Skip to content

Anthropic

Hub testing in Claude adds a switch-back feedback prompt

2 min read

Written by an AI model from the linked sources and checked against them by a second model. How we work

Image: threads.com

Anthropic is still testing Hub internally in Claude, and the latest build now asks for feedback when someone leaves it and switches back to a normal chat, TestingCatalog said in a post on Threads on August 28. The same build moves the Connected Devices option straight into a tool widget, where it is easier to reach. Anthropic has not announced Hub publicly.

The feedback prompt only makes sense if Hub is a separate mode that replaces the usual chat. TestingCatalog calls it presumably a task orchestrator and thinks Anthropic may plan a broad release. Since mid-August the working theory has been that Hub lets Claude coordinate several sub-agents and show their progress in one interface.

In the Claude mobile app, Hub Mode is marked as internal and sits next to a planet icon. A similar icon appeared earlier under the name Orbit and was then removed. The mode’s description says it applies to new tasks and that the app has to be restarted after you turn it on. When TestingCatalog enabled it in its August 18 report, it saw no visible changes.

The likely engine underneath is already shipping. Claude Opus 4.8 introduced Dynamic Workflows, which let Claude orchestrate up to 1,000 sub-agents running in parallel: Claude writes a JavaScript orchestration script, and a runtime executes it in the background for codebase-scale tasks. Memeburn says those workflows stay hidden from the user, who sees the output but not the individual agents, and that Hub Mode could expose that orchestration layer. The outlet also points to Cowork’s expansion to mobile and web in July 2026, saying more than 90% of Cowork sessions aren’t about coding.

Moving Connected Devices into the tool widget led TestingCatalog to ask whether Hub might also orchestrate tasks across different devices. It’s an open question, raised in the post itself. The outlet has also compared the possible feature to Grok Bot, where users can create several specialized agents, and said it would fit Anthropic’s push toward agent-based workflows across Claude Code, Cowork and collaborative projects.

Hub’s only related appearance on the web is in Claude Code. There, a Hub destination opens a list of sub-agents, and selecting one leads to a page saying Claude will generate a task board for the user and their agents. In its earlier report, TestingCatalog said nothing confirms that this Claude Code page and Hub Mode in the mobile app are the same feature.

No comments yet. Be the first.

Join the conversation

Sign in to leave a comment. It appears after moderation.