AI Trading Co-Pilot MCP Server

AI Trading Co-Pilot MCP Server

AI-powered cryptocurrency trading assistant for OKX exchange, featuring a multi-agent analysis system, real-time market data, and autonomous trading capabilities.

Category
Visit Server

README

#AI Trading Co-Pilot — MCP Server

AI-powered cryptocurrency trading assistant for OpenClaw and any MCP-compatible client (Claude Desktop, etc.).

4-agent analysis system (macro + technical + risk + portfolio manager) with real-time OKX exchange integration, smart rules engine, and autonomous trading capabilities.

⚡ Quick Start (3 minutes)

1. Install

npm install @okx-trade/mcp-server

Or clone and install locally:

cd mcp/
npm install

2. Configure API Keys

# OKX Exchange API Key (required for trading & account data)
export OKX_DEMO_API_KEY="your-api-key"
export OKX_DEMO_SECRET_KEY="your-secret-key"
export OKX_DEMO_PASSPHRASE="your-passphrase"

# LLM API Key (required for AI analysis)
export MINIMAX_API_KEY="your-minimax-key"

# Optional settings
export OKX_MODE="demo"            # "demo" (default) or "live"
export OKX_TRADE_CONFIRM="true"   # Require confirmation before trades

3. Start the MCP Server

node bin/okx-trade-mcp.mjs

Or add to your MCP client config:

Claude Desktop (claude_desktop_config.json):

{
  "mcpServers": {
    "okx-trade": {
      "command": "node",
      "args": ["/path/to/mcp/bin/okx-trade-mcp.mjs"],
      "env": {
        "OKX_DEMO_API_KEY": "your-key",
        "OKX_DEMO_SECRET_KEY": "your-secret",
        "OKX_DEMO_PASSPHRASE": "your-pass",
        "MINIMAX_API_KEY": "your-minimax-key",
        "OKX_MODE": "demo"
      }
    }
  }
}

OpenClaw: Install via ClawHub or add manually in Settings → MCP Servers.

4. Start Chatting

"What's the BTC price?"
"Run a full analysis on BTC"
"Show me ETH technical analysis with 4H candles"
"Open a 10x long BTC position"

🔑 Configuration

OKX API Key (Required for Trading)

  1. Log in to OKX
  2. Go to API Management → Create API Key
  3. Permissions needed: Read + Trade (do NOT enable Withdraw)
  4. Important: Do NOT set IP whitelist (MCP runs locally with dynamic IPs)

Demo mode (default): Uses OKX demo trading environment — no real money involved.

Env Variable Description
OKX_DEMO_API_KEY Demo API Key
OKX_DEMO_SECRET_KEY Demo Secret Key
OKX_DEMO_PASSPHRASE Demo Passphrase
OKX_LIVE_API_KEY Live API Key (when OKX_MODE=live)
OKX_LIVE_SECRET_KEY Live Secret Key
OKX_LIVE_PASSPHRASE Live Passphrase

LLM API Key (Required for AI Analysis)

The AI co-pilot uses MiniMax M2.5 for multi-agent analysis.

  1. Sign up at MiniMax Platform
  2. Create an API Key
  3. Set MINIMAX_API_KEY environment variable

Without LLM key: Market Data, Account, Trading, and Rules tools still work. Only AI Copilot tools require the LLM key.

Optional Settings

Env Variable Default Description
OKX_MODE demo demo or live — must explicitly set live for real trading
OKX_TRADE_CONFIRM false true = always preview before executing trades
DATA_DIR ~/.okx-trade-mcp Custom data directory for rules, prompts, config

🛠 Available Tools (19)

Market Data (4 tools) — No API key needed

Tool Description Example
okx_get_ticker Real-time price, 24h change, volume "What's the BTC price?"
okx_get_candles OHLCV candle data (1m~1D, up to 300) "Show me ETH 4H candles"
okx_technical_analysis RSI, MACD, Bollinger Bands, Moving Averages "Technical analysis for SOL"
okx_get_market_data Open Interest, Long/Short Ratio, Funding Rate "What's the BTC funding rate?"

Account (2 tools) — Requires OKX API key

Tool Description Example
okx_get_balance All currencies with USD equivalent "What's my balance?"
okx_get_positions Open positions + PnL + leverage + margin "Show my positions"

AI Analysis (3 tools) — Requires LLM key

Tool Description Example
okx_run_copilot Full 4-agent analysis → signal + conviction "Analyze BTC for me"
okx_run_single_agent Run one agent (macro/technical/risk) "What does the macro agent think about ETH?"
okx_generate_trading_plan Detailed plan: entry, SL, TP, risk/reward "Generate a trading plan"

Trading (2 tools) — Requires OKX API key

Tool Description Example
okx_place_order Market order with leverage + SL/TP "Open 10x long BTC, SL at 70000"
okx_close_position Close an existing position "Close my ETH position"

Rules (3 tools)

Tool Description Example
okx_list_rules View all hard + soft trading rules "List my rules"
okx_evaluate_rules Check if a trade passes hard rules "Can I go long BTC now?"
okx_add_rule Add rule via natural language "Block trades when funding > 0.1%"

Skills & Config (5 tools)

Tool Description Example
okx_list_skills View 6 analysis skills and status "What skills are available?"
okx_configure_skills Enable/disable skills per agent "Enable fibonacci for technical agent"
okx_get_prompts View agent system prompts "Show me the current prompts"
okx_update_prompt Customize an agent's prompt "Make technical agent focus on volume"
okx_set_llm_config Configure temperature, max tokens "Set temperature to 0.3"

💬 Usage Examples

Market Analysis

"What's the current BTC price and 24h change?"
"Show me ETH technical analysis with 4H candles"
"What's the current funding rate for BTC?"
"Get the open interest and long/short ratio for SOL"

AI Co-Pilot

"Run a full analysis on BTC"
"What does the macro agent think about ETH?"
"Run technical analysis only for SOL"
"Generate a trading plan based on the analysis"

Trading

"Open a 10x long BTC position with 10 contracts"
"Place a short ETH order with SL at 4000 and TP at 3500"
"Close my BTC-USDT-SWAP position"
"Show my current positions and P&L"

Rules Management

"Add a rule: block all trades when funding rate > 0.1%"
"Add a rule: don't go long when L/S ratio > 2"
"List my trading rules"
"Check if a BTC long trade would pass my rules"

Customization

"Show me the current agent prompts"
"Update the technical agent prompt to focus more on volume analysis"
"Enable fibonacci and candlestick-patterns skills for technical agent"
"Disable the fear-greed-index skill"
"Set LLM temperature to 0.3"

🤖 AI Agents

The co-pilot runs 4 AI agents:

Agent Role Focus
🌐 Macro Analyst Macro environment Fed policy, DXY, ETF flows, on-chain data
📊 Technical Analyst Chart analysis RSI, MACD, Bollinger Bands, trend, key levels
🛡️ Risk Manager Risk assessment Position sizing, leverage, stop-loss, max risk
👔 Portfolio Manager Final decision Synthesizes all reports → LONG/SHORT/HOLD + conviction

Workflow: Macro + Technical + Risk run in parallel → Portfolio Manager synthesizes → Final signal


⚙️ Analysis Skills (6)

Skill Description Agents
fibonacci-levels Fibonacci retracement from swing H/L Technical, Manager
candlestick-patterns Doji, Engulfing, Hammer detection Technical, Manager
volume-profile Volume distribution + POC Technical, Risk
orderflow-analysis OI changes, L/S ratio trends Technical, Risk, Manager
volatility-regime Classify Low/Normal/High/Extreme Risk, Manager
fear-greed-index Crypto market sentiment (0-100) Macro, Manager

All skills are enabled by default. Use okx_configure_skills to customize.


🎨 Canvas Dashboard

For OpenClaw users, two HTML templates are included for visual rendering:

  • dashboard.html — Main trading dashboard (price, positions, signal, chart)
  • analysis-report.html — Full analysis report (agent cards, trading plan, reasoning)

Ask the agent: "Show me the dashboard" or "Render the analysis report"


📋 Supported Instruments

Instrument SWAP SPOT
BTC-USDT
ETH-USDT
SOL-USDT
OKB-USDT

🔒 Security

  • All API keys stay local — stored in environment variables, never transmitted
  • Zero hardcoded credentials — npm package contains no secrets
  • Tool responses never include API keys — all sensitive data stripped
  • Default demo mode — must explicitly opt-in to live trading
  • Trade confirmation — optional preview before execution
  • No third-party data sharing — only connects to OKX API and LLM provider
  • Hard rules auto-enforcement — programmatic blocks cannot be bypassed

🔧 Testing

MCP Inspector

npx @modelcontextprotocol/inspector node bin/okx-trade-mcp.mjs

Claude Desktop

Add to claude_desktop_config.json and restart Claude Desktop.

OpenClaw

Install the plugin from ClawHub or add as custom MCP server.


📄 License

MIT


🌐 Links

Recommended Servers

playwright-mcp

playwright-mcp

A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.

Official
Featured
TypeScript
Magic Component Platform (MCP)

Magic Component Platform (MCP)

An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.

Official
Featured
Local
TypeScript
Audiense Insights MCP Server

Audiense Insights MCP Server

Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.

Official
Featured
Local
TypeScript
VeyraX MCP

VeyraX MCP

Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.

Official
Featured
Local
graphlit-mcp-server

graphlit-mcp-server

The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.

Official
Featured
TypeScript
Kagi MCP Server

Kagi MCP Server

An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.

Official
Featured
Python
E2B

E2B

Using MCP to run code via e2b.

Official
Featured
Neon Database

Neon Database

MCP server for interacting with Neon Management API and databases

Official
Featured
Exa Search

Exa Search

A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.

Official
Featured
Qdrant Server

Qdrant Server

This repository is an example of how to create a MCP server for Qdrant, a vector search engine.

Official
Featured