The Slack MCP Server, developed by Korotovsky, is a Model Context Protocol (MCP) server designed to enable Large Language Model (LLM) agents to interact with Slack Workspaces. Built primarily with Go 1.21 or newer, this free MCP server acts as a bridge, allowing AI clients to perform various actions within Slack. It supports multiple transport protocols and authentication methods, making it adaptable for different development needs. Engineers and developers building LLM agents, especially for hobby projects or single-workspace tools, will find this server valuable for integrating AI capabilities into their Slack environments.

What does Slack MCP Server connect to?

The Slack MCP Server functions as a Model Context Protocol (MCP) server, specifically connecting LLM agents to Slack Workspaces. Its primary role is to enable these agents to interact directly with Slack, performing actions like reading messages, posting new content, and managing reactions. It supports various AI clients, including Claude Desktop, Cursor, Perplexity, and Copilot, alongside custom agents. The server is designed to be compatible with any MCP client that utilizes Stdio, SSE, or HTTP transport protocols, ensuring broad applicability for developers integrating AI with Slack.

Core features and tools

The Slack MCP Server exposes a specific set of callable tools for LLM agents to interact with Slack. These include conversations_history, conversations_replies, conversations_add_message, and conversations_search_messages for managing conversational content. Agents can also use channels_list to retrieve channel information, and reactions_add and reactions_remove for managing message reactions. Further capabilities include attachment_get_data for accessing attachment content and conversations_unreads for checking unread messages. These tools provide granular control over Slack interactions.

Who is it for?

This server targets engineers and developers focused on building LLM agents, particularly for hobby projects and single-workspace tools. It provides the necessary interface for AI agents to perform actions such as reading and searching Slack messages, posting new messages, listing channels, and getting unread messages. Its design supports both stealth and OAuth authentication, offering flexibility for different development needs. Skip this if your project requires a multi-workspace enterprise solution or does not involve LLM agent integration with Slack.

How to get started

Access is granted through a standard OAuth authorization flow, or by using classic bot tokens or browser session tokens for Stealth Mode. Authentication can involve setting the SLACK_BOT_TOKEN environment variable for bot tokens, and it supports Stdio, SSE, and HTTP transport protocols.

Use cases

The Slack MCP Server supports several concrete workflows for LLM agents:

• Reading and searching Slack messages across channels, direct messages, and threads.
• Posting new messages to specific conversations.
• Listing available channels and users, and retrieving unread messages across channels.

These capabilities enable agents to monitor discussions, contribute information, and manage their interaction context within Slack environments effectively.

The Slack MCP Server by Korotovsky offers a direct way for LLM agents to engage with Slack Workspaces. Its support for various authentication methods and transport protocols makes it adaptable for developers. The server provides a focused set of tools for managing messages, channels, and reactions, catering to specific integration needs.