The 'Claude Tag' from Slack is coming to Claude Desktop

What is Claude Tag

Claude Tag is a feature that invites Claude into a Slack channel so any team member can call on Claude mid-conversation. Claude follows the codebase tied to the channel's conversation context, and when mentioned, it responds or picks up work on the spot. In other words, Claude functions not as a personal assistant but as a colleague on the team channel — and until now, this experience was only possible within organizations using Slack.

The news now emerging is that this experience is moving outside Slack. Signs have been confirmed that Anthropic is preparing 'Projects for Claude Code' inside Claude Desktop — a structure where a team shares one workspace within the Claude app itself, without Slack. TestingCatalog, which first spotted this, described it as "essentially Slack built into the app."

What's changing, and how

This builds on the previously spotted 'Managed Projects,' taking it a step further. Three things are changing.

Element	Before	This update
Repository connection	Attached temporarily as needed, per session	Fixed as permanent context at project creation
Session structure	One conversation per project	Multiple threads branch off within one project
Context management	Handled manually by the user	Claude periodically reorganizes it itself

The parallel to Slack lines up precisely. The project session is the channel, individual tasks branch off as threads, and multiple people plus Claude stay together in that space. Since a GitHub repository is permanently attached to the project, there's no need to re-explain the codebase every time, and Claude periodically reorganizes context files and memory on its own — less like a conversation log that's written once and set, and more like a workspace that keeps getting tidied up.

There are signs Anthropic has also been refining the Slack-side Claude Tag around the same time. Noah Zweben noted on X regarding Claude Tag improvements that "unprompted messages are down about 45%" — suggesting Claude has gotten better at judging when to jump in and when to stay quiet. It's a separate feature, but it points in the same direction: Claude embedded in team spaces.

Can you use it now — not yet

The rollout appears to be expanding beyond an early alpha, and some early-access organizations seem to already have access, but there's no schedule yet for general availability. At this point, there's no menu confirmed to be accessible to ordinary users. Still, the direction is clear — Anthropic is pushing Claude from a coding assistant that helps one person toward organizational infrastructure that teams use together.

Editor's take

The core of this feature isn't what's newly built, but what's being removed. Until now, AI coding tools have defaulted to single-player use. One developer would ask and receive answers, while team collaboration was left to tools outside the AI — Slack, GitHub comments, and the like. The moment a repository is permanently attached to a project and multiple people split threads within the same session, that boundary disappears.

Zooming out to the broader market, this overlaps with Cursor's private beta testing of codebase and review tabs — a phase where the center of gravity for AI coding tools is shifting from "personal assistant" to "team workspace." For teams in Korea, it's not yet a stage to put on immediate adoption plans, but there's one thing worth noting: if you're currently re-explaining repository context every single time in the tools you use, that repetition is a cost that's about to disappear. There's a good chance that within the next few weeks, the rollout will expand to the organization level or be officially announced integrated with Slack's Claude Tag.