Human MCP

🚀 human-mcp

human-mcp serves as an MCP server provided by human as an MCP tool, enabling AI assistants to leverage human capabilities.

🚀 Quick Start

human-mcp is an MCP server that allows AI assistants to utilize human capabilities. It receives requests from AI assistants, displays instructions on the human-computer interaction interface, and returns human responses to the AI assistants.

✨ Features

• Receive tool execution requests sent by MCP clients via standard input.
• Write the required instructions into the SQLite database.
• The Streamlit application monitors the SQLite database, displays instructions to humans, and prompts for input.
• Receive human input through Streamlit and write it into the SQLite database.
• The MCP server reads the results from the database and returns them to the MCP client via standard output.

🛠️ Provided Tools

1. human_eye_tool: Humans observe the environment with their eyes and describe the situation or find specific objects.
2. human_hand_tool: Manually perform simple physical operations.
3. human_mouth_tool: Issue verbal instructions or say specified words.
4. human_weather_tool: Query and report the current weather conditions at the location.
5. human_ear_tool: Use hearing to perceive the environment and describe the sounds heard.
6. human_nose_tool: Perceive odors through the sense of smell and describe the scents.
7. human_taste_tool: Taste food and describe its flavor.

📦 Installation

Prerequisites

• Python 3.12 or higher
• uv tool
• SQLite3

Installation Steps

1. Clone the repository:

   git clone https://github.com/yourusername/human-mcp.git
   cd human-mcp
   

2. Create and activate a virtual environment:

   uv venv
   source .venv/bin/activate
   

3. Install dependencies:

   uv pip install .
   

💻 Usage Examples

Basic Usage

0. Install the MCP server:

task install-mcp

1. Connect to the MCP server in Claude:

   "human-mcp": {
     "command": "uv",
     "args": [
       "run",
       "--with",
       "mcp[cli]",
       "mcp",
       "run",
       "$PATH_TO_REPOSITORY/human_mcp/mcp_server.py"
     ]
   }
   

2. Start the Streamlit UI in a second terminal:

   task run-streamlit
   

3. Open a browser and access the Streamlit UI (usually at http://localhost:8501).
4. When the MCP client (e.g., Claude Desktop) sends a request, the task will be displayed on the Streamlit UI.
5. Enter the response in the Streamlit UI and click the "Submit Response" button. The response will be returned to the MCP client.

📚 Documentation

Project Structure

human-mcp/
├── human_mcp/          # Core functionality code
│   ├── tools/         # Implementations of various tools
│   └── ui/           # Streamlit user interface
├── config.py          # Configuration file
├── requirements.txt    # Dependency management
└── README.md          # Project description

📄 License

[License information]

⚠️ Important Note

This project is for entertainment and testing purposes only. It is not recommended for use in a production environment.