Claude Tag is a new Slack integration announced by Anthropic in June 2026. It allows Claude to join team channels as a full member. Users simply mention @Claude to delegate work while the AI maintains full conversation context across the channel and connected tools.
📑Table of Contents
According to the official announcement, 65% of code written by Anthropic’s internal product team is now generated by their internal version of Claude Tag. This demonstrates how deeply the system integrates into daily collaborative workflows rather than functioning as a simple chatbot. It is positioned as an evolution of Claude Code, making the model more proactive and team-oriented. The feature is currently in beta for Claude Enterprise and Team plan customers.
Key Features and Capabilities
The core value lies in the shared Claude instance per channel. Anyone can inspect progress and continue threads, preventing information silos. Context learning accumulates tacit knowledge from channel history and approved data sources without accessing private channels.
With “ambient” mode enabled, Claude proactively flags relevant information and follows up on stalled tasks. Asynchronous execution supports scheduling work across hours or days in parallel. Direct messages are also supported but can be disabled organization-wide. The underlying model is Opus 4.8. Channel activity bills to the organization while personal DM usage may bill to the individual account.
Source: Anthropic Official Announcement (as of June 2026)
Setup and Onboarding Steps
Setup requires four steps. First, pair Claude Tag with your Slack workspace. Next, grant access to selected tools and data sources. Then configure monthly token spend limits at the organization or channel level. Finally, test in a private channel.
The existing Claude app in Slack is a prerequisite. Admins provision an identity and access bundle once at claude.ai/admin-settings/claude-tag. Only organization owners can perform this action. A 30-day opt-in migration path from the legacy Claude in Slack app is available.
Source: Claude.ai Documentation (as of June 2026)
Admin Controls and Security
Admins receive full activity logs with requester attribution. Scoped memories prevent cross-domain leakage between sales and engineering instances. Permissions are tied to the channel rather than individual users. Granular access controls and data-handling policies are configurable from the admin dashboard. Only owners can perform initial setup and define token limits and access scopes to meet enterprise governance requirements.
Practical Use Cases
Common scenarios include reproducing bugs and opening PRs, converting decision threads into documentation, assembling project state snapshots, and investigating incidents. Claude posts visible checklists and keeps the entire exchange inside the thread for full auditability. The beta is particularly valuable for distributed teams and large-scale projects.
Enterprise and Team plan customers can try the beta with introductory launch credits. It runs on the Opus 4.8 model.
Frequently Asked Questions (FAQ)
| Question | Answer |
|---|---|
| Which plans support Claude Tag? | Beta access is limited to Enterprise and Team plans. |
| Is installation required? | No. Simply mention @Claude in a channel where it has been added. |
| How is context maintained? | Claude learns from channel history and permitted data sources. |
| Can it be used in DMs? | Yes, but organizations can disable DM usage globally. |
| How is billing handled? | All channel activity is billed to the organization. |
Sources: Anthropic Official, Claude.ai Documentation
Summary and Future Outlook
Claude Tag turns Claude into a persistent, context-aware team member inside Slack. Its emphasis on asynchronous and proactive collaboration is expected to deliver significant value for distributed teams and complex projects. We will continue to track official Anthropic updates and share practical guidance.
Related articles:
- Omnigent: Databricks Open-Sources Meta-Harness for Multi-Agent Control Across Claude Code and Codex
- Claude Fable 5 Vanished 8 Days After an AI Self-Creation Claim
- How to Fix AI-Generated “Looks Good But Unusable” Diagrams Using draw.io Skills
Author
krona23
Over 20 years in the IT industry, serving as Division Head and CTO at multiple companies running large-scale web services in Japan. Experienced across Windows, iOS, Android, and web development. Currently focused on AI-native transformation. At DevGENT, sharing practical guides on AI code editors, automation tools, and LLMs in three languages.
Leave a Reply