Human

v1.0.0Communicationstable

An MCP server that allows AI assistants to utilize human capabilities by sending requests to humans and receiving their responses through a Streamlit UI.

claudecodeclawgoclawmcpopenclaw
Share:
289
Stars
0
Downloads
0
Weekly
0/5

What is Human?

Human is a Model Context Protocol (MCP) server that allows AI assistants like Claude, Cursor, and VS Code to mcp server that allows ai assistants to utilize human capabilities by sending requests to humans and receiving their responses through a streamlit ui.

An MCP server that allows AI assistants to utilize human capabilities by sending requests to humans and receiving their responses through a Streamlit UI.

This server falls under the Communication category on MCPgee, the world's largest MCP server directory with 33,000+ servers.

Features

  • An MCP server that allows AI assistants to utilize human cap

Use Cases

Request human input for AI agents through a web interface.
Delegate decisions to humans when AI needs validation.
Create hybrid workflows combining AI and human expertise.
upamune

Maintainer

LicenseMIT License
Languagetypescript
Versionv1.0.0
UpdatedMay 16, 2026
Statushealthy
Maintenanceactive

Works with

ClaudeOpenAIwindowsmacoslinux

Installation

NPM

npx -y human-mcp

Manual Installation

npx -y human-mcp

Configuration

Configuration Details

Config File

claude_desktop_config.json

Performance

Response Metrics

Response Time< 200ms
ThroughputMedium

Resource Usage

Memory UsageLow
CPU UsageLow

How to Set Up and Use Human

The Human MCP server enables AI agents to delegate tasks that require human perception or physical interaction by routing requests through a Streamlit web interface where a human operator can respond in real time. It exposes seven sensory tools — covering sight, hearing, smell, taste, touch, speech, and local weather — allowing AI assistants to incorporate human judgment and real-world observation into otherwise automated workflows. This server is particularly useful for hybrid human-AI pipelines in Claude Code or similar agent frameworks where certain actions require a human in the loop.

Prerequisites

  • Python 3.12+ and uv package manager installed
  • Git to clone the repository
  • SQLite3 (usually included with Python)
  • An MCP-compatible client such as Claude Code or Claude Desktop
  • A web browser to access the Streamlit operator interface at http://localhost:8501
1

Clone the repository

Clone the human-mcp repository from GitHub to get the server and Streamlit UI code.

git clone https://github.com/upamune/human-mcp.git
cd human-mcp
2

Set up the virtual environment and install dependencies

Create a uv virtual environment and install all required packages including the MCP CLI and Streamlit.

uv venv
source .venv/bin/activate
uv pip install .
3

Start the Streamlit operator interface

Launch the Streamlit web UI that the human operator uses to receive AI requests and submit responses.

uv run streamlit run human_mcp/streamlit_app.py
4

Configure the MCP server in your client

Add the human-mcp server to your claude_desktop_config.json, pointing to the mcp_server.py file in the cloned repository.

5

Test with an agent prompt

Ask your AI assistant to use a human sense tool. The request will appear in the Streamlit UI at http://localhost:8501 for the operator to respond to.

Human Examples

Client configuration

Claude Desktop config that runs the human-mcp MCP server via uv, pointing to the local repository path.

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

Prompts to try

Example prompts that trigger human-mcp tools, routing the request to the Streamlit operator interface.

- "Look at the object on the desk and describe what you see" (triggers human_eye_tool)
- "What does the weather feel like outside right now?" (triggers human_weather_tool)
- "Can you smell anything unusual in the room?" (triggers human_nose_tool)
- "Please say 'Hello, your package has arrived' out loud" (triggers human_mouth_tool)
- "What sounds can you hear in the environment right now?" (triggers human_ear_tool)

Troubleshooting Human

Streamlit UI is not accessible at http://localhost:8501

Ensure the Streamlit app is running (uv run streamlit run human_mcp/streamlit_app.py). Check that port 8501 is not in use by another process with: lsof -i :8501.

Tool calls from the AI agent time out waiting for human response

The server waits for a human operator to respond via the Streamlit UI. Ensure someone is watching the UI and submitting responses. For testing, open http://localhost:8501 in a browser tab while running the agent.

mcp run command fails with module not found

Activate the uv virtual environment (source .venv/bin/activate) and install the MCP CLI: uv pip install 'mcp[cli]'. Then re-run the server command.

Frequently Asked Questions about Human

What is Human?

Human is a Model Context Protocol (MCP) server that mcp server that allows ai assistants to utilize human capabilities by sending requests to humans and receiving their responses through a streamlit ui. It connects AI assistants to external tools and data sources through a standardized interface.

How do I install Human?

Install via npm with the command: npx -y human-mcp. Then add the server configuration to your AI client's JSON config file (e.g., claude_desktop_config.json or .cursor/mcp.json).

Which AI clients work with Human?

Human works with all major MCP-compatible AI clients including Claude Desktop, Claude Code, Cursor, VS Code (GitHub Copilot), Windsurf, and Cline.

Is Human free to use?

Yes, Human is open source and available under the MIT License license. You can use it freely in both personal and commercial projects.

Browse More Communication MCP Servers

Explore all communication servers available in the MCPgee directory. Each server includes setup guides for Claude, Cursor, and VS Code.

Quick Config Preview

{ "mcpServers": { "human": { "command": "npx", "args": ["-y", "human-mcp"] } } }

Add this to your claude_desktop_config.json or .cursor/mcp.json

Read the full setup guide →

Ready to use Human?

Browse our complete directory of 33,000+ MCP servers, read setup guides for your editor, and start building with the Model Context Protocol.

33,000+ ServersFree & Open SourceStep-by-Step Guides