slack-mcp-ai-assistant

A Slack bot that integrates with Model Context Protocol (MCP) servers to provide AI-powered assistance with access to various tools and services. It summarize requirements in a Slack thread. Output is a Merge Request with analytics.

Architecture

The bot is designed with four core components:

1. SlackMCPBot: Core class managing Slack events and message processing
2. LLMClient: Handles communication with OpenAI API
3. Server: Manages communication with MCP servers
4. Tool: Represents available tools from MCP servers

Features

• 🔌 MCP Integration: Connect to multiple MCP servers and discover their tools
• 🤖 AI-Powered Responses: Uses OpenAI GPT models with function calling
• 💬 Slack Integration: Responds to mentions and direct messages in Socket Mode
• 🧵 Thread Support: Maintains conversation context within Slack threads
• 📝 Comprehensive Logging: Structured logging with Winston
• 🔄 Graceful Shutdown: Properly disconnects from all services on shutdown
• 🛠️ Tool Execution: Executes tools from MCP servers based on conversation context

Setup

Prerequisites

• Node.js 18.0.0 or higher
• A Slack app with Socket Mode enabled
• OpenAI API key
• MCP servers configured and accessible

Installation

1. Clone and install dependencies:

   npm install
   

2. Configure environment variables:

   cp env.example .env
   

   Edit .env with your configuration:

   SLACK_BOT_TOKEN=xoxb-your-bot-token-here
   SLACK_APP_TOKEN=xapp-your-app-token-here
   SLACK_SIGNING_SECRET=your-signing-secret-here
   OPENAI_API_KEY=sk-your-openai-api-key-here
   

3. Configure MCP servers:

   Edit servers_config.json to add your MCP servers:

   {
     "mcpServers": {
       "mcp-gitlab": {
         "command": "uv",
         "args": [
           "--directory",
           "/path/to/mcp-gitlab",
           "run",
           "mcp-gitlab",
           "--gitlab-url",
           "https://your-gitlab.com",
           "--gitlab-token",
           "your-gitlab-token"
         ]
       }
     }
   }
   

Slack App Configuration

1. Create a Slack App at https://api.slack.com/apps

2. Enable Socket Mode in your app settings

3. Add Bot Token Scopes:

   • app_mentions:read
   • chat:write
   • im:read
   • im:write
   • channels:read
   • groups:read
   • mpim:read

4. Subscribe to Bot Events:

   • app_mention
   • message.im

5. Install the app to your workspace

Usage

Starting the Bot

# Development mode
npm run dev

# Production mode
npm start

Interacting with the Bot

1. Mention the bot in any channel:

   @your-bot-name help me create a merge request
   

2. Direct message the bot:

   What's the status of project X?
   

3. Use in threads for contextual conversations:

   • The bot maintains conversation history within threads
   • Tools are executed based on the conversation context

Enhanced Tool Execution Details

The bot now features an enhanced View Details popup that provides comprehensive information about tool executions:

View Details Features

• Visual Tool Icons: Each tool displays with contextual icons (🔧 for GitLab, 💬 for Slack, 📋 for JIRA, etc.)
• Execution Metadata: Shows timestamps, execution status, and performance information
• Enhanced Code Blocks: Properly formatted JSON, code, and text with syntax highlighting
• Structured Layout: Clear sections for parameters, results, and metadata
• Error Handling: Detailed error information with troubleshooting suggestions
• Batch Execution Summary: Overview of multiple tool executions with success/failure counts

Tool Icon Legend

Icon	Tool Type	Description
🦊	GitLab	Repository management, MRs, pipelines
💎	JIRA	Task management, updates, transitions
💬	Slack	Messaging, channels, user management
📚	TechDocs	Documentation search and retrieval
💭	Think	AI reasoning and analysis
🔧	Generic	General-purpose tools

Code Block Formatting

The popup automatically formats different content types:

• JSON Objects: Properly indented with syntax highlighting
• Long Text: Multi-line code blocks for readability
• Thinking Results: Special formatting for AI reasoning
• Compact Mode: Summarized view for batch operations

Example Interactions

GitLab Integration (if GitLab MCP server is configured):

@bot create a merge request from feature-branch to main
@bot show me recent pipelines for project myproject
@bot what are the failing jobs in the latest pipeline?

General Assistance:

@bot help me understand this error message: [paste error]
@bot summarize the recent activity in this project

Viewing Tool Details:

• Click the "View Details" button on any tool execution
• Get comprehensive information about parameters, results, and execution metadata
• Perfect for debugging, auditing, and understanding tool behavior

Architecture Details

SlackMCPBot Class

• Manages Slack app initialization and event handling
• Maintains conversation state per thread
• Coordinates between LLM and MCP servers
• Handles graceful shutdown

LLMClient Class

• Wraps OpenAI API with function calling support
• Manages conversation history and context
• Converts MCP tools to OpenAI function format
• Handles tool execution responses

Server Class

• Manages individual MCP server connections
• Spawns and communicates with MCP server processes
• Discovers and validates available tools
• Executes tool calls with proper error handling

Tool Class

• Represents tools available from MCP servers
• Validates tool arguments against schemas
• Converts to OpenAI function calling format
• Provides metadata for LLM context

Configuration

Environment Variables

Variable	Required	Description
SLACK_BOT_TOKEN	Yes	Slack bot token (starts with xoxb-)
SLACK_APP_TOKEN	Yes	Slack app token (starts with xapp-)
SLACK_SIGNING_SECRET	No	Slack signing secret for verification
OPENAI_API_KEY	Conditional	OpenAI API key (required if using OpenAI)
ANTHROPIC_API_KEY	Conditional	Anthropic API key (required if using Anthropic)
LLM_MODEL	No	Model to use for either provider (default: gpt-4.1 for OpenAI, claude-3-5-sonnet-20241022 for Anthropic)
LLM_MAX_TOKENS	No	Max tokens per response (default: 4000)
LLM_TEMPERATURE	No	Response randomness (default: 0.7)
NODE_ENV	No	Environment (development/production)
LOG_LEVEL	No	Logging level (default: info)

LLM Provider Selection

The bot automatically selects the LLM provider based on which API key is provided:

• Anthropic Claude: Used when only ANTHROPIC_API_KEY is provided
• OpenAI: Used when only OPENAI_API_KEY is provided, or when both API keys are provided (OpenAI takes precedence)

At least one API key (either OPENAI_API_KEY or ANTHROPIC_API_KEY) must be provided.

Use LLM_MODEL to specify the model for either provider. If not specified, defaults to:

• OpenAI: gpt-4.1
• Anthropic: claude-3-5-sonnet-20241022

MCP Server Configuration

The servers_config.json file defines MCP servers:

{
  "mcpServers": {
    "server-name": {
      "command": "executable-command",
      "args": ["arg1", "arg2"],
      "env": {
        "OPTIONAL_ENV_VAR": "value"
      }
    }
  }
}

Logging

Logs are written to the logs/ directory:

• combined.log: All log messages
• error.log: Error messages only
• exceptions.log: Uncaught exceptions
• rejections.log: Unhandled promise rejections

Log levels: error, warn, info, debug

Development

Project Structure

src/
├── bot/
│   └── SlackMCPBot.js     # Main bot class
├── llm/
│   └── LLMClient.js       # OpenAI & Anthropic integration
├── mcp/
│   └── Server.js          # MCP server management
├── types/
│   └── Tool.js            # Tool representation
├── utils/
│   ├── config.js          # Configuration loader
│   └── logger.js          # Logging setup
└── index.js               # Entry point

Adding New Features

1. New MCP Server: Add configuration to servers_config.json
2. Custom Tools: Extend the Tool class for special handling
3. Event Handlers: Add new Slack event handlers in SlackMCPBot
4. LLM Customization: Modify prompts and behavior in LLMClient

Troubleshooting

Common Issues

1. "Missing required environment variables"

   • Ensure all required env vars are set in .env

2. "Failed to connect to MCP server"

   • Check MCP server command and arguments
   • Verify server executable is available
   • Check logs for detailed error messages

3. "Tool execution failed"

   • Verify tool arguments match expected schema
   • Check MCP server logs
   • Ensure server is still connected

4. Slack connection issues

   • Verify bot tokens are correct
   • Check Slack app configuration
   • Ensure Socket Mode is enabled

Debug Mode

Set LOG_LEVEL=debug for detailed logging:

LOG_LEVEL=debug npm start

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

License

MIT License - see LICENSE file for details