MCP Paradex Server

Model Context Protocol (MCP) server implementation for the Paradex trading platform.

Quickstart

New here? See the Quickstart Guide for step-by-step setup instructions for Claude Desktop, Claude Code, Cursor, Windsurf, VS Code, Claude.ai web, and ChatGPT.

Need a remote HTTP endpoint for ChatGPT or Claude.ai web? See the HTTP Deployment Guide for Railway, Render, Fly.io, AWS Lambda, and Docker/VPS options.

Overview

This project provides a bridge between AI assistants (like Claude) and Paradex, a privacy-preserving multi-asset exchange on Starknet with spot, perpetuals, options, and commodity futures. Using the MCP standard, AI assistants can:

• Retrieve market data from Paradex
• Manage trading accounts and vaults
• Place and manage orders
• Monitor positions and balance

Prerequisites

• Python 3.10+

• uv — required for the uvx run method

  macOS: Install via brew install uv so the binary lands in /usr/local/bin, which GUI apps like Claude Desktop can find. The curl-pipe install puts uv in ~/.local/bin, which GUI apps cannot see.

Installation

Quick Start

Cursor IDE

Click to automatically configure this MCP server in Cursor:

Claude Code CLI

claude mcp add paradex uvx mcp-paradex

Smithery (Claude Desktop)

npx -y @smithery/cli@latest mcp add @tradeparadex/mcp-paradex-py --client claude

Standard Installation

PyPI

pip install mcp-paradex

uvx (Recommended)

uvx mcp-paradex

Development Setup

1. Clone the repository:

   git clone https://github.com/tradeparadex/mcp-paradex-py.git
   cd mcp-paradex-py
   

2. Install development dependencies:

   uv sync --dev --all-extras
   

3. Run locally:

   uv run mcp-paradex
   

Configuration

Environment Variables

Set these environment variables for authentication:

• PARADEX_ENVIRONMENT: Set to prod, testnet, or nightly (default: prod)
• PARADEX_ACCOUNT_PRIVATE_KEY: Your Paradex account private key

Using .env File

cp .env.template .env
# Edit .env with your credentials

Client Configuration

Claude Desktop

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "paradex": {
      "command": "uvx",
      "args": ["mcp-paradex"],
      "env": {
        "PARADEX_ENVIRONMENT": "testnet",
        "PARADEX_ACCOUNT_PRIVATE_KEY": "your_private_key"
      }
    }
  }
}

Docker (local / stdio)

# Build image
docker build . -t tradeparadex/mcp-paradex-py

# Run (public only)
docker run --rm -i tradeparadex/mcp-paradex-py

# Run with trading capabilities
docker run --rm -e PARADEX_ACCOUNT_PRIVATE_KEY=your_key -i tradeparadex/mcp-paradex-py

Docker (AWS Lambda / HTTP)

Use Dockerfile.aws for remote deployments via AWS Lambda with the Lambda Web Adapter. The adapter bridges Lambda invocations to the server's HTTP endpoint, so no Lambda-specific code is needed.

# Build
docker build -f Dockerfile.aws -t tradeparadex/mcp-paradex-py-aws .

# Test locally (mirrors Lambda config)
docker run --

... [View full README on GitHub](https://github.com/tradeparadex/mcp-paradex-py#readme)