How do I install DataDoe MCP?

Copy the http config for Cursor, Claude Desktop, Claude Code, or VS Code on this page. It uses the publisher's install, not a PlugMCP proxy.

Does DataDoe MCP need an API key?

No required environment variable is listed on this record.

What transport does DataDoe MCP use?

DataDoe MCP is reached over HTTP.

When was DataDoe MCP last checked?

Last checked 2026-10-23. Hosted Amazon Seller and Vendor MCP server for Claude, ChatGPT, Cursor, Codex, Gemini, Copilot.

DataDoe MCP server

Amazon Seller MCP Server - DataDoe

Hosted Amazon Seller Central & Vendor Central MCP server with read and write access. Connect Claude, ChatGPT, Cursor, Codex, Gemini, and GitHub Copilot to live Amazon SP-API and Amazon Ads API data, then let your AI agent act on it: update listings, manage orders, and optimize Amazon Ads campaigns. DataDoe handles the SP-API developer approval, OAuth, and rate limits so your AI agent starts working in under a minute.

🔗 Start a free trial · 📘 Documentation · ⚡ Actions · 📊 Amazon data schema · 🎥 Video demo

🚀 Quick start

Sign up at DataDoe and connect your Amazon Seller Central or Vendor Central account.
Create a DataDoe MCP API key in DataDoe MCP Integrations.

Paste the config below into your AI client (Claude, Cursor, Codex, Gemini, GitHub Copilot, ChatGPT, or any MCP-capable tool):

{
  "mcpServers": {
    "datadoe": {
      "url": "https://mcp.datadoe.com/mcp/v1",
      "headers": {
        "datadoe-mcp-key": "<YOUR_DATADOE_MCP_KEY>"
      }
    }
  }
}

Ask your AI agent: "Show my top 10 ASINs by revenue last month across all Amazon marketplaces."

That's it. DataDoe runs the MCP server on hosted infrastructure, so your team doesn't need to deploy anything locally or wait for Amazon SP-API developer approval.

What is DataDoe MCP?

DataDoe MCP is a hosted Model Context Protocol (MCP) server for Amazon sellers, vendors, and agencies. It exposes your live Amazon Selling Partner API (SP-API) and Amazon Ads API data through MCP tools that work with Claude, ChatGPT, Cursor, Codex CLI, Gemini CLI, GitHub Copilot, Claude Desktop, n8n, NanoClaw, and any other MCP-capable client.

DataDoe MCP gives your AI agent two layers over your Amazon account:

A read layer - SKU-level orders, sales, ads spend, traffic, inventory, listings, returns, settlements, brand analytics, and catalog, returned as structured tool responses or downloadable CSV and JSON exports.
A write layer (Actions) - your agent can change your Amazon account through the SP-API and Amazon Ads API: update listings, cancel orders, confirm shipments, and manage Amazon Ads campaigns, ad groups, targets, and ads.

Building your own Amazon SP-API integration typically requires SP-API developer registration, OAuth refresh-token flow, marketplace-specific endpoints, throttling logic, and 2-4 weeks of Amazon approval. DataDoe takes care of all of that. You get a single authenticated MCP URL for both reading and acting on your Amazon data.

Who is DataDoe MCP for?

Amazon sellers (FBA, FBM, multi-marketplace) - get instant answers from your own seller data without context-switching to Seller Central.
Amazon agencies managing multiple client accounts - query across every connected Seller Central / Vendor Central account from one MCP server.
Vendors with Vendor Central - read 1P data alongside 3P data with the same tools.
AI builders and developers - ship Amazon-aware AI agents, dashboards, and internal tools without writing your own SP-API integration.
Operations teams - automate recurring reports via Claude Code, Cursor, n8n, or any MCP-capable workflow tool.

Why use DataDoe MCP?

✅ No SP-API approval needed - DataDoe handles SP-API developer registration, OAuth, refresh tokens, and rate limits on your behalf.
✅ 30-second setup - paste the MCP URL and your API key into your AI client config. DataDoe runs the server on hosted infrastructure.
✅ 20+ documented integrations out of the box: Claude, ChatGPT, Cursor, Codex, Gemini CLI, GitHub Copilot, n8n, CrewAI, the Claude & OpenAI Agent SDKs, Excel / Word / PowerPoint via Claude, and any other MCP-capable client.
✅ SKU-level resolution - drill into individual ASINs, parent listings, marketplaces, time periods, ad campaigns, keyword reports, settlements, returns.
✅ Multi-marketplace, multi-account - one MCP server covers every Amazon marketplace (US, UK, DE, FR, IT, ES, CA, AU, JP, MX, and more) across Seller Central and Vendor Central.
✅ AI-native by design - exports_create accepts SQL-like filter groups, GROUP BY, aggregations, and date intervals, so your AI agent can build complex reports from one tool call.
✅ Read and write - with Actions, your agent doesn't just report, it updates listings, manages orders, and optimizes Amazon Ads, with a dryRun validation step and per-type controls.
✅ Always-on hosted infrastructure - DataDoe manages SP-API rate limits, token rotation, and ongoing maintenance.

What can you ask DataDoe MCP?

Example questions your AI agent can answer with DataDoe MCP connected:

"What were my top 20 ASINs by revenue last month across all Amazon marketplaces?"
"Reconcile my Amazon settlements against orders for Q1, and show any discrepancies."
"Which Amazon Ads campaigns had ACoS over 40% last week, and what was their total spend?"
"Show inventory units at risk of stocking out in the next 14 days."
"Build me a daily KPI dashboard with sales, traffic, and ad spend for the last 90 days."
"Which search terms in my PPC reports drove the most clicks but zero conversions?"
"Pull every Amazon return for SKU ABC-123 in the last 60 days and summarize the return reasons."
"Compare my brand analytics search term share-of-voice month over month."

And with Actions enabled, your agent can act on what it finds:

"Raise the daily budget on my top-ACoS Sponsored Products campaign by 20%."
"Pause every campaign with ACoS over 50% last week."
"Update the price of SKU ABC-123 to 19.99 and refresh its bullet points."
"Confirm shipment for order 123-4567890-1234567 with UPS tracking 1Z999..."

Actions: write to Amazon

Actions let your AI agent make changes on your connected Amazon Seller Central, Vendor Central, and Amazon Ads accounts through the SP-API and Amazon Ads API. Every Action is recorded and auditable.

What your agent can do:

Listings - update the title, bullet points, description, generic keyword, item-type keyword, and images (AMAZON_LISTINGS_UPDATE), update prices for Seller Central accounts (AMAZON_LISTINGS_PRICING_UPDATE), and preview a listing write with the matching *_VERIFY Action before submitting it.
Orders - cancel an order item, confirm shipment, buy shipping, solicit a review, message the buyer, or search current orders (AMAZON_ORDERS_FIND) and order metrics (AMAZON_ORDERS_GET_METRICS).
Vendor Central - acknowledge purchase orders, submit shipments, and retrieve labels, packing slips, and transaction status.
Amazon Ads - add, update, remove, and find campaigns, ad groups, targets, ads, ad associations, and Sponsored Products bid rules across Sponsored Products, Brands, Display, TV, and Amazon DSP.

How it works - the agent runs each Action through these MCP tools:

actions_details_schema_get - get the payload schema for the Action type.
actions_start with dryRun=true - validate the payload without executing.
actions_start - run the Action and get an action id.
actions_get - poll until it completes, then read the result.

Use actions_list to review past Actions (filter by status, type, creator, and date).

Action types are disabled by default and enabled per type in Settings > Actions. When a type is disabled, actions_start rejects live runs but still allows dryRun validation.

Example details payload for AMAZON_LISTINGS_UPDATE:

{
  "type": "AMAZON_LISTINGS_UPDATE",
  "sellerOrVendorId": "<SELLER_OR_VENDOR_UUID>",
  "dryRun": true,
  "details": {
    "changes": [
      {
        "sku": "ABC-123",
        "language_tag": "en_US",
        "name": "Stainless Steel Water Bottle 750ml",
        "bulletPoints": ["Keeps drinks cold for 24 hours", "Leak-proof lid"]
      }
    ]
  }
}

Running an Action costs 3 AI Tokens for up to 100 changes, plus 1 AI Token per additional 100 changes. See the Actions docs for the full catalog and payload schemas.

Available MCP tools

DataDoe MCP exposes the following tools to your AI client:

Tool	Category	What it does
`sellers_and_vendors_list`	Account	Lists Amazon sellers and vendors connected to your DataDoe organization, with account type, marketplace, and connection details.
`sellers_and_vendors_get`	Account	Returns one seller or vendor, including whether a connection needs reauthorization.
`sellers_and_vendors_rename`	Account	Renames a seller or vendor. Names are unique within the organization.
`sellers_and_vendors_invitation_create`	Account	Creates a one-time link to connect a new Amazon account or reconnect an existing one.
`sellers_and_vendors_invitations_get`	Account	Lists pending invitations, or returns one invitation and its status.
`sellers_and_vendors_invitation_revoke`	Account	Revokes a pending invitation so its link can no longer be used.
`organization_and_subscription_details_get`	Account	Returns your DataDoe organization profile, subscription plan, billing health, and AI Token balances.
`exports_sources_get`	Data	Searches DataDoe's catalog of Amazon data export sources (orders, sales and traffic, ads performance, inventory, listings, settlements, returns, brand analytics, and more) by keyword.
`exports_source_get`	Data	Returns the columns (paginated) and metadata of one export source, including whether a date period is required.
`exports_create`	Data	Creates an Amazon data export from any source. Supports SQL-like filters, HAVING, GROUP BY, aggregations (sum / avg / count / countDistinct / min / max), date intervals (DAY / WEEK / MONTH), pagination, and CSV or JSON output.
`exports_get`	Data	Returns status and metadata for an in-flight or completed export job.
`exports_list`	Data	Lists export jobs for the organization with pagination and optional export ID filters.
`exports_raw_url_get`

Amazon Marketing Cloud (AMC) tools are available by request only and are listed by the live server only after access is granted. See Amazon Marketing Cloud.

Quick setup per AI client

The snippets below are the minimum config you need. For step-by-step guides per AI client, see the Per-client setup guides list at the end of this section.

Claude.ai · Claude Desktop

Add DataDoe as a custom connector: open Customize > Connectors, click Add, and paste https://mcp.datadoe.com/mcp/v1 as the MCP Server URL. Claude signs in to DataDoe with OAuth, so you do not need an API key. See Using Claude.

Claude Code

claude mcp add datadoe "https://mcp.datadoe.com/mcp/v1" --transport http --header "datadoe-mcp-key: <YOUR_DATADOE_MCP_KEY>"

The server name and URL must come before --header. You can also omit --header and sign in with OAuth by running /mcp in Claude Code and choosing Authenticate.

Cursor

Add to ~/.cursor/mcp.json (global) or .cursor/mcp.json (per project):

{
  "mcpServers": {
    "datadoe": {
      "url": "https://mcp.datadoe.com/mcp/v1",
      "headers": {
        "datadoe-mcp-key": "<YOUR_DATADOE_MCP_KEY>"
      }
    }
  }
}

GitHub Copilot (VS Code)

Add to .vscode/mcp.json (per project) or your user MCP configuration (MCP: Open User Configuration):

{
  "servers": {
    "datadoe": {
      "type": "http",
      "url": "https://mcp.datadoe.com/mcp/v1",
      "headers": {
        "datadoe-mcp-key": "<YOUR_DATADOE_MCP_KEY>"
      }
    }
  }
}

Codex CLI · Gemini CLI · ChatGPT · n8n · NanoClaw · any other MCP client

DataDoe MCP works as a generic remote MCP server. Configure your client with:

URL: https://mcp.datadoe.com/mcp/v1
Transport: HTTP Streamable
Auth header: datadoe-mcp-key: <YOUR_DATADOE_MCP_KEY> (create a key in DataDoe MCP Integrations)

Per-client setup guides

For step-by-step setup guides per AI client, see the dedicated DataDoe documentation pages:

Full documentation root: www.datadoe.com/hub/docs

What Amazon data is available?

DataDoe MCP exposes every Amazon data table available in DataDoe, including:

Orders & sales: order line items, order performance, sales and traffic, refunds
Amazon Ads: campaigns, ad groups, keywords, search terms, sponsored products / brands / display, ACoS / ROAS / impression-share metrics
Inventory: FBA inventory, restock recommendations, stranded inventory, age, units at risk
Listings & catalog: ASIN catalog, product attributes, buy box ownership, variations
Finance: settlements, fees, reserves, reimbursements, deposits
Returns: customer returns, return reasons, FBA returns
Brand analytics: search query performance, market basket, repeat purchase, demographics
Traffic: sessions, page views, conversion rate, by ASIN, by marketplace

Full schema: www.datadoe.com/hub/data-scheme

What DataDoe MCP is NOT

To avoid confusion when evaluating Amazon MCP servers:

❌ Not a self-hosted MCP server. DataDoe MCP is hosted at mcp.datadoe.com. This GitHub repository is the schema facade used to publish DataDoe MCP to public MCP registries (Official MCP Registry, mcpservers.org, glama.ai, mcpmarket.com, Cline, smithery.ai, and others).
❌ Not an SP-API wrapper you operate yourself. DataDoe owns the SP-API developer registration, OAuth flow, refresh-token rotation, marketplace endpoints, and rate-limit handling. You bring your Amazon account; we handle the rest.
❌ Not a free open-source server. DataDoe MCP requires a DataDoe subscription. For a free self-hosted alternative, see the community Amazon MCP servers indexed at mcpservers.org.

Documentation & resources

License

MIT. The DataDoe MCP schema facade in this repository is open-source. The hosted MCP server, DataDoe application, and DataDoe infrastructure are proprietary and operated by Deltologic.

How to start and connect locally (for indexing and listing)

This server is just a schema facade of the actual DataDoe MCP server, made for exposing DataDoe MCP to various MCP registries. It is a no-op server: it does not do anything beyond exposing the schema of DataDoe MCP. If you want to actually use DataDoe MCP, see the DataDoe MCP documentation.

1. Install dependencies

yarn install

2. Build the local server

yarn build

This compiles the TypeScript source into dist/index.js.

3. Start the server

yarn start

The server communicates over stdio, so the process is intended to be launched by your MCP client rather than opened in a browser.

4. Connect from an MCP client

For a local setup, point your client at the built entrypoint. A typical mcp.json looks like this:

{
  "mcpServers": {
    "datadoe": {
      "command": "node",
      "args": ["/absolute/path/to/datadoe-mcp/dist/index.js"]
    }
  }
}

If your client supports environment-based workspace paths, you can usually swap the absolute path for the actual clone location on your machine.

DataDoe is operated by Deltologic · datadoe.com

DataDoe MCP

Add this server

Host a copy with our keys and logs.

Server facts

Tools

Questions