What is DataDoe MCP?
DataDoe MCP is a hosted MCP server that gives AI agents, assistants, and applications access to your Amazon data from Seller Central, Vendor Central, and Amazon Ads. It works with all major MCP clients, including chat apps such as ChatGPT and Claude, and development tools such as Claude Code, Cursor, and Codex.
Why would I use DataDoe MCP?
DataDoe MCP turns your Amazon Seller, Vendor, and Ads data into a live data source for AI. Instead of exporting CSVs or maintaining your own SP-API and Ads API pipelines, connect once and let your AI agent pull SKU-level sales, ads, inventory, and traffic data on demand, in natural language. Connect your Amazon accounts, add DataDoe MCP to your AI client, and create insights and apps in minutes instead of weeks.
What can I do with DataDoe MCP?
Once DataDoe MCP is connected, it becomes a live data layer for your AI agent. You can use it to:
- Explore your Amazon data in plain English: ask questions about sales, ads, traffic, inventory, and listings without writing SQL or pulling exports.
- Generate reports and reconciliations on demand: let your AI agent build sales summaries, ad performance reviews, or settlement-vs-order reconciliations whenever you need them.
- Vibe-code custom dashboards and tools: scaffold KPI dashboards, internal apps, or one-off analyses directly inside your IDE, with live Amazon data connected from the start.
- Automate recurring workflows: use CrewAI, n8n, or any MCP-capable automation tool to schedule AI-generated briefings, alerts, and exports straight to Slack, email, or your file system.
- Build custom AI agents on top of Amazon data: ship client-facing assistants, internal copilots, or agency-grade reporting bots without writing your own SP-API and Ads API integration.
- Bring Amazon into the AI tools you already use: work with your data inside the same agent you use for code, docs, or operations, instead of context-switching between platforms.
- Manage sellers and vendors: agents can create invitation links for new accounts, send reconnect links when Amazon stops accepting an account’s authorization, and rename sellers and vendors. See Sellers and vendors API for the matching MCP tools.
- Run Amazon Marketing Cloud queries: agents can run ready-made AMC query templates (new-to-brand, keyword halo, path to conversion, customer value, and more) or their own SQL, schedule them as workflows, poll results, and download files. AMC is in beta and works with AMC accounts and self-service AMC; see Amazon Marketing Cloud.
Learn more in our video demo:
What data can I use with DataDoe MCP?
DataDoe MCP supports all data available in DataDoe. Review the tables, columns, and refresh cadences in our interactive Data Scheme.
Agents should search with exports_sources_get using Amazon or business keywords, inspect whyMatched, then load columns with exports_source_get before exports_create. exports_source_get returns at most 40 columns per call. If columnsMeta.hasNextPage is true, more columns exist and the schema is incomplete until those pages are loaded. If the first source-search page is wrong, retry with table names or paginate. An empty recommendation list does not mean the data is missing.
What does it cost?
DataDoe MCP is included with every DataDoe subscription. Usage counts against your monthly AI Tokens. See Subscription and pricing for details.
How do I use DataDoe MCP with ChatGPT or Claude?
If you are using ChatGPT or Claude, you can connect DataDoe MCP directly in your assistant. Both sign in to DataDoe with OAuth, so you do not need to create an MCP key. Your role must allow MCP OAuth. For details, see the ChatGPT and Claude pages.
A connector added in Claude follows you across Claude surfaces. It is available in the Claude web app, Claude Desktop, the Microsoft 365 add-in, and Claude Code. One sign-in covers them all.
You can also sign in to DataDoe from Codex, Genspark, Superhuman, Muse, Glama, Littlebird, Cursor, GrokBot, Grok, and Antigravity. Follow their guides below.
How do I connect DataDoe MCP using other MCP clients?
If your Organization has a white-label domain prefix, copy the full server URL from Integrations > MCP (opens in a new tab). For example, total-store uses https://total-store.mcp.datadoe.com/mcp/v1. Hub examples use https://mcp.datadoe.com/mcp/v1; substitute your Organization’s server URL when following a guide. Authentication stays the same, except for Codex sign-in, which accepts only the standard server address. For a white-label address in Codex, use an MCP key. Update your client configuration after changing or clearing the prefix.
For other MCP clients, follow the dedicated guides:
Coding agents and IDEs
- Using Claude Code
- Using Codex and Codex Sites
- Using Cursor
- Using Antigravity
- Using Gemini CLI
- Using VS Code
- Using OpenCode
Autonomous agents and automation
Office apps
Agent SDKs
If your MCP client is not listed, add DataDoe MCP as a generic remote server:
- URL:
https://mcp.datadoe.com/mcp/v1 - Transport: Streamable HTTP. The server does not offer a legacy SSE endpoint. If your client only supports SSE or stdio, bridge it with
mcp-remote(opens in a new tab). - Auth header:
datadoe-mcp-key: YOUR_DATADOE_MCP_KEY(create a key in DataDoe MCP Integrations (opens in a new tab))
1{
2 "mcpServers": {
3 "datadoe": {
4 "url": "https://mcp.datadoe.com/mcp/v1",
5 "headers": {
6 "datadoe-mcp-key": "YOUR_DATADOE_MCP_KEY"
7 }
8 }
9 }
10}Limit access with roles and service accounts
Personal MCP keys and OAuth connections follow their owner's current roles and account, Action, and data table access. Service account keys use their configured permissions and resource assignments, limited by their creator's current access. Catalog names and schemas remain visible, while using a resource or reading its saved results requires access to it and the selected accounts.
Personal data in tables, Action requests and results, and generated Files requires Access PII data and Amazon PII Data Access for the Organization. All tables or All Actions access does not bypass this check. New non-Admin Roles and Service accounts start with the permission disabled. The Organization requirement also applies to Admins.
To create an MCP key, open Integrations → MCP (opens in a new tab) and select New key. Choose a key type, enter a name, and choose when it expires. For a service account key, select an active service account. Copy the key before closing the dialog; its full value is shown only once.
Your role must allow Personal MCP keys to create a personal key. Admins can create keys for any Service account. Other Users need Manage service accounts and can create keys only for Service accounts they created. DataDoe sends email reminders before keys expire. Select a service account's name in the keys table to open its settings.
See Manage API keys, MCP keys, and service accounts for key management. See Manage roles and access to limit a connection to selected Groups, accounts, Actions, and data tables.
Contact DataDoe support
Ask your AI agent to use datadoe_support_ticket_create to send a request, datadoe_support_ticket_check to read a reply, or datadoe_support_ticket_update to add information. Your connection must use a person's account with Contact support via MCP permission. Service accounts cannot use these tools.
Read only lets you check tickets. Read & write also lets you create or update them.
Only the ticket creator and current Organization Admins can read or update a ticket. Older tickets without a recorded creator are available only to Admins. Adding someone as an email contact does not give them access through these tools.
Optional contactEmails must belong to active members of your Organization. These contacts receive the full submitted ticket and follow-up text by email, including related details. DataDoe checks membership before sending each notification. External addresses, inactive users, and people who have left the Organization do not receive ticket emails, even if their addresses remain on an older ticket.

