Slack Search

🚀 From Slack Logs to Intelligent Search - Powered by MCP and Claude Desktop

A local, AI-driven Slack search interface that enhances the Claude Desktop experience using the Model Context Protocol (MCP). This project helps you search and gain meaningful insights from Slack export logs locally and privately.

🚀 Quick Start

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 through a JSON configuration file (claude_desktop_config.json), allowing you to directly query Slack data using natural language.

✨ Features

• Set up a local MCP server to parse Slack export logs.
• Register custom MCP resources for intelligent search.
• Integrate with Claude Desktop via a JSON configuration file.
• Use natural language to query Slack data directly.

📦 Installation

1. Clone the repository

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 the Claude Desktop configuration

Paste the following content into your Claude Desktop configuration 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 the file paths to match your local machine settings.

💻 Usage Examples

Basic Usage

Open Claude Desktop and try the following queries:

Search for onboarding discussions in February.
Did we mention the pricing model in the #general channel?
Find mentions of "release" in the random channel.

Claude will use MCP URIs in the search:// and summary:// styles to call your local server behind the scenes to get the results.

📚 Documentation

Project Structure

├── mcp_server.py               # MCP server for registering search resources
├── claude_desktop_config.json # Claude Desktop configuration file for connecting to this server
├── Slack-dataset/              # Slack dataset (not included)
└── README.md                   

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

How it Works

• The MCP server parses Slack JSON files and indexes the content.
• It registers searchable resources such as search://{query} or summary://{channel}.
• Claude Desktop uses MCP calls to get context results from your server.

Motivation

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

📄 License

Acknowledgments

• Claude Desktop
• Model Context Protocol (MCP)