Hyperliquid MCP
Enables natural language control of Hyperliquid perpetual futures, including querying positions, prices, orderbook, and executing trades like market and limit orders, all from MCP-compatible clients.
README
🟢 Hyperliquid MCP
Control your Hyperliquid perps from Claude (or any MCP client) using natural language.
"What's my BTC PnL?" → Claude fetches your positions and answers in seconds.
"Buy 0.1 ETH at market" → Claude places the order via Hyperliquid API.
Built with the Model Context Protocol — the open standard for connecting AI to external tools.
✨ Features
| Tool | Description |
|---|---|
hl_get_all_mids |
Prices for all assets |
hl_get_orderbook |
L2 bids/asks for any asset |
hl_get_meta |
Market info (leverage, tick size) |
hl_get_candles |
OHLCV history (1m → 1d) |
hl_get_user_state |
Positions, equity, margin, PnL |
hl_get_open_orders |
Active orders |
hl_get_fills |
Trade history |
hl_get_funding_history |
Funding rate history |
hl_place_order |
Place limit or market orders |
hl_cancel_order |
Cancel specific order |
hl_cancel_all_orders |
Cancel all orders (optional: by coin) |
hl_close_position |
Close entire position |
hl_set_leverage |
Set leverage (cross or isolated) |
🚀 Quick Start
1. Install
pip install hyperliquid-mcp
Or run directly without installing:
uvx hyperliquid-mcp
2. Configure Claude Desktop
Open ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows) and add:
Read-only mode (positions, prices, history — no private key needed):
{
"mcpServers": {
"hyperliquid": {
"command": "uvx",
"args": ["hyperliquid-mcp"],
"env": {
"HL_WALLET_ADDRESS": "0xYourWalletAddressHere"
}
}
}
}
Trading mode (place orders, cancel, close positions):
{
"mcpServers": {
"hyperliquid": {
"command": "uvx",
"args": ["hyperliquid-mcp"],
"env": {
"HL_PRIVATE_KEY": "0xYourPrivateKeyHere"
}
}
}
}
⚠️ Never share your private key. It stays on your machine — this server runs locally via stdio, no data is sent anywhere except Hyperliquid's official API.
3. Restart Claude Desktop and start chatting
What are my open positions?
What's the ETH funding rate this week?
Place a limit order to buy 0.5 SOL at $150
🔧 Manual Setup (from source)
git clone https://github.com/erscoder/hyperliquid-mcp
cd hyperliquid-mcp
pip install -e .
cp .env.example .env
# Edit .env with your wallet address or private key
Then in claude_desktop_config.json:
{
"mcpServers": {
"hyperliquid": {
"command": "python",
"args": ["-m", "hyperliquid_mcp"],
"cwd": "/path/to/hyperliquid-mcp",
"env": {
"HL_WALLET_ADDRESS": "0xYourWalletAddressHere"
}
}
}
}
🔒 Security
- Your keys never leave your machine. The server runs locally via stdio transport.
- Read-only by default. Set only
HL_WALLET_ADDRESSif you don't want trading enabled. - Trading requires
HL_PRIVATE_KEY. Without it, all trading tools return an error. - Use a dedicated trading wallet with limited funds for extra safety.
🤖 Use in AI Agents
MCP is not just for chat clients. Use hyperliquid-mcp as the trading layer inside any autonomous agent.
LangChain / LangGraph
from langchain_mcp_adapters.client import MultiServerMCPClient
client = MultiServerMCPClient({
"hyperliquid": {
"command": "uvx",
"args": ["hyperliquid-mcp"],
"transport": "stdio",
"env": {"HL_WALLET_ADDRESS": "0xYourWallet"}
}
})
tools = await client.get_tools()
# tools now includes hl_get_user_state, hl_get_orderbook, etc.
# Pass them to any LangChain agent or LangGraph node
CrewAI
from crewai import Agent
from crewai_tools import MCPTool
hl_tools = MCPTool.from_server(
command="uvx",
args=["hyperliquid-mcp"],
env={"HL_WALLET_ADDRESS": "0xYourWallet"}
)
trader = Agent(
role="Trading Analyst",
goal="Monitor Hyperliquid positions and funding rates",
tools=hl_tools
)
Custom agent (MCP Python SDK)
from mcp import ClientSession, StdioServerParameters
from mcp.client.stdio import stdio_client
server_params = StdioServerParameters(
command="uvx",
args=["hyperliquid-mcp"],
env={"HL_WALLET_ADDRESS": "0xYourWallet"}
)
async with stdio_client(server_params) as (read, write):
async with ClientSession(read, write) as session:
await session.initialize()
result = await session.call_tool("hl_get_user_state", {})
print(result.content)
Agent ideas:
- Portfolio monitor that alerts when funding rates spike above threshold
- Risk manager that auto-closes positions when drawdown exceeds limit
- Arbitrage scanner comparing funding across assets
- Morning briefing agent that summarizes overnight PnL and open positions
💬 Other MCP Clients
Works with any MCP-compatible client:
- Claude Desktop — see Quick Start above
- VS Code (Copilot) — add to
.vscode/mcp.json - Cursor — add to MCP settings
- Continue.dev — add to config
📄 License
MIT — free to use, fork, and contribute.
🌟 Star History
If this saved you time, a ⭐ goes a long way!
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。