Amazon Orders MCP

MCP server for fetching historical Amazon consumer orders from multiple accounts.

Database

Orders are stored in ~/.amazon-orders-mcp/orders.db (SQLite).

• Import orders — one click; pulls anything not already in the database (skips known orders, refreshes recent ones for status/tracking)

• View — reads from the database; use the toolbar to filter by date, search, or hide cancelled

Related MCP server: Amazon Seller MCP Server

MCP tools

• sync_orders — import new orders into the database

• get_orders — read from the database (optional filters)

• list_accounts, login_account, remove_account

Setup

Already registered in ~/.cursor/mcp.json. Restart Cursor or reload MCP servers to pick it up.

First run installs dependencies automatically via run-amazon-orders-mcp.sh.

Tools

• list_accounts — saved profiles and session status

• login_account — open browser to sign in (complete 2FA manually)

• get_orders — fetch orders (accounts: "all" or comma-separated names)

• remove_account — delete a saved profile

Example

1. login_account with account_name: "personal", region: "us"

2. get_orders with accounts: "all", year: 2025, max_orders: 10

Output fields

Each row: account, order_id, item_name, order_date, order_status, tracking_number

Limitations

• Unofficial scraping — Amazon HTML changes may break parsers

• Tracking fetch adds ~2s per shipment

• US (amazon.com) is the primary target; other regions supported via region code

• Re-run login_account when sessions expire

Sessions stored in ~/.amazon-orders-mcp/profiles/ (mode 700).