tasty-agent: A TastyTrade MCP Server

A Model Context Protocol server for TastyTrade brokerage accounts. Enables LLMs to monitor portfolios, analyze positions, and execute trades. Features AI-powered trade idea generation, automated IV analysis prompts, and built-in rate limiting (2 requests/second) to prevent API errors.

Authentication

OAuth Setup:

Create an OAuth app at https://my.tastytrade.com/app.html#/manage/api-access/oauth-applications
Check all scopes, save your client ID and client secret
Create a "New Personal OAuth Grant" in your OAuth app settings (check all scopes)
Copy the generated refresh token
Configure the MCP server with your credentials (see Usage section below)

MCP Tools

Account & Portfolio

get_balances() - Account balances and buying power
get_positions() - All open positions with current values
get_net_liquidating_value_history(time_back='1y') - Portfolio value history ('1d', '1m', '3m', '6m', '1y', 'all')
get_transaction_history(days=90, underlying_symbol=None, transaction_type=None) - All transactions: trades + cash flows (default: last 90 days, transaction_type: 'Trade' or 'Money Movement')
get_order_history(days=7, underlying_symbol=None) - Order history including filled, canceled, and rejected orders (default: last 7 days)

Market Data & Research

get_quotes(instruments, timeout=10.0) - Real-time quotes for multiple stocks and/or options via DXLink streaming
get_greeks(options, timeout=10.0) - Greeks (delta, gamma, theta, vega, rho) for multiple options via DXLink streaming
get_market_metrics(symbols) - IV rank, percentile, beta, liquidity for multiple symbols
market_status(exchanges=['Equity']) - Market hours and status ('Equity', 'CME', 'CFE', 'Smalls')
search_symbols(symbol) - Search for symbols by name/ticker
get_current_time_nyc() - Current time in New York timezone (market time)

Order Management

get_live_orders() - Currently active orders
place_order(legs, price=None, time_in_force='Day', dry_run=False) - Place multi-leg orders with automatic price discovery from market quotes
- Stock actions: 'Buy', 'Sell'
- Option actions: 'Buy to Open', 'Buy to Close', 'Sell to Open', 'Sell to Close'
replace_order(order_id, price) - Modify existing order price (for complex changes, cancel and place new order)
delete_order(order_id) - Cancel orders by ID

Watchlist Management

get_watchlists(watchlist_type='private', name=None) - Get watchlists ('public'/'private', all if name=None)
manage_private_watchlist(action, symbols, name='main') - Add/remove multiple symbols from private watchlists
delete_private_watchlist(name) - Delete private watchlist

AI-Powered Analysis

generate_trade_ideas(focus_symbols=None, risk_tolerance='moderate', max_ideas=5) - Generate specific, actionable trade ideas using AI analysis of current positions, watchlists, market metrics, and volatility environment

MCP Prompts

IV Rank Analysis - Automated prompt to analyze IV rank extremes across positions and watchlists for entry/exit opportunities

Key Features

Smart Order Placement

Automatic price calculation from real-time market quotes when no price specified
Multi-leg options strategies (spreads, strangles, etc.) with single function call
Dry-run mode for testing orders without execution

AI-Powered Trading Assistant

Trade Idea Generation: Uses AI to analyze current positions, watchlists, market metrics, and volatility to generate specific actionable trade ideas
IV Analysis Prompts: Pre-built prompts to identify extreme IV rank conditions for entry/exit opportunities
Risk-adjusted suggestions based on conservative/moderate/aggressive tolerance levels

Rate Limiting & Reliability

Built-in rate limiting (2 requests/second) prevents API throttling
Comprehensive error handling and logging

MCP Client Configuration

Add to your MCP client configuration (e.g., claude_desktop_config.json):

{
  "mcpServers": {
    "tastytrade": {
      "command": "uvx",
      "args": ["tasty-agent"],
      "env": {
        "TASTYTRADE_CLIENT_SECRET": "your_client_secret",
        "TASTYTRADE_REFRESH_TOKEN": "your_refresh_token",
        "TASTYTRADE_ACCOUNT_ID": "your_account_id"
      }
    }
  },
  "capabilities": {
    "sampling": {}
  }
}

Note: The sampling capability enables AI-powered trade analysis features. Currently supported in VS Code, not yet in Claude Desktop.

Examples

"Get my account balances and current positions"
"Get real-time quotes for SPY and AAPL"
"Get quotes for TQQQ C option with strike 100 expiring 2026-01-16"
"Get Greeks for AAPL P option with strike 150 expiring 2024-12-20"
"Buy 100 AAPL shares" (auto-pricing)
"Buy 100 AAPL at $150"
"Buy to open 17 TQQQ calls, strike 100, exp 2026-01-16"
"Place a call spread: buy to open AAPL 150C and sell to open AAPL 155C, both exp 2024-12-20"
"Close my AAPL position: sell to close 10 AAPL calls"
"Modify order 12345 to price $10.05"
"Cancel order 12345"
"Get my trading history from January"
"Get my private watchlists"
"Add TSLA and NVDA to my tech watchlist"
"Remove AAPL from my tech watchlist"
"Generate trade ideas for my portfolio"
"Generate aggressive trade ideas for AAPL and TSLA"

Background Trading Bot

Run automated trading strategies:

# Run once with instructions
uv run background.py "Check my portfolio and rebalance if needed"

# Run every hour
uv run background.py "Monitor SPY and alert on significant moves" --hourly

# Run every day
uv run background.py "Generate daily portfolio summary" --daily

# Custom period (seconds)
uv run background.py "Scan for covered call opportunities" --period 1800  # every 30 minutes

# Schedule start time (NYC timezone)
uv run background.py "Execute morning trading strategy" --schedule "9:30am" --hourly

# Market open shorthand (9:30am)
uv run background.py "Buy the dip strategy" --market-open --hourly

Development

Testing with chat.py

For interactive testing during development:

# Set up environment variables in .env file:
# TASTYTRADE_CLIENT_SECRET=your_secret
# TASTYTRADE_REFRESH_TOKEN=your_token
# TASTYTRADE_ACCOUNT_ID=your_account_id (defaults to the first account)
# OPENAI_API_KEY=your_openai_key (you can provide alternative provider of your choice as supported by pydantic-ai)
# MODEL_IDENTIFIER=model_provider:model_name (defaults to openai:gpt-5-mini)


# Run the interactive client
uv run chat.py

The client provides a chat interface to test MCP tools directly. Example commands:

"Get my account balances"
"Get quote for SPY"
"Place dry-run order: buy 100 AAPL at $150"

Debug with MCP inspector

npx @modelcontextprotocol/inspector uvx tasty-agent

License

MIT