NewFree course: Learn how to runRun Amazon with Claude
DataDoe for AI builders

Build your own Amazon tools with the AI you already use.

Point Claude Code, Cursor, Codex or n8n at one MCP endpoint and describe what you need. Your AI finds the right Seller Central, Vendor Central and Ads tables, then writes the script, dashboard or alert.

Start free trial Get the MCP setup
  • One MCP endpoint
  • 130+ Amazon tables
  • 14-day free trial
• Under the hood

Real tool names, real tables, nothing to guess

DataDoe MCP gives your AI a short set of tools and a published schema of 130+ Amazon tables. It searches for the table, loads its columns, pulls the rows and only then writes code, so the script uses columns that exist instead of ones it made up.

• Connect

One endpoint, every builder you use

Chat apps and Cursor sign in to DataDoe. Coding agents, automations and SDKs send the same address with a datadoe-mcp-key header. Either way the tools and tables are identical.

Claude Code

Sign in

terminal

claude mcp add datadoe \
  "https://mcp.datadoe.com/mcp/v1" \
  --transport http

Then run /mcp and choose Authenticate. If DataDoe is already a connector in your Claude account, Claude Code has it.

Claude Code setup

Cursor

Sign in

.cursor/mcp.json

{
  "mcpServers": {
    "datadoe": {
      "url": "https://mcp.datadoe.com/mcp/v1"
    }
  }
}

Cursor opens a DataDoe sign-in. Cloud Agents take the same address.

Cursor setup

Codex

MCP key

~/.codex/config.toml

[mcp_servers.datadoe]
url = "https://mcp.datadoe.com/mcp/v1"
http_headers = { "datadoe-mcp-key" = "…" }

Create the key under Integrations → MCP. Keep it in the home-level file, not in the repo.

Codex setup

n8n

MCP key

MCP Client Tool

Endpoint   https://mcp.datadoe.com/mcp/v1
Transport  HTTP Streamable
Auth       Header Auth
Name       datadoe-mcp-key

Add the tool to an AI Agent node. Schedule or webhook triggers run it without you.

n8n setup

Claude and ChatGPT

Sign in

connector URL

https://mcp.datadoe.com/mcp/v1

In Claude: Customize → Connectors → Add. In ChatGPT, add DataDoe as an app. Both sign in, no key.

ClaudeChatGPT

Any MCP client or agent SDK

MCP key

remote server

URL        https://mcp.datadoe.com/mcp/v1
Transport  Streamable HTTP
Header     datadoe-mcp-key

Gemini CLI, VS Code, the OpenAI Agents SDK and the Claude Agent SDK have their own guides in the docs.

Every client
• Access

Your build reads. You decide what it writes.

Access is set per Amazon account, not per key or per tool. A script someone vibe-coded on Friday can’t do more than the account allows, and nothing changes on Amazon until a write type is switched on.

  • Each Amazon account is read only or read and write.
  • Every Action type starts off. An owner turns them on one by one.
  • Most AI clients ask before a change. Dry runs are optional.
  • Every change is logged with what Amazon answered.

Buyer names and addresses stay out unless you add buyer data (PII), $49/mo on request. How Actions work

• Shipping it

From one prompt to a tool your team runs daily

The first version usually lives in a chat or an editor. When people start relying on it, it moves to a schedule, an Amazon event or a shared dashboard, still on the same tables.

  1. 01

    Ask it

    MCP in your AI client

    Explore and prototype in Claude, ChatGPT, Cursor or Claude Code. The agent exports what it needs as it goes.

    https://mcp.datadoe.com/mcp/v1

    Per export2 AI Tokens

  2. 02

    Schedule it

    REST API from your code

    Move the script to a server or CI job. Same tables, an API key, and writes through the Actions endpoint.

    GET /api/v1/util/sellers-and-vendors
POST /api/v1/actions
datadoe-api-key: …

    Per Action, 100 changes3 AI Tokens

  3. 03

    Trigger it

    Webhooks to your endpoint

    Let Amazon events start the work: a listing status change, an inventory change, an order update.

    AMAZON_LISTINGS_ITEM_STATUS_CHANGE
→ POST https://your-app/hooks

    Per message delivered0.05 AI Tokens

  4. 04

    Share it

    BigQuery for the team

    Point Looker Studio, a notebook or your warehouse at the same tables, so the dashboard and the agent agree.

    SELECT sku, SUM(profit)
FROM amazon_profit_by_sku_and_date
GROUP BY sku

    Per query0.2 AI Tokens

Exports from premium tables cost 5. The Base Plan includes 2,000 AI Tokens a month, and most sellers use well under 1,000. Token pricing

• On camera

Six builds, prompt to working tool

Each video starts with a request in plain English and ends with a dashboard, report or Slack alert running on Seller Central or Vendor Central data. Nothing is cut between the two.

All builds on YouTube

FBA restock planner

Claude Code
FAQ

AI builders, answered.

A hosted MCP server and a REST API over your Amazon Seller Central, Vendor Central and Amazon Ads data, joined into one schema of 130+ tables. Claude Code, Cursor, Codex, n8n, ChatGPT, Claude or any other MCP client can find a table, read its columns, export rows and, once you allow it, start Actions that change things on Amazon. You describe the tool you want and your AI writes it on top of that data.

No. If you can paste a URL into Claude or ChatGPT, or a short config into Cursor or Codex, you can connect. After that you describe the dashboard, alert or report in plain English and the AI writes and runs the code. It helps to read what it built, and you can ask it to explain any part before you rely on it.

All of them use the same MCP address, https://mcp.datadoe.com/mcp/v1. Claude, ChatGPT and Cursor sign in to DataDoe with OAuth, so no key is needed, and Claude Code can sign in or use a key. Codex, n8n, agent SDKs and other clients send a datadoe-mcp-key header with a key you create under Integrations → MCP.

Seller Central, Vendor Central and Amazon Ads: orders, profit by SKU, settlements, FBA inventory and restock data, listings, Brand Analytics, Vendor purchase orders and more, over 100 tables in all. Data is fresh daily and key tables update intraday, with about two years of history backfilled when you connect. The Data Scheme lists every table and column.

Neither. There is no DataDoe SDK to install and no Amazon developer registration on your side. Your AI calls the MCP tools directly, or writes plain HTTP calls against the REST API using the OpenAPI spec and the API guide we publish for coding agents. Amazon authorization, report polling, retries and history are handled by DataDoe.

Yes, through Actions: bids, budgets, listings, prices, A+ Content, orders and Vendor purchase orders. Every Action type is off until an organization owner switches it on, and the Amazon account must be set to read and write. Most AI clients ask before a change, you can preview it as a dry run, and every change is logged with Amazon's answer.

Access is set per Amazon account: read only, or read and write. It is not scoped per key or per field, so give a key only to people and tools you would trust with that account. Buyer names and addresses are excluded unless you add buyer data (PII), a $49/mo add-on available on request.

The Base Plan is $97 a month with 2,000 AI Tokens, after a 14-day free trial. An export costs 2 AI Tokens (5 from a premium table), an Action 3 tokens for up to 100 changes, a BigQuery query 0.2 and a webhook message 0.05. Most sellers use well under 1,000 tokens a month, and nothing is charged beyond your tokens unless you opt in.

DataDoe does not use your data to train models. Your prompts and the rows your AI reads are processed by the AI client you connect, under that provider's terms, so check its data and training settings before you connect a business account, especially on personal plans.

Building something specific?

Book a call and we'll connect DataDoe to your builder of choice and walk through the tables your idea needs.

Book a demo
• Get started

Describe the tool. Your AI builds it.

Connect Seller Central, Vendor Central and Amazon Ads once, add one MCP address to the builder you use, and start with a question. Writes stay off until you switch them on.

Start your free trial MCP docs
  • Claude Code, Cursor, Codex, n8n
  • 130+ Amazon tables
  • Read or read and write per account
  • $97/mo, 14-day free trial