for the user, while others questioned the long-term impact on individual agency. comment ↗
- Skeptical: One user raised questions about the practical returns of parallel management, noting that the primary bottlenecks remain in providing sufficient project detail and predicting token costs. comment ↗
Summarized opinions from the Hacker News thread, not verified facts. Read the full discussion ↗