Slack MCP Client: Use Any MCP Server Directly in Slack

Are you looking to bring Model Context Protocol (MCP) tools to your Slack workspace? Runbear is an MCP client for Slack, so your team can use an AI agent with MCP tools directly where your team already works. No coding required. For how this differs from Slack's own MCP server, see our Slack MCP overview.

What is MCP and Why You Need a Slack MCP Client

Model Context Protocol (MCP) transforms AI assistants from simple conversational tools into powerful agents that can take action. With a proper MCP client in Slack, you can:

• Access external data sources and services
• Connect to workplace tools seamlessly
• Complete tasks independently
• Deliver information directly to your team

Introducing Runbear: A Hosted MCP Client for Slack

Runbear connects your agent to MCP servers from Slack, so your team uses AI tools without local installations. Everyone in the channel works with the same agent inside your existing Slack workspace.

Real-World Slack MCP Client Use Cases

Example: Calendar Scheduling

With a calendar MCP server connected, ask the agent to look up open meeting slots without leaving Slack.

Example: Email Summaries

With an email MCP server connected, ask the agent to summarize a long email thread right in your Slack channel.

Simple Steps to Set Up Your Slack MCP Client

1. Create your Runbear account: Sign up and create an agent.
2. Choose a model and write instructions: On the default Claude Agent SDK runtime, pick Claude, GPT, or Gemini in the Settings tab, then describe the agent's job in the Instructions tab.
3. Connect MCP servers: In the agent's Tools tab, pick a server from the managed catalog and sign in to it, or choose Add Custom MCP to add your own remote server by its HTTPS URL.
4. Connect it to Slack: Connect your agent to your Slack workspace with a few clicks, and enjoy!

Connecting accounts to AI agents

There are two ways to connect service accounts to your AI agent's MCP servers: shared and individual authorization.

Option A: Per-User (individual) authorization

Each Slack user connects their own account. When a user requests something that requires authorization, the Slack bot will prompt them with a secure link to connect.

No worries: connected accounts are private and accessible only to the individual user.

Option B: Shared authorization

Connect a shared account that your AI agents can use to answer questions and complete tasks.

With this setup, users won't need to authorize access individually; the AI agents will handle requests using the shared connection. Every user of the agent inherits that account's access, so grant it deliberately.

Transform Your Slack Workspace with an MCP Client Today

Runbear brings AI agents with MCP tools to everyone on your team through Slack. Start free and see what an agent with MCP tools can do in your Slack workspace.

Slack MCP Client vs Slack MCP Server

The MCP client is the application that sends requests to MCP servers. Runbear acts as the Slack MCP client, sitting between your team in Slack and the MCP servers that provide tools and data. Your team talks to the client (Runbear in Slack), and the client talks to the servers on their behalf. Slack also runs its own MCP server, which works in the opposite direction: it lets AI clients outside Slack read and act in your workspace. Our Slack MCP overview explains both.

Without a hosted MCP client, each person would need to install and configure MCP servers on their own machine. Runbear removes that requirement by hosting the client connection and making the servers connected to an agent available to everyone who uses that agent in Slack (on Per-User servers, each person connects their own account the first time).

Frequently Asked Questions

Do I need to install anything locally to use MCP in Slack?

No. Runbear handles the MCP server connections remotely. Your team just uses Slack as normal, and the AI agent connects to MCP servers behind the scenes. There is nothing to install on any team member's computer.

Which MCP servers work with Slack through Runbear?

Runbear's managed catalog includes servers such as Apollo.io and Attio, and hundreds of further integrations are reachable through its wider connector catalog in the same Tools tab. You can also connect any remote MCP server, including ones built by your engineering team, as a custom server over Streamable HTTP or SSE.

Is MCP data shared between Slack users?

It depends on your configuration. Individual authorization keeps each user's connected accounts private and separate. Shared authorization lets the AI agent use one shared account for the whole team. You can mix both approaches for different MCP servers.

How is this different from Slack's built-in AI?

As of October 2026, Slack's developer docs describe a Slackbot MCP Client that lets developers connect remote MCP servers to Slackbot through a Slack app. Runbear is a team agent you configure in a dashboard, with its own instructions, knowledge sources, and tools, and it runs in Slack and Microsoft Teams.