AskOnSlackMCP
A Human-in-the-Loop MCP server that enables AI assistants to request information or clarification from humans via Slack. It uses real-time, thread-based conversations with socket mode integration to bridge the gap between AI systems and human experts.
README
Human-in-the-Loop Slack MCP Server
A Model Context Protocol (MCP) server that enables AI assistants to request information from humans via Slack. This server acts as a bridge between AI systems and human experts, allowing AI to ask questions and receive responses through Slack when it needs human knowledge or clarification.
Quick Start with npx
Run directly from GitHub without installation:
npx github:trtd56/AskOnSlackMCP \
--slack-bot-token "xoxb-your-bot-token" \
--slack-app-token "xapp-your-app-token" \
--slack-channel-id "C1234567890" \
--slack-user-id "U1234567890"
Example with Claude Desktop
Add to your Claude Desktop configuration (~/Library/Application Support/Claude/claude_desktop_config.json):
{
"mcpServers": {
"slack-human": {
"command": "npx",
"args": [
"github:trtd56/AskOnSlackMCP",
"--slack-bot-token", "xoxb-your-actual-token",
"--slack-app-token", "xapp-your-actual-token",
"--slack-channel-id", "C1234567890",
"--slack-user-id", "U1234567890"
]
}
}
}
Features
- 🤖 MCP-compliant server for AI assistant integration
- 💬 Real-time Slack integration via Socket Mode WebSocket connection
- 🧵 Thread-based conversations for maintaining context
- ⏱️ 60-second timeout for human responses
- 📢 User mentions (
@username) for notifications - 🔍 Comprehensive debugging and logging capabilities
- 🔐 Secure token handling
- 🚀 Dynamic handler initialization for faster startup
- ⚡ Optimized for instant response detection with event-driven architecture
Prerequisites
-
Slack App Setup
- Create a new Slack app at https://api.slack.com/apps
- Enable Socket Mode in your app settings
- Generate an App-Level Token with
connections:writescope - Install the app to your workspace
-
Bot Token Scopes
chat:write- Send messageschannels:read- Access channel informationusers:read- Access user information
-
Socket Mode
- Enable Socket Mode in your app settings
- This is required for the app to receive events in real-time
-
Event Subscriptions
- Enable Events API
- Subscribe to bot events:
message.channels- Messages in public channelsmessage.groups- Messages in private channelsmessage.im- Direct messages (optional)
- Save changes and reinstall the app to your workspace
Installation (Optional)
If you want to install locally instead of using npx:
- Clone the repository:
git clone https://github.com/trtd56/AskOnSlackMCP.git
cd AskOnSlackMCP
- Install dependencies:
npm install
- Build the TypeScript code:
npm run build
Configuration
All configuration is passed via command-line arguments:
--slack-bot-token- Bot User OAuth Token (xoxb-...)--slack-app-token- App-Level Token for Socket Mode (xapp-...)--slack-channel-id- Channel ID where the bot will operate--slack-user-id- User ID to mention when asking questions--log-level- (Optional) Logging level (default: INFO)
Usage
Development Mode
Run with hot-reloading:
npm run dev
Production Mode
Build and run:
npm run build
npm start
With MCP Client (Using npx)
Configure your MCP client to use this server directly from GitHub:
{
"mcpServers": {
"human-in-the-loop-slack": {
"command": "npx",
"args": [
"github:trtd56/AskOnSlackMCP",
"--slack-bot-token", "xoxb-your-token",
"--slack-app-token", "xapp-your-token",
"--slack-channel-id", "C1234567890",
"--slack-user-id", "U1234567890"
]
}
}
}
With MCP Client (Local Installation)
If you've installed locally:
{
"mcpServers": {
"human-in-the-loop-slack": {
"command": "node",
"args": [
"/path/to/AskOnSlackMCP/dist/index.js",
"--slack-bot-token", "xoxb-your-token",
"--slack-app-token", "xapp-your-token",
"--slack-channel-id", "C1234567890",
"--slack-user-id", "U1234567890"
]
}
}
}
Available Tools
ask_on_slack
Main tool for asking questions to humans via Slack.
Parameters:
question(string): The question to ask the human. Be specific and provide context.
Example:
{
"tool": "ask_on_slack",
"arguments": {
"question": "What is the API endpoint for the production server?"
}
}
Usage Notes:
- The bot will mention the specified user in the Slack channel
- The human has 60 seconds to respond in a thread
- The tool will return the human's response or timeout after 60 seconds
Development
Scripts
npm run build- Compile TypeScriptnpm run dev- Run with hot-reloadingnpm start- Run compiled codenpm test- Run tests with Vitestnpm run test:ci- Run tests with coveragenpm run lint- Run ESLintnpm run format- Format code with Prettiernpm run clean- Clean build artifacts
Project Structure
src/
├── index.ts # Main MCP server implementation
├── bin.ts # Binary entry point for npx execution
├── human.ts # Abstract Human interface
├── slack-client.ts # Socket Mode Slack implementation
└── types.ts # TypeScript type definitions
tests/
├── human.test.ts # Human abstract class tests
├── index.test.ts # CLI argument parsing tests
├── slack-client.test.ts # Slack client tests
└── types.test.ts # Type definition tests
Testing
The project uses Vitest for testing. Tests are located in the tests/ directory.
To run tests:
npm test # Run tests in watch mode
npm run test:ci # Run tests once with coverage
CI/CD
The project uses GitHub Actions for continuous integration and deployment.
-
CI Workflow (
ci.yml): Runs on every push and pull request- Tests on Node.js 18.x, 20.x, and 22.x
- Runs linting and type checking
- Generates code coverage reports
- Builds the project
-
Release Workflow (
release.yml): Runs on version tags- Builds and tests the project
- Creates GitHub releases
- Publishes to npm (requires NPM_TOKEN secret)
Troubleshooting
-
Connection Issues
- Verify all tokens are correct
- Check that the bot is invited to the channel
- Ensure Socket Mode is enabled in your Slack app
-
No Response Received
- Verify the user ID is correct (format: U1234567890)
- Ensure the user responds in the message thread, not the main channel
- Check that the bot has permission to read messages in the channel
-
Authentication Errors
- Bot token should start with
xoxb- - App token should start with
xapp- - Regenerate tokens if needed
- Verify bot has required scopes:
chat:write,channels:read,users:read
- Bot token should start with
-
Performance Optimization
- The server uses event-driven architecture for instant response detection
- WebSocket connection ensures real-time message delivery
- Detailed timing logs available with
[TIMING]prefix for debugging
License
MIT
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。