Slack-Search

Slack-Search

2
research_and_datacommunicationknowledge_and_memory

This project provides an AI-powered search interface for Slack logs by setting up a local MCP server. It enhances Claude Desktop's capabilities by allowing intelligent, natural language search and retrieval of information from exported Slack data. All operations are performed locally, ensuring privacy.

From Slack Logs to Smart Search — Powered by MCP and Claude Desktop

A local, AI-powered Slack search interface that uses Model Context Protocol (MCP) to enhance Claude Desktop experience. This project helps you search and retrieve meaningful insights from Slack export logs, all locally and privately.

What is This?

This project sets up a local MCP server (mcp_server.py) that parses Slack export logs and registers custom MCP resources for intelligent search. It integrates with Claude Desktop via a JSON config (claude_desktop_config.json), allowing you to query Slack data directly using natural language.

Project Structure

├── mcp_server.py               # The MCP server that registers search resources
├── claude_desktop_config.json # Config file for Claude Desktop to connect to this server
├── Slack-dataset/              # Slack dataset (not included)
└── README.md

Dataset found at: https://github.com/preethac/Software-related-Slack-Chats-with-Disentangled-Conversations

⚙️ Setup Instructions

1. Clone the repo

git clone https://github.com/AdilFayyaz/Slack-Search.git
cd Slack-Search

2. Set up a virtual environment

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

3. Update Claude Desktop Config

Paste the following into your Claude Desktop config file:

{
  "mcpServers": {
    "search_slack": {
      "command": "/bin/zsh",
      "args": [
        "-c",
        "source /Users/yourname/path-to-project/.venv/bin/activate && python3 /Users/yourname/path-to-project/mcp_server.py"
      ],
      "host": "127.0.0.1",
      "port": 5000,
      "timeout": 10000
    }
  }
}

Make sure to update file paths to your local machine setup.

Usage

Open Claude Desktop, and try queries like:

Search for onboarding discussions from February.
What did we say about the pricing model in #general?
Find mentions of "launch" in random.

Claude will use search:// and summary:// style MCP URIs behind the scenes to get you results from your local server.

How it Works

  • The MCP server parses Slack JSON files and indexes content.
  • It registers searchable resources like search://{query} or summary://{channel}.
  • Claude Desktop makes MCP calls to your server to fetch and return contextual results.

Motivation

Slack is where decisions happen and get buried. This tool helps resurface important context from old threads without uploading your data to the cloud. It's a small step toward smarter, local-first search powered by AI.

Acknowledgments