tiger-slack

Tiger Slack exposes an MCP HTTP server that lets AI assistants query and analyze your Slack data with full threading context, channel and user browsing, and message permalink generation. The server is designed to work with tools like Claude Code and the MCP Inspector to run queries, fetch contextual data, and drive AI-powered insights over your Slack workspace. To start using it, connect your MCP client to the server’s HTTP endpoint (for example via the Claude Code integration or via the MCP Inspector web UI) and begin issuing memory-aware queries, such as retrieving recent conversations in a channel or listing participants in a thread. The server is designed to work alongside the ingest service and TimescaleDB backend to provide fast, context-rich responses that reference past discussions and project history.

Prerequisites:

• Docker and Docker Compose (for the recommended docker-compose workflow) or a local Node.js runtime if you prefer running the MCP server directly.
• A Slack app configured with the necessary tokens for the ingest service (SLACK_BOT_TOKEN and SLACK_APP_TOKEN).
• Git with submodules access since this repo uses submodules.

1. Clone the repository with submodules:

git clone --recurse-submodules git@github.com:timescale/tiger-slack.git
cd tiger-slack

2. Ensure environment configuration for Slack and database is prepared. Copy example env into place and fill tokens:

cp inges t/.env.sample .env

3. Build and run the system using Docker Compose (recommended):

docker compose up -d --build

4. If you prefer running the MCP server directly (Node.js):

# Install dependencies (from mcp directory)
cd mcp
npm install

# Run the MCP server (adjust path as needed)
node mcp/dist/index.js

5. Verify the MCP endpoint is reachable (default port 3001 as shown in docs) and connect your MCP clients (Claude Code, MCP Inspector) to http://localhost:3001/mcp

Notes and tips:

• The ingest service uses a Python app; ensure your Slack app tokens are configured and available in the .env file before starting ingestion.
• The MCP server communicates with TimescaleDB; ensure the database is provisioned and accessible by the MCP server.
• If you use Claude Code, you can add the MCP server with commands like: claude mcp add -s project tiger-slack http://localhost:3001/mcp. You can also run the server in stdio mode if you’ve built a local JS bundle.
• Optional observability: configure LOGFIRE_TOKEN for tracing if you use Logfire with Claude Code or MCP Inspector.
• For production usage, consider securing the MCP endpoint and controlling access via networking/firewall rules.
• Environment variables for Slack and DB connections are typically set in the .env file consumed by the ingest and MCP services; review the ingest/README.md for specifics on per-service env vars.