Query live market data, manage perp trading positions, and execute orders on Paradex directly from Claude. Designed for traders and developers building algorithmic trading workflows.
io.github.tradeparadex/mcp-paradex-py
Local install
STDIO
No auth required
How models use it and what it is built for.
Query live market data, manage perp trading positions, and execute orders on Paradex directly from Claude. Designed for traders and developers building algorithmic trading workflows.
Local install — runs as a subprocess.
Configuration this server reads at startup.
Ethereum private key for your Paradex account (enables trading). Without this, only public market data is accessible.
Paradex environment to connect to. Defaults to 'prod'. Use 'testnet' for testing.
Where to find authoritative docs and source for Paradex Trading.
Paste any of these into Agent Studio after connecting Paradex Trading.
Common questions about connecting and running Paradex Trading.
Do I need an Ethereum private key to use this MCP server?
Only if you want to execute trades. Set PARADEX_ACCOUNT_PRIVATE_KEY to enable order placement and position management. Without it, you can still query public market data.
Can I test this server before trading on mainnet?
Yes. Set PARADEX_ENVIRONMENT=testnet to connect to Paradex's testnet. The default is 'prod' for mainnet trading.
What market data can I access?
The server provides order books, funding rates, and other live market data for Paradex perpetuals. You can also query your account balances, open orders, positions, and vault information.
Is my private key stored or transmitted securely?
The private key is passed via the PARADEX_ACCOUNT_PRIVATE_KEY environment variable and used locally to sign transactions. Refer to Paradex's security documentation for details on key handling best practices.
What happens if I don't set PARADEX_ACCOUNT_PRIVATE_KEY?
The server will operate in read-only mode, allowing you to query market data and account information but preventing any trades or position changes.
MCP Playground runs 10,000+ hosted MCP servers — GitHub, Linear, Notion, Stripe, Sentry and more — across Claude, GPT, Gemini, DeepSeek and 30+ AI models. Compare model answers side-by-side, save agent presets, share runs. Zero install.
Open Agent Studio