World ID And MiniApps Documentation

🚀 World MiniApp MCP Server

This project sets up an MCP server for querying World Mini Apps documentation and integrating it with Cursor.

🚀 Quick Start

The main idea of this project is to create an agent that can read the MiniApps documentation and assist you in coding when writing World MiniApps using Cursor.

✨ Features

• Create an agent to read MiniApps documentation.
• Assist in coding World MiniApps with Cursor.

📦 Installation

Prerequisites

• Python 3: Download from python.org or install via Homebrew:

  brew install python
  

• Cursor: Installed on your system.

Installation Steps

1. Set up the Python environment

Open your terminal and run the following commands:

# Create a virtual environment
python3 -m venv .venv

# Activate the virtual environment
source .venv/bin/activate

2. Install dependencies

After activating the virtual environment, install the required packages:

pip install langchain_community langchain-openai langchain-anthropic scikit-learn bs4 pandas pyarrow matplotlib lxml langgraph "mcp[cli]" tiktoken

3. Build the document vector store

Before running the build script, export your OpenAI API key in the terminal:

export OPENAI_API_KEY="your-openai-api-key"

Then run:

python3 worldMiniApp-MCP/miniApp_build_docs.py

Note: Mateo (from the World team) mentioned that you can use https://docs.world.org/llms-full.txt to support all World documentation. This script will create the llms_full.txt and sklearn_vectorstore.parquet files.

4. Run the MCP server

Start the MCP server by running the following command:

python3 worldMiniApp-MCP/miniApp_mcp.py

The server will run on standard input/output and wait for MCP requests.

5. Configure Cursor to use your MCP server

1. Open Cursor settings:
   • In Cursor, go to "Settings" and click on the MCP tab.
2. Activate MCP and add configuration:
   • Copy and paste the following JSON into the MCP configuration area:

{
  "mcpServers": {
    "world-mini-app-mcp": {
      "command": "your-absolute-path-to-repo-folder/WorldMCP/.venv/bin/python",
      "args": [
        "your-absolute-path-to-repo-folder/WorldMCP/worldMiniApp-MCP/miniApp_mcp.py"
      ],
      "env": {
        "OPENAI_API_KEY": "sk-xxxxxxxxxxxxxxxxxxxx"
      }
    }
  }
}

• Replace "your-openai-api-key" with your actual API key.

3. Save the configuration:
   • Cursor will now detect your MCP server.

6. Chat with the MCP agent in Cursor

• Switch to agent mode:
  • Instead of using the standard chat, activate the agent mode in Cursor.
• When you chat, Cursor will use your MCP tool (world_mini_app_query_tool) to fetch World MiniApp documentation as needed.

1. Remember to use Claude Sonnet:
   • Especially the model "claude-3.7-sonnet-thinking".
2. Get an Anthropic API key:
   • Go to https://console.anthropic.com/.
   • Create a profile.
   • Get an API key (it's recommended to top up $10 for testing and building your MiniApp, as Claude 3.7 has higher usage requirements).
3. Open Cursor settings:
   • In Cursor, go to "Settings" and click on the "Models" tab.
   • Select the Claude model.
   • Enter your Anthropic API key.

💡 Optional: Test with MCP Inspector

To manually check your MCP server, install and run the MCP Inspector:

npm install -g @modelcontextprotocol/inspector
mcp-inspector