MetaTrader 5 MCP Server
A Model Context Protocol (MCP) server for MetaTrader 5, allowing AI assistants to interact with the MetaTrader 5 platform for trading and market data analysis. Documentation
Features
- Connect to MetaTrader 5 terminal
- Access market data (symbols, rates, ticks)
- Place and manage trades
- Analyze trading history
- Integrate with AI assistants through the Model Context Protocol
Installation
From PyPI
uvx --from mcp-metatrader5-server mt5mcp
From Source
git clone https://github.com/Qoyyuum/mcp-metatrader5-server.git
cd mcp-metatrader5-server
uv sync
uv run mt5mcp
Requirements
- uv (recommended) or pip
- Python 3.11 or higher
- MetaTrader 5 terminal installed on Windows
- MetaTrader 5 account (demo or real)
Usage
Quick Start
The server runs in stdio mode by default for MCP clients like Claude Desktop:
uv run mt5mcp
Development Mode (HTTP)
For testing with HTTP transport, create a .env file:
MT5_MCP_TRANSPORT=http
MT5_MCP_HOST=127.0.0.1
MT5_MCP_PORT=8000
Then run:
uv run mt5mcp
The server will start at http://127.0.0.1:8000
Installing for MCP Clients
Method 1: Using uvx (Simplest - No Installation Required) ⭐
Add this configuration to your MCP client's config file:
For Claude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"mcp-metatrader5-server": {
"command": "uvx",
"args": [
"--from",
"git+https://github.com/Qoyyuum/mcp-metatrader5-server",
"mt5mcp"
]
}
}
}
Method 2: Using FastMCP Install (Recommended)
git clone https://github.com/Qoyyuum/mcp-metatrader5-server
cd mcp-metatrader5-server
After git cloning the repo, run the following commands:
For MCP JSON format
uv run fastmcp install mcp-json src/mcp_mt5/main.py
For Claude Desktop
uv run fastmcp install claude-desktop src/mcp_mt5/main.py
For Claude Code
uv run fastmcp install claude-code src/mcp_mt5/main.py
For Cursor
uv run fastmcp install cursor src/mcp_mt5/main.py
For Gemini CLI
uv run fastmcp install gemini-cli src/mcp_mt5/main.py
Method 3: Manual Configuration
Add this to your claude_desktop_config.json or whatever LLM config file:
{
"mcpServers": {
"mcp-metatrader5-server": {
"command": "uvx",
"args": [
"--from",
"mcp-metatrader5-server",
"mt5mcp"
]
}
}
}
API Reference
Connection Management
initialize(): Initialize the MT5 terminallogin(account, password, server): Log in to a trading accountshutdown(): Close the connection to the MT5 terminal
Market Data Functions
get_symbols(): Get all available symbolsget_symbols_by_group(group): Get symbols by groupget_symbol_info(symbol): Get information about a specific symbolget_symbol_info_tick(symbol): Get the latest tick for a symbolcopy_rates_from_pos(symbol, timeframe, start_pos, count): Get bars from a specific positioncopy_rates_from_date(symbol, timeframe, date_from, count): Get bars from a specific datecopy_rates_range(symbol, timeframe, date_from, date_to): Get bars within a date rangecopy_ticks_from_pos(symbol, start_pos, count): Get ticks from a specific positioncopy_ticks_from_date(symbol, date_from, count): Get ticks from a specific datecopy_ticks_range(symbol, date_from, date_to): Get ticks within a date range
Trading Functions
order_send(request): Send an order to the trade serverorder_check(request): Check if an order can be placed with the specified parameterspositions_get(symbol, group): Get open positionspositions_get_by_ticket(ticket): Get an open position by its ticketorders_get(symbol, group): Get active ordersorders_get_by_ticket(ticket): Get an active order by its tickethistory_orders_get(symbol, group, ticket, position, from_date, to_date): Get orders from historyhistory_deals_get(symbol, group, ticket, position, from_date, to_date): Get deals from history
Example Workflows
Connecting and Getting Market Data
# Initialize MT5
initialize()
# Log in to your trading account
login(account=123456, password="your_password", server="your_server")
# Get available symbols
symbols = get_symbols()
# Get recent price data for EURUSD
rates = copy_rates_from_pos(symbol="EURUSD", timeframe=15, start_pos=0, count=100)
# Shut down the connection
shutdown()
Placing a Trade
# Initialize and log in
initialize()
login(account=123456, password="your_password", server="your_server")
# Create an order request
request = OrderRequest(
action=mt5.TRADE_ACTION_DEAL,
symbol="EURUSD",
volume=0.1,
type=mt5.ORDER_TYPE_BUY,
price=mt5.symbol_info_tick("EURUSD").ask,
deviation=20,
magic=123456,
comment="Buy order",
type_time=mt5.ORDER_TIME_GTC,
type_filling=mt5.ORDER_FILLING_IOC
)
# Send the order
result = order_send(request)
# Shut down the connection
shutdown()
Resources
The server provides the following resources to help AI assistants understand how to use the MetaTrader 5 API:
mt5://getting_started: Basic workflow for using the MetaTrader 5 APImt5://trading_guide: Guide for placing and managing tradesmt5://market_data_guide: Guide for accessing and analyzing market datamt5://order_types: Information about order typesmt5://order_filling_types: Information about order filling typesmt5://order_time_types: Information about order time typesmt5://trade_actions: Information about trade request actions
Prompts
The server provides the following prompts to help AI assistants interact with users:
connect_to_mt5(account, password, server): Connect to MetaTrader 5 and log inanalyze_market_data(symbol, timeframe): Analyze market data for a specific symbolplace_trade(symbol, order_type, volume): Place a trade for a specific symbolmanage_positions(): Manage open positionsanalyze_trading_history(days): Analyze trading history
Development
Project Structure
mcp-metatrader5-server/
├── src/
│ └── mcp_mt5/
│ ├── __init__.py # Entry point with main()
│ ├── main.py # FastMCP server with all tools
│ └── test_client.py # Test client for development
├── docs/
│ ├── getting_started.md
│ ├── market_data_guide.md
│ ├── trading_guide.md
│ └── publishing.md
├── .env # Environment configuration (create from .env.example)
├── README.md
├── pyproject.toml # Project metadata (using hatchling)
└── uv.lock # Dependency lock file
Building the Package
Using uv (recommended):
uv build
This will create wheel and source distributions in the dist/ directory.
Publishing to PyPI
Using uv:
# Build first
uv build
# Publish to PyPI
uv publish
# Or publish to TestPyPI first
uv publish --publish-url https://test.pypi.org/legacy/
License
MIT
Acknowledgements
- MetaQuotes for the MetaTrader 5 platform
- FastMCP for the MCP server implementation