korotovsky/slack-mcp-server

What it solves

This project provides a Model Context Protocol (MCP) server that allows AI models to interact with Slack workspaces. It bridges the gap between an LLM and Slack, enabling an AI to read messages, search history, manage user groups, and post messages or reactions within a workspace.

How it works

The server acts as an intermediary that connects to Slack via various authentication methods (OAuth or "stealth mode" using browser session tokens). It exposes a set of tools and resources to an MCP-compatible client, allowing the AI to perform actions like fetching channel history, searching for messages, and managing "Save for Later" items. It supports multiple transport layers (Stdio, SSE, and HTTP) and can be configured with environment variables to enable or disable specific high-risk actions like posting messages.

Who it’s for

This is for developers and users of MCP-compatible AI clients who want to give their AI assistants the ability to integrate deeply with their Slack communications and workspace management.

Highlights

• Flexible Authentication: Supports both secure OAuth tokens and a stealth mode that doesn't require bot installations.
• Comprehensive Toolset: Includes tools for reading history, searching messages, managing threads, and handling unread messages with priority sorting.
• Workspace Management: Ability to create, update, and manage user groups and memberships.
• Smart History & Search: Supports pagination by date or count and advanced filtering for message searches.
• Safety Controls: High-risk tools like conversations_add_message and reactions_add are disabled by default and must be explicitly enabled via environment variables.
• Enterprise Support: Compatible with Slack Enterprise Grid setups.