This MCP workflow sets up a Slack server that lets Claude read channels, search messages, post replies, and summarize conversations. Perfect for getting AI-powered answers about your team discussions without leaving your chat.
TL;DR — Key Takeaways
- Slack AI Assistant via MCP — Intermediate level
- Tool: MCP
- 5 steps with detailed instructions
- Prerequisites: Slack workspace admin access, Claude Desktop or API, MCP Slack server, Node.js 18+
Use Case
Teams using Slack who want AI-powered search, summarization, and Q&A over their workspace conversations.
Prerequisites
Slack workspace admin access, Claude Desktop or API, MCP Slack server, Node.js 18+
Workflow Diagram
Slack AI Assistant via MCP
Step-by-Step Guide
-
Create Slack App
Go to api.slack.com/apps and create a new app. Add bot token scopes: channels:read, channels:history, chat:write, search:read. Install the app to your workspace. Save the Bot User OAuth Token.
-
Install MCP Slack Server
Run: npx @modelcontextprotocol/server-slack. Set SLACK_BOT_TOKEN environment variable. Configure in Claude Desktop config file as a new MCP server.
-
Test Basic Operations
Ask Claude: List all channels in my workspace. Then: Search for messages about project deadline. Then: Summarize the last 50 messages in the general channel. Verify Claude can read and search Slack data.
-
Build Analysis Workflows
Ask Claude: What were the main decisions made in the engineering channel this week? or: Find all unresolved questions from the last 3 days and create a summary. Claude will search, read, and synthesize across channels.
-
Set Up Daily Digest
Use n8n or a cron job to trigger Claude each morning with: Summarize important updates from all channels since yesterday. Claude reads via MCP, generates a digest, and posts it to a summary channel or sends via email.
Tags
This workflow answers:
- How to set up slack ai assistant via mcp
- What tools and access you need
- How to troubleshoot common issues