Slack MCP — Team Messaging Server for AI Agents

Available Tools

slack_send_message

"Post to #engineering: 'Deploy to staging complete, ready for QA'"
"Send a DM to @alice: 'PR #234 is ready for your review'"

slack_read_channel

"Show the last 20 messages from #incidents"
"What was discussed in #product-planning today?"

slack_search

"Search Slack for discussions about rate limiting"
"Find messages from @bob about the database migration"

slack_reply_thread

"Reply to the deployment thread: 'All tests passing on staging'"

slack_list_channels

"Show all public channels and their topics"

Use Cases

Deploy Notifications

"Post to #releases: 'v2.4.1 deployed — changelog: fixed auth bug, added rate limiting'"
→ Agent formats and posts the release notification

Incident Context

"Search #incidents for the database outage last Tuesday and summarize what happened"
→ Agent searches, reads threads, and provides a summary

Standup Automation

"Read my recent git commits and post a standup update to #daily-standup"
→ Agent reads git log, formats standup, posts to Slack

Key Stats

• 1,500+ GitHub stars
• Send, read, search, thread management
• Channel and DM support
• Full message history access
• Bot token scoping for security

FAQ

Q: What is Slack MCP? A: An MCP server that gives AI agents full access to Slack — sending messages, reading channels, searching history, and managing threads via natural language.

Q: Is it safe? A: Use a bot token with minimal scopes. The agent can only access channels the bot is invited to.

Q: Is Slack MCP free? A: The MCP server is open-source. Slack is free for small teams.