Nova MCP for AI assistants
Give your AI assistant every Amazon number
Nova MCP is the decision layer for Amazon sellers: the bridge between AI and the most advanced Amazon data model on the market.
Connect your Amazon data to your favourite AI assistant
Josh Grillo, Founder of Sell Through, on the Nova MCP
Measured coverage
Why an assistant pointed at Nova answers better
Amazon does not hand you a profit figure. It hands you settlements, order reports and fee lines spread across separate endpoints, in different shapes, that only mean something once they are matched up.
Nova does that matching before the question is asked. 73 charge lines, 45 refund lines and 21 reimbursement types, 139 money movements in all, reconciled down to one SKU on one day. An assistant asking Nova asks once and gets the reconciled answer. An assistant querying Amazon directly spends its context paginating and joining, and can still get it wrong.
- 73
- Charge lines
- 45
- Refund lines
- 21
- Reimbursement types
- 236
- P&L measures
- 84
- Listing health fields
- 75
- Search Query Performance fields
- 34
- Advertising targeting fields
- 34
- Inventory fields
- 27
- Live data models
Fees, commissions, chargebacks and surcharges, each matched to the SKU and day it hit
The refund side of those charges, so a returned unit lands on the right day
Amazon paying you back for lost, damaged or miscounted stock
Everything you can ask of the profit model, from net margin to fee share of sales
Suppressions, buy box state, content and image checks on your own ASINs
Impression, click, add to cart and purchase share for the queries you rank on
Campaign, ad group, keyword and target level spend and return
Stock on hand, inbound, reserved, unfulfillable and days of cover
Modelled and joined before your assistant sees them, across 23 Amazon marketplaces
The P&L contains the current trading day, not yesterday's close. Counted from the Nova warehouse schema, September 2026.
Your AI assistants, plugged into every Amazon data source
Data sources
AI assistants
ClaudeDesktop, Code, webChatGPTConnectors & custom GPTsGeminiMCP-capable clients500+ metrics. Hourly refresh. 23 Amazon marketplaces.
How to choose the right MCP - 8 criteria to decide
| Dimension | Nova MCP | Typical Amazon MCP connector | Edge |
|---|---|---|---|
| 01Data accuracy | 99.9% guaranteed. 370 Amazon transaction lines reconciled into 154 P&L lines | Clean API copies. Fee reconciliation depth rarely documented | Nova |
| 02Data freshness | Several refreshes a day, up to every 30 minutes on money metrics like Buy Box lost | Usually once a day, so answers run on yesterday's numbers | Nova |
| 03Data coverage | SP-API, Ads, Search Query Performance, search terms, inventory, listing health, BSR, plus Vendor Central (1P) | Broad API coverage, served as tables rather than decisions | Nova |
| 04Token consumption | Structured tables and pre-aggregated answers. No raw dumps | Raw tables, so the model burns tokens filtering and aggregating | Nova |
| 05SKU cost allocation | Profit to the cent per SKU, attributed unit by unit | Sales and fees per SKU, no true landed cost | Nova |
| 06Smart data enrichment | COGS per SKU, FBM cost, 24 months of backfilled history, your own tags | Only what Amazon publishes. Your costs live elsewhere | Nova |
| 07Speed of answers | One question, one low-token call against a structured table | Several round trips per question | Nova |
| 08Marketplace coverage | 23 marketplaces, currency-normalized across accounts and regions | Marketplace count and currency normalization rarely documented | Nova |
Nova MCP
99.9% guaranteed. 370 Amazon transaction lines reconciled into 154 P&L lines
Typical Amazon MCP connector: Clean API copies. Fee reconciliation depth rarely documented
Nova MCP
Several refreshes a day, up to every 30 minutes on money metrics like Buy Box lost
Typical Amazon MCP connector: Usually once a day, so answers run on yesterday's numbers
Nova MCP
SP-API, Ads, Search Query Performance, search terms, inventory, listing health, BSR, plus Vendor Central (1P)
Typical Amazon MCP connector: Broad API coverage, served as tables rather than decisions
Nova MCP
Structured tables and pre-aggregated answers. No raw dumps
Typical Amazon MCP connector: Raw tables, so the model burns tokens filtering and aggregating
Nova MCP
Profit to the cent per SKU, attributed unit by unit
Typical Amazon MCP connector: Sales and fees per SKU, no true landed cost
Nova MCP
COGS per SKU, FBM cost, 24 months of backfilled history, your own tags
Typical Amazon MCP connector: Only what Amazon publishes. Your costs live elsewhere
Nova MCP
One question, one low-token call against a structured table
Typical Amazon MCP connector: Several round trips per question
Nova MCP
23 marketplaces, currency-normalized across accounts and regions
Typical Amazon MCP connector: Marketplace count and currency normalization rarely documented
Everything you need to 3x your Amazon business
Search 500+ Amazon metrics
Can't find what you need?
Setup, in one minute
How to connect Nova to Claude.
Four steps, no engineering. The same server URL works in ChatGPT, Gemini and Cursor: add it in that assistant's connectors settings.
- 01
Open Claude, go to Settings then Connectors
In Claude, open your profile, then Settings, then the Connectors tab. Works on the desktop app and on claude.ai.
- 02
Click Add custom connector
In the Connectors panel, click Add, then Add custom connector. If you do not see the option, check that you are on a Claude paid plan.
- 03
Fill in the connector details
Name
NovaRemote MCP server URL
https://mcp.novadata.io/api/mcpNo authentication is required in the current beta. Click Add, then Save.
- 04
Toggle Nova on and start asking questions
Toggle Nova on, open a new conversation, and ask about your live Amazon business.
Screenshots, troubleshooting and example prompts.
Enrich your data from Claude
A P&L is only as good as the costs behind it. Those costs no longer have to be typed into a settings screen. Ask Claude, ChatGPT or Gemini to set them, one SKU at a time or a whole file at once, and Nova writes them back.
1. Ask, or paste your file
One SKU or a whole cost file. Nova matches your rows to real SKUs and ASINs before anything else happens.
2. Read the preview
You get the rows to import with the region, the currency and the mode, plus a count of what is added, updated, unchanged and removed, with a warning on any row that looks wrong.
3. Confirm, then it writes
Nothing is saved until you approve that exact preview. Missing-cost counters update straight away, and the P&L picks the new values up on the next refresh.
COGS
Landed cost per SKU, as a fixed cost for a period or as a full FIFO batch ladder.
FBM rate book
Shipping per delivery country, a rest-of-world fallback, packing per parcel and picking per unit.
VAT categories
The product tax category per SKU, for the regions where it actually changes the P&L.
Stock and supplier orders
Stock parameters, open supplier orders, stock held outside Amazon, and your own product tags.
The same connector that reads your data can now fill it in. See how it feeds the P&L.
More on the Nova MCP
Setup walkthroughs, the assistant-by-assistant data catalog, and the background on how MCP works for Amazon sellers.
- What is an Amazon MCP server?Plain-English explainer: MCP, SP-API wrappers, Amazon Ads MCP, and what each one can actually see.
- Amazon AI agentsThe agent layer: what you can build once an assistant can read your seller data.
- Connect Claude in 4 stepsThe setup walkthrough, from Seller Central auth to your first prompt.
- Amazon data for ClaudeThe data catalog Claude reads: sources, grains, and the SKU profit model behind them.
- Amazon data for ChatGPTThe same model, read from ChatGPT: custom GPTs, the API, or your own agent.
- Amazon data for GeminiGemini, AI Studio and Vertex reading the same reconciled Amazon dataset.
- Amazon Ads MCP vs Nova MCPWhat Amazon's official Ads MCP can act on, and what it cannot see.
- Seller Central MCP serverWhat an MCP layer over the Selling Partner API can and cannot answer today.
- Amazon MCP prompt library32 copy-and-paste prompts for profit, PPC, keyword share, inventory and listings.
- Nova MCP tool referenceAll 14 tools, the question each answers, and what comes back.
- Amazon MCP for agenciesMulti-account operators: one connection across every client, and per-client profit after fees.
Frequently asked questions
How many Amazon data points does the Nova MCP serve?
The Nova MCP serves 500+ data points across sales, profitability, PPC campaigns, ad groups and keywords, customer search terms, search query performance (SQP), Best Seller Rank, fees, refunds, FBA stock health, listing health, and your own cost inputs. All refreshed continuously from Seller Central, SP-API, and the Ads API.
Which AI assistants can connect to Nova?
Any assistant that supports the Model Context Protocol. Claude, ChatGPT and Gemini are the ones sellers use most. You add Nova once as a custom connector and the assistant can query your Amazon business from then on.
Why does token consumption matter with an Amazon MCP?
Connectors that hand the model raw tables force it to filter and aggregate thousands of rows itself, which burns tokens and slows the answer down. Nova returns structured, pre-aggregated tables, so one question is usually one low-token call.
Which categories are covered?
Sales and orders, profitability, PPC and advertising, customer search terms, search query performance, Best Seller Rank, buy box, inventory, fees and reimbursements, refunds, subscribe and save, listing health, and your own cost inputs (COGS, freight, 3PL, and more).
Is the data library free to browse?
Yes, browse and search the full library freely. Connecting your assistant to live data requires a Nova account (14-day free trial, no card).
How often is the data refreshed?
Several times a day for every data point, and up to every 30 minutes for the metrics that move money, pulled from the Amazon Selling Partner API, Advertising API, Brand Analytics, and Seller Central business reports.
Ready to give your assistant every Amazon number?
The Nova MCP serves these 500+ data points to Claude, ChatGPT and Gemini, straight from Seller Central, SP-API and the Ads API. Refreshed several times a day, ready to query.