概览
What is Slack MCP Host?
A Slack bot that integrates MCP (Machine Control Protocol) servers with Slack, providing LLM-powered interactions in dedicated channels. It can execute tools from configured MCP servers and present results in a user-friendly format.
How to use Slack MCP Host?
Install dependencies, configure environment variables (Slack token, OpenAI key, MCP server definitions), and deploy to Cloudflare Workers or run locally with uvicorn. Interact with the bot by mentioning it or messaging in designated Slack channels.
Key features of Slack MCP Host
- LLM-powered conversations in Slack
- Integration with multiple MCP servers
- Markdown rendering support
- Collapsible tool execution results
- Conversation history maintenance
- Secure configuration management
Use cases of Slack MCP Host
- Query a Neo4j movie database for film details
- Retrieve current time or date via MCP time server
- Execute any tool from configured MCP servers in Slack
- Enable team collaboration with an LLM-powered Slack bot
FAQ from Slack MCP Host
What prerequisites are needed to run Slack MCP Host?
Python 3.11+, the uv package manager, a Slack workspace with admin access, an OpenAI API key, and one or more MCP servers.
How do I configure MCP servers?
Define each MCP server as an environment variable with a prefix like SLACK_MCP_ (e.g., SLACK_MCP_NEO4J_MOVIES=uvx mcp-neo4j-cypher --url ...).
How do I deploy Slack MCP Host?
You can deploy to Cloudflare Workers using Wrangler, or run locally with uvicorn src.api:app --reload --port 8000 and expose via ngrok.
What LLM model does the bot use?
By default it uses GPT-4, as configured in the .env file with OPENAI_MODEL=gpt-4.
Does the bot support formatting and collapsible results?
Yes, it renders Markdown and shows tool execution results inside collapsible <details> elements.
