The Ultimate MCP Server

for Amazon Sellers

Doe gives you the complete Amazon data layer — every metric, clean and AI-ready — so your AI runs on your terms.

From Amazon login to your first AI query in under 5 minutes — connect Seller Central, Vendor Central, and Amazon Ads, then talk to your data in Claude, ChatGPT, Cursor, or any MCP-compatible assistant.

BOOK A DEMO
Book a demo
What's inside

Six Amazon APIs. One MCP endpoint.

Connect Amazon once, get every Seller, Vendor and Ads API exposed as named MCP tools your AI can call out of the box. Forty-plus endpoints. Zero glue code.

Amazon Seller MCP

Orders, listings, FBA inventory, fees, settlements, reimbursements and Brand Analytics — through the official Amazon SP-API across every marketplace.

get_ordersget_fba_inventorybrand_analytics

Amazon Vendor MCP

1P Vendor Central programmable for the first time — POs, ASNs, shortages, chargebacks, OTIF and Retail Analytics (ARA Basic + Premium).

get_purchase_ordersget_chargebacksget_ara

Amazon Ads MCP

Sponsored Products, Brands, Display, Brand Stores, Amazon DSP and Amazon Marketing Cloud — read performance, change bids, run queries by chat.

list_campaignsupdate_bidrun_amc_query

Reconciled True Profit

Native primitives that net fees, FBA charges, returns, Ads spend and COGS together — so profit_by_sku returns real net profit.

profit_by_skuprofit_by_brandtacos_by_campaign

Multi-account & multi-marketplace

One scoped MCP token covers every Seller, Vendor and Ads account you authorize across NA, EU, UK, JP, AU — all reconciled into one schema.

5 marketplacesn accounts1P + 3P unified

Self-describing schema

Your AI calls tools/list on connect and learns every Amazon endpoint — names, parameters, types. No schema mapping, no SDK install.

tools/listJSON SchemaMCP spec
On the AI side

What every AI client gets out of the box

The same DataDoe Amazon MCP endpoint works in Claude, ChatGPT, Cursor, Codex, Copilot and Gemini. Each one unlocks the same four capabilities.

Conversational Amazon BI

Ask anything in plain English. "Which ASINs lost money last month after fees, FBA and Ads?" — your AI calls the right MCP tool, federates the API calls, returns a typed answer.

  • Multi-turn conversations keep account scope
  • Answers cite the tool calls they made
  • Works with 40+ tools without prompt engineering

Build Amazon agents

Point Claude Code, Cursor, Codex or Copilot at the MCP endpoint and let it ship Amazon tooling — repricer, FBA reimbursement bot, chargeback dispute agent — on real Seller, Vendor and Ads data.

  • Schema auto-discovery, no SP-API wrappers to write
  • Live data from line one, no fixtures or mocks
  • Replace the SaaS stack you were renting

Scheduled prompts & reports

Save a prompt, run it every Monday at 8am. Weekly ACoS digest, monthly P&L recap, daily Buy Box risk — delivered to email or Slack.

  • Daily, weekly, monthly cadences
  • Markdown reports or raw data export
  • Different prompts for different stakeholders

Webhook-driven actions

Subscribe to events — Buy Box lost, ACoS spike, low stock, negative review — and trigger AI agents the moment they fire. No polling, no missed signals.

  • Event-driven, not poll-based
  • Pipe into Slack, email, n8n, Zapier or your own webhook
  • Replay missed events from any window
In the wild

What teams actually ship on it

Three real workflows running on the DataDoe Amazon MCP today — same endpoint, same schema, different shape of answer.

PPC Manager

"ACoS just spiked. Find why."

Tuesday morning ACoS jumps 28%. Instead of opening six dashboards, asks the AI.

claude · amazon mcp
why did acos jump tuesday
3 SP campaigns spent 4× normal on one search term:
"organic dog treats"
spend $1,840 (avg $420)
clicks 312, orders 8
ACoS 84% (target 22%)
Finance

"Close the month in 20 minutes."

Reconciled P&L by brand. Fees, FBA, Ads, returns — all netted. Done before standup.

P&L · Oct close
Brand A Brand B Brand C
Net rev
$284K+12%
True profit
$96K+8%
FBA fees
$42K+18%
Ad spend
$58K+4%
Agency · Multi-account

"Risk across 12 clients, now."

Vendor PO shortages and FBA stock-outs across every client account, color-coded.

multi-account · risk
SKU / PODOCSTATUS
B0QW4K3M268dOK
B0NM7YPXL21dLOW
B09R7ZQBC9dRESTOCK
PO 88JX2L−180uSHORT
PO 7QM9YA−940uCHARGE
B07LM2NKJ142dAGED
Day One

From OAuth click to first answer in five minutes

No SDK install, no schema mapping, no onboarding call. The MCP protocol does the handshake for you.

01
Authorize Amazon

Official LWA OAuth click. Seller, Vendor and Ads — all marketplaces, all accounts.

02
Get your MCP URL

DataDoe issues your private endpoint. Every Amazon API already wired in.

03
Paste into your AI

Add as MCP server in Claude, ChatGPT, Cursor or any MCP-aware client.

04
Ask anything

Tools auto-discovered. Your AI knows every Amazon endpoint. You're live.

Inside the MCP

Six things the protocol handles so you don't

Build directly on Amazon's APIs and you own a thousand edge cases. The MCP absorbs every one of them.

LWA OAuth · automatic token rotation

SP-API, Vendor and Ads refresh tokens managed in the background. Your AI session never breaks mid-conversation.

Multi-marketplace · home-currency normalized

Each marketplace has its own SP-API endpoint, currency and locale. The MCP collapses NA, EU, UK, JP, AU into one schema.

Reconciled profit · fees, refunds, Ads, COGS

Revenue is what Amazon shows you. The MCP reconciles every fee, return and ad cost into profit_by_sku primitives.

SP-API rate limits · smart queuing

Burst + restore quotas per role, fanout across accounts. Your AI gets a clean answer instead of a 429.

1P + 3P joined on ASIN

Brands selling the same ASIN as 3P in some marketplaces and 1P in others. One query, total Amazon revenue across both.

Amazon-audited application

Listed in the Amazon Selling Partner Appstore. Annual penetration tests, 180-day vuln scans, encryption at rest and in transit.

Works with every MCP-aware AI

Pick your client. The Amazon layer stays the same.

One DataDoe Amazon MCP endpoint. Six native AI integrations. Zero per-client integration work.

Anthropic · native MCP

Claude

Native Model Context Protocol in Claude Desktop, Claude Code and Claude on the web.

Amazon MCP for Claude OpenAI · agent mode

ChatGPT

Add the Amazon MCP as a custom connector in ChatGPT Plus, Pro, Team or Enterprise.

Amazon MCP for ChatGPT Google · MCP + CLI

Gemini

Plug the Amazon MCP into Gemini and Gemini CLI through MCP support.

Amazon MCP for Gemini AI-native IDE

Cursor

Drop the Amazon MCP into Cursor's .mcp.json and ship Amazon agents fast.

Amazon MCP for Cursor OpenAI · IDE · CLI

Codex

Connect Codex to the Amazon MCP and let it build agents on real Seller, Vendor and Ads data.

Amazon MCP for Codex GitHub · agent mode

GitHub Copilot

Copilot agent mode reads every Amazon endpoint, straight from VS Code.

Amazon MCP for Copilot

FAQ

What is Amazon MCP?
Faq Plus
What is the difference between Amazon Seller MCP, Amazon Vendor MCP, and Amazon Ads MCP?
Faq Plus
How do I connect Amazon to ChatGPT or Claude using MCP?
Faq Plus
Is DataDoe an alternative to Sellerlabs MCP?
Faq Plus
Is Amazon MCP secure? What data does DataDoe see?
Faq Plus
How long does setup take?
Faq Plus
Can I use Amazon MCP for multiple Amazon accounts and marketplaces?
Faq Plus
What can I actually do with Amazon MCP in Claude or ChatGPT?
Faq Plus
Do I still need Helium 10, Jungle Scout, or other Amazon SaaS tools?
Faq Plus
How is DataDoe Amazon MCP priced, and is there a free trial?
Faq Plus

Set up in under 5 minutes.
Try free for 7 days. Then $97/month.

Every integration. Full onboarding support. If it’s not the best decision you made in 2026, you can cancel anytime.

GET STARTED FOR FREEBook a demo

Know what makes you money

Catch problems instantly

Connect anything with API & MCP

Replace tools with your own apps

Access Amazon-audited infrastructure

Set up in under
5 minutes.
Try free for 7 days. Then $97/month.

Every integration. Full onboarding support. If it’s not the best decision you made in 2026, you can cancel anytime.

GET STARTED FOR FREEBook a demo