Slack MCP

Official Slack MCP server — read channels, post messages, search history, and manage your Slack workspace through AI.

TRUSTED · Official

// Add to your client

{
  "mcpServers": {
    "slack": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-slack"],
      "env": {
        "SLACK_BOT_TOKEN": "xoxb-<your-bot-token>",
        "SLACK_TEAM_ID": "T<your-team-id>"
      }
    }
  }
}

{
  "mcpServers": {
    "slack": {
      "command": "npx",
      "args": ["-y", "@modelcontextprotocol/server-slack"],
      "env": {
        "SLACK_BOT_TOKEN": "xoxb-<your-bot-token>",
        "SLACK_TEAM_ID": "T<your-team-id>"
      }
    }
  }
}

claude mcp add slack -e SLACK_BOT_TOKEN=xoxb-<token> -e SLACK_TEAM_ID=T<team> -- npx -y @modelcontextprotocol/server-slack

Paste into your client's MCP configuration file.

// Try it now

"Summarize the last 20 messages in #general and list any action items"

! Requires a Slack app with OAuth scopes — admin approval may be needed in enterprise workspaces
! Posting messages uses your identity — review before enabling automated posts
! Rate limits apply per workspace; heavy channel scraping will hit throttling

Setup

Easy Setup

API key or simple config

Token usage

Light usage

// When to use

You need to catch up on a channel after being away for hours or days
You want to post status updates or announcements from conversation context
You're searching message history for a specific decision or link
You want your AI to monitor mentions and summarize them periodically

// When NOT to use

You need to react with emoji or manage complex workflows (limited)
You use Microsoft Teams, Discord, or another chat platform
You need to handle voice huddles, clips, or Slack Canvas editing

// Usage Scenarios

Channel Catch-Up

Get a summary of what happened in a channel while you were focused on deep work.

Example prompt

"Summarize the last 50 messages in #product-launch and list any open questions"

Find Past Decisions

Search across channel history to find where a decision was made or a link shared.

Example prompt

"Search #engineering for where we decided on the authentication library"

Post Updates from Context

Generate and post a status update based on what you just discussed with the AI.

Example prompt

"Post a summary of today's deploy results to #eng-updates"

// About

Plain English

Slack is where most teams talk all day, and it's where most useful context ends up — decisions, customer feedback, deploy notes, who said yes to what. Catching up after a day off, or finding the thread where a decision was made, can take longer than the actual work. This server lets your AI read and write Slack on your behalf. Ask Claude 'summarize what happened in #product yesterday,' 'find the last message about the pricing change,' or 'post a status update in #eng with the deploy summary' — and it does it. It can read channel history, search across messages, and post updates without you opening Slack at all. For managers, this is a fast way to catch up on a busy day. For developers, it's a way to drop a deploy summary into a channel without tab-switching. For anyone, it makes Slack searchable in plain English instead of through Slack's own search box. A careful note: posting uses your identity. Treat automated posts the way you'd treat any message you sign your name to — review what's about to go out, and limit the bot's permissions to the channels it actually needs.

The official Slack MCP server connects AI assistants to your Slack workspace. Read channel messages, post updates, search conversation history, and interact with Slack's full API — all through natural language.

Maintained by the Slack team with OAuth authentication. A key integration for teams that live in Slack.

GitHub 42.0kUpdated Feb 10, 2025MIT

View on GitHub