Setting Up Claude Tag in Slack: A Practical Guide

Holographic robot AI teammate joining a Slack-style team thread to delegate and complete tasks in the cloud.

Most teams already live inside Slack all day. Claude Tag turns that same workspace into a place where you can tag @Claude the way you would tag a real teammate, hand off a task, and let the work happen in the background. Claude does the work in a cloud sandbox, uses whatever tools it has been given, and posts the finished result back into the Slack thread.

It is useful for research, summaries, content drafts, internal documentation, task follow-ups, and operations work. Setup requires a Claude Team or Enterprise plan, Slack admin permissions, and a few careful configuration choices. This guide walks through what Claude Tag actually does, how to install it, and how to roll it out safely without overspending or exposing the wrong data.

What Claude Tag actually does

Claude Tag connects Claude directly to Slack. Anyone in a connected channel can mention @Claude, describe a task, and watch Claude work in a cloud sandbox hosted by Anthropic, separate from the chat itself. Claude can pull from connected tools such as Google Drive, GitHub, Notion, Jira, Salesforce, or HubSpot, depending on the permissions the admin has granted. It posts a progress checklist as it works, accepts mid-task steering from anyone in the channel, and returns the result to the thread when it is done.

In practice this turns Claude from a chat assistant into a cloud-based AI teammate. It is not magic, though. Claude still needs clear instructions, a defined scope, and human review on anything that matters.

What you need before you start

Before installing anything, line up the following:

A Claude Team or Claude Enterprise plan. Claude Tag is not available on Free, Pro, or Max individual plans.
Slack workspace admin permissions, since most workspaces require an admin to approve new apps.
Permission to install Slack apps in your workspace.
Access to the Claude Tag admin settings at claude.ai/admin-settings/claude-tag as a Claude organization Owner or Admin.
A dedicated test channel where Claude can run without touching live client or production data.
A clear idea of which tools Claude should be allowed to use, and which it should not.
Funded usage credits on the Claude Team plan, since Claude Tag will not run without them.
A short internal policy covering what employees can and cannot delegate to Claude.

Step-by-step setup

Step 1: Install Claude in Slack

Go to claude.com/claude-for-slack and click Add to Slack. Confirm the correct Slack workspace. Most workspaces require an admin to approve the app, so Slack admin permissions are effectively required. An earlier, more limited “Claude in Slack” app existed before Claude Tag, so make sure the team is installing the current product.

Step 2: Connect your Claude account and tools

Open the Claude app inside Slack, then open the Claude Tag admin settings in your browser. Connect only the tools Claude actually needs for the work you plan to delegate. There are more than twenty pre-built connectors, including Google Drive, Gmail, Google Calendar, Notion, Confluence, GitHub, GitLab, BigQuery, Snowflake, Linear, Asana, Jira, HubSpot, Salesforce, Datadog, and Sentry, plus custom tools and custom MCP servers via Connect another tool. Slack channel history, web search, and sandboxed code execution work without any extra credentials.

Warning: do not hand Claude access to sensitive systems, finance tools, or client data during initial testing. Start with read-only or low-risk tools, then expand once you trust the setup.

Step 3: Pair Claude Tag with Slack

From a Slack workspace admin account, post @Claude connect as a new top-level message in a channel. Claude will reply with a single-use pairing code that expires in 15 minutes. Open claude.ai/admin-settings/claude-tag in the browser, click Set up, and paste the code into the wizard.

The wizard will ask for:

A channel scope: either Whole workspace or Specific channels. Choose Specific channels and start with one test channel, not the whole workspace.
An access bundle, which is a named set of tool credentials attached to that scope. Different channels can have different bundles, so an engineering channel can have GitHub access while a marketing channel only has document access.
A spend limit for the scope.

Click Launch to finish the connection. Remember that configuration changes such as new tools or new settings only apply to new threads. An existing thread keeps the connections it started with, so test changes in a fresh thread.

Step 4: Lower costs by setting the model

Under Customize at claude.ai/admin-settings/claude-tag, set the default model for the scope. Sonnet is a sensible default for most business work such as summaries, drafts, research, and internal documentation, and it costs meaningfully less per token than Opus. Reserve Opus for high-stakes or genuinely complex tasks.

Users can still switch models for a single thread by simply asking, for example @Claude use Sonnet for this. The reply footer shows which model answered, which makes it easy to audit.

Step 5: Run a test in a dedicated channel

Tag @Claude with a simple, real task. A good first prompt is:

@Claude please summarize this thread, identify the open questions, and create a short action list for the team.

Confirm that Claude responds, uses the right context, touches only the expected tools, and produces output a human would actually use. Review the result before granting access to more channels.

Pro tip: review spend, tools, and memory after the first run

After the first few tasks, open the Claude Tag admin pages and check three things:

Token spend per channel. The breakdown lives at claude.ai/admin-settings/usage/claude-tag, and the limit resets each billing period. Work done in channels bills to the organization’s shared usage credit balance, not to individual seats. Direct messages with Claude bill to the sender’s own seat instead.
Which connected tools were actually used. Drop the ones that are not pulling their weight, and tighten the access bundle if Claude is reaching into systems you did not expect.
The memory Claude has saved. Claude Tag keeps persistent, channel-scoped memory. Memory from public channels is shared across the workspace, while memory from private channels stays in that channel. Anyone in the channel can ask @Claude what do you remember about this channel? and can correct or remove entries by talking to Claude. Organization Owners can edit or delete a scope’s memory in admin settings, and Admins can view it.

If multiple teams use Claude Tag, schedule a regular memory review. Persistent memory is genuinely useful, but it should be monitored.

Safety and permissions

Claude Tag is powerful, so a few practical guardrails go a long way:

Start with limited channel access and expand slowly.
Avoid connecting sensitive tools at first, especially anything that writes to production systems.
Do not give Claude access to private client data without an approved internal policy.
Write simple internal rules for what employees can delegate, and which channels Claude is allowed in.
Keep a human in the loop on any output that goes to a customer or affects revenue.
Monitor usage and costs weekly at first, then monthly once the spend pattern is clear.

Two guest-related behaviors are worth flagging. Slack channels shared across multiple workspaces are not supported. In channels that include guests, Claude stays silent unless the admin enables Allow Claude to respond to guests.

Best use cases

Claude Tag fits the kind of work that already happens in Slack threads:

Summarizing long Slack threads into decisions and action items.
Capturing meeting follow-ups and assigning next steps.
Drafting client updates from internal discussions.
Turning open conversations into structured task lists.
Researching internal questions across Drive, Notion, Confluence, and the web.
Creating SOPs from Slack conversations.
Producing first drafts of blog posts, emails, and reports.
Running project status reviews across connected tools.
Preparing sales or support responses from prior context.
Organizing team knowledge into searchable documentation.

Example Slack prompts

These prompts are designed to be copied and pasted into a Slack thread:

@Claude summarize this thread and list the decisions made.
@Claude create a task list from this conversation and group it by owner.
@Claude turn this discussion into a client-friendly update.
@Claude review this idea and identify risks or missing details.
@Claude draft an SOP based on the process described in this thread.
@Claude create a blog outline from the points above.
@Claude compare the options discussed here and recommend the best one.
@Claude find the open questions we still need to answer.
@Claude rewrite this into a professional email.
@Claude create a concise executive summary of this thread.

Troubleshooting

When something goes wrong, the cause is usually one of the following:

Claude does not respond. The app may not be installed or approved, the channel may not be in the scope, Claude Tag may be switched off for that scope, or guests in the channel may be blocking the reply.
Pairing code expired or already used. Codes are single-use and expire in 15 minutes. Post @Claude connect again and paste the new code promptly.
Only admins can run @Claude connect. This is by design. A regular member will not be able to complete pairing.
Usage credits not funded. On Team plans, the organization must fund usage credits before Claude Tag will run.
New tool connections only apply to new threads. Start a fresh thread to test new settings.
“Still waiting for available capacity” is transient and resolves by retrying in the same thread.
Costs higher than expected. Check the default model and the per-channel spend page at claude.ai/admin-settings/usage/claude-tag.
Claude using more tools than expected. Tighten the access bundle for that scope.
Memory needs review or deletion. Owners can edit or delete memory in admin settings, and channel members can correct it by talking to Claude.

Cost control

Cost control is mostly about choices made before launch. Model choice matters: Sonnet is significantly cheaper than Opus and is strong enough for most everyday work. Monitoring matters too: check token spend after the first few runs at claude.ai/admin-settings/usage/claude-tag, and review the per-channel breakdown to catch outliers early.

Three habits keep the bill predictable. Use a dedicated test channel so early experiments do not burn through credits. Expand from one channel to more channels slowly, and set a spend limit before launch, not after the first surprise.

Conclusion

Claude Tag makes Slack more productive by letting teams delegate work where the conversation already happens. The best setup is careful and controlled: start with one channel, connect only the tools you need, set Sonnet as the default model, set a spend limit, and review spend, tools, and memory after testing. Once configured properly, Claude becomes a useful teammate for summaries, documentation, follow-ups, content, and internal operations, working in the background while the team stays focused in the thread.

Setting Up Claude Tag in Slack: A Practical Guide for Business Teams