> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.paradex.trade/agentic-ai-hub/mcp/quickstart/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.paradex.trade/_mcp/server.

# Quickstart guide

**Repository:** [github.com/tradeparadex/mcp-paradex-py](https://github.com/tradeparadex/mcp-paradex-py)

Once configured, your AI assistant can:

* Fetch live and historical market data
* Manage trading accounts and vaults
* Place, monitor, and cancel orders
* Track positions, balances, and funding payments

Your setup choice depends on:

* **Which MCP features you need** — trading is only available with a local setup at the moment. Remote and Paradex-hosted options provide market data only.
* **Which AI client you prefer** — Claude Desktop, Cursor, ChatGPT, and others are supported. Not every client supports every setup path.

### Pick your client

Use the table below to see which setup paths are available for your client, then pick your setup path.

| Client                                                   | Local MCP setup | Remote MCP deployment | Paradex-hosted MCP |
| -------------------------------------------------------- | :-------------: | :-------------------: | :----------------: |
| [ChatGPT Desktop](/agentic-ai-hub/mcp/chatgpt-desktop)   |                 |                       |                    |
| [Claude Code CLI](/agentic-ai-hub/mcp/claude-code)       |                 |                       |                    |
| [Claude Desktop](/agentic-ai-hub/mcp/claude-desktop)     |                 |                       |                    |
| [Claude.ai (web)](/agentic-ai-hub/remote-mcp/quickstart) |                 |                       |                    |
| [Codex CLI](/agentic-ai-hub/mcp/codex-cli)               |                 |                       |                    |
| [Cursor](/agentic-ai-hub/mcp/cursor)                     |                 |                       |                    |
| [Gemini CLI](/agentic-ai-hub/mcp/gemini-cli)             |                 |                       |                    |
| [VS Code (GitHub Copilot)](/agentic-ai-hub/mcp/vscode)   |                 |                       |                    |
| [Windsurf](/agentic-ai-hub/mcp/windsurf)                 |                 |                       |                    |

### Pick your setup path

#### Local MCP setup (Recommended)

MCP server runs on your machine. Full API support including trading.

1. Complete the [Prerequisites](/agentic-ai-hub/mcp/prerequisites) — install Python, uvx, and get your Paradex private key.
2. Click your client name in the table above and follow the setup instructions.

#### Remote MCP deployment

You host your own MCP server on a cloud platform (Railway, Render, Fly.io, etc.) and connect your AI client to it via a URL. Market data only — no trading support at the moment.

1. Follow the [Remote MCP quickstart](/agentic-ai-hub/remote-mcp/quickstart) to deploy your server.
2. Click your client name in the table above and enter your endpoint URL.

#### Paradex-hosted MCP

The fastest way to get started — no installation, no cloud deployment. Market data only — no trading support at the moment.

Follow the [Paradex-hosted MCP quickstart](/agentic-ai-hub/paradex-hosted-mcp/quickstart) to connect via URL or Smithery.

> The Paradex MCP server implements the [Model Context Protocol](https://modelcontextprotocol.io/) standard, giving AI assistants structured, authenticated access to the Paradex trading platform.