Skip to main content

A Model Context Protocol server for interacting with TastyTrade brokerage accounts

Project description

tasty-agent: A TastyTrade MCP Server

Trust Score

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

Authentication

OAuth Setup:

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

MCP Tools

Account & Portfolio

  • account_overview(include=["balances","positions"]) - Account balances (including net liquidating value) and open positions.

Market Data & Research

  • get_quotes(instruments, timeout=10.0) - Real-time quotes for stocks, options, futures, and indices via DXLink streaming
  • get_greeks(options, timeout=10.0) - Greeks (delta, gamma, theta, vega, rho) for equity and futures options via DXLink streaming
  • get_market_metrics(symbols) - IV rank, percentile, beta, liquidity for multiple symbols
  • market_status(exchanges=['Equity']) - Market hours, status, holidays, and current NYC time ('Equity', 'CME', 'CFE', 'Smalls')
  • search_symbols(symbol, limit=10) - Search for symbols by name/ticker

History

  • get_history(type, days=None, underlying_symbol=None, transaction_type=None, page_offset=0, limit=25) - Transaction history (type="transactions", default 90 days) or order history (type="orders", default 7 days). Paginated — use page_offset and limit for large result sets. Filter transactions by "Trade" or "Money Movement".

Order Management

  • place_order(legs, target_value=None, time_in_force="Day", dry_run=false) - Place multi-leg orders with quote-derived mid pricing only. The tool fetches live quotes for the exact resolved instruments, computes the signed net mid, validates the final limit against bid/ask guardrails, and optionally sizes quantity from target_value.
    • quantity is the actual share/contract count. target_value=50000 sizes an equity or equity-option order from quote-derived pricing; omit quantity for single-leg target-value orders. For multi-leg spreads with target_value, use quantity only to express the leg ratio, such as 1:1 or 2:1.
    • Order prices are aligned to the broker's valid tick grid before submission. Option orders require tastytrade option tick-size data; if it is unavailable, the tool fails before placement instead of submitting an invalid price increment.
    • Order actions follow the tastytrade Python SDK contract: equities and options use Buy to Open, Buy to Close, Sell to Open, or Sell to Close; futures use Buy or Sell.
  • replace_order(order_id) - Reprice an existing live order at the current quote-derived mid.
  • cancel_order(order_id) - Cancel an order.
  • list_orders() - Get all live orders.
  • Tool outputs are compact: quote tables include actionable bid/ask/mid/size fields; order results include compact order, buying-power, fee, warning/error, and sizing summaries.

Watchlist Management

  • watchlist(action, ...) - Unified watchlist management:
    • action="list" - No name returns compact watchlist metadata (name, group, symbol_count); with name, returns compact symbol entries.
    • action="add" - Add symbols to a watchlist (creates if doesn't exist)
    • action="remove" - Remove symbols from a watchlist
    • action="delete" - Delete a watchlist

MCP Prompts

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

Usage

Local (stdio)

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"
      }
    }
  }
}

Remote (Modal)

Deploy as a remote MCP server on Modal with proxy auth:

uvx modal setup

# Create secrets
uvx modal secret create tasty-agent-secrets \
  TASTYTRADE_CLIENT_SECRET=your_secret \
  TASTYTRADE_REFRESH_TOKEN=your_token

# Create a proxy auth token at https://modal.com/settings/proxy-auth-tokens

# Export your workspace host
export MODAL_HOST=<workspace>--tasty-agent-mcp-server.modal.run

# Deploy
uvx modal deploy examples/modal_deploy.py

Clients authenticate with Modal-Key and Modal-Secret headers. See examples/modal_deploy.py.

Programmatic (Python client)

Connect to a remote tasty-agent and call tools directly:

# List all tools
uv run examples/mcp_client.py

# Call a tool
uv run examples/mcp_client.py market_status
uv run examples/mcp_client.py get_market_metrics '{"symbols": ["AAPL", "SPY"]}'

See examples/mcp_client.py for the full client code.

Examples

"Get my account balances and current positions"
"What's my net liquidating value?"
"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"
"Get Greeks for /ES C option with strike 5800 expiring 2026-05-30"
"Buy to open 100 AAPL shares at mid"
"Buy to open 17 TQQQ calls, strike 100, exp 2026-01-16"
"Buy $50K of TSLA calls, strike 300, exp 2026-01-16"
"Place a call spread: buy to open AAPL 150C and sell to open AAPL 155C, both exp 2024-12-20"
"Buy 1 /ESM26 future at mid"
"Reprice order 12345 at mid"
"Cancel order 12345"
"Show my live orders"
"Get my trading history from January"
"Get my order history for SPY"
"Get my private watchlists"
"Add TSLA and NVDA to my tech watchlist"

Development

# Run tests
uv run pytest

# Interactive chat client (requires .env with credentials + OPENAI_API_KEY)
uv run examples/chat.py

# Debug with MCP inspector
npx @modelcontextprotocol/inspector uvx tasty-agent

License

MIT

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

tasty_agent-6.0.1.tar.gz (137.5 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

tasty_agent-6.0.1-py3-none-any.whl (26.8 kB view details)

Uploaded Python 3

File details

Details for the file tasty_agent-6.0.1.tar.gz.

File metadata

  • Download URL: tasty_agent-6.0.1.tar.gz
  • Upload date:
  • Size: 137.5 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for tasty_agent-6.0.1.tar.gz
Algorithm Hash digest
SHA256 96ab95645f0dc3402a4458ad8b170c87b2596282fae724d5db991cf66ea5e58d
MD5 aebee504907a420fa2358b8c22f0ceae
BLAKE2b-256 a7eb8847ae515a477507afdce97065e5cc99a23e1eb79ec77c0976f30c5add7f

See more details on using hashes here.

File details

Details for the file tasty_agent-6.0.1-py3-none-any.whl.

File metadata

  • Download URL: tasty_agent-6.0.1-py3-none-any.whl
  • Upload date:
  • Size: 26.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.12.13

File hashes

Hashes for tasty_agent-6.0.1-py3-none-any.whl
Algorithm Hash digest
SHA256 6bbf619d449de45acf4e2ce7672520b7fecf09772ac3dc3bb188267e0714d9e5
MD5 6bcbeb969aef9023c9d4f36f4e6c9d91
BLAKE2b-256 d6b0affb4cde7efdf3c3f68f8bdd5369d07be60fcca18398ec2087745a1977dc

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page