Skip to main content
Nova Analytics

Tool reference

Every Nova MCP tool, and the question it answers

You never call these by name. You ask, the assistant picks the tool, and the ranking happens on our side so the answer comes back small.

Fourteen tools. The whole seller P&L, one question away.

14
tools
23
marketplaces
40+
fee types reconciled
Try for Free
Rated 4.7/5 onTrustpilotTrusted by 3,000+ Amazon sellers
Rated 4.7/5 onTrustpilotfrom 23 reviews

Josh Grillo, Founder of Sell Through, on the questions he runs through the Nova MCP

14

tools

19,000

tokens saved on discovery

24 mo

history queryable

30 min

refresh on money metrics

The tools are useful because they chain

One call answers a question. Three in a row answer a problem, and the assistant picks the order.

Which SKUs lost the most net profit last month?

find_moversstep 1
SKUChange
NV-4021-$4,120
NV-1180-$2,330
NV-7734-$980

What moved on the worst one?

explain_movementstep 2
SignalVariance
Ad spend+38%
Conversion rate-12%
Buy Box share-9 pts

Which search terms wasted spend on it?

query_search_termsstep 3
TermSpend
cordless drill set$412
drill bits titanium$188
18v battery pack$121

Example answers. Figures shown are illustrative, the shape of the response is not.

Profit and movement

The P&L side: what you made, what changed, and why it changed.

query_metrics

"What was net profit by brand in the UK last month versus the month before?"

find_movers

"Which 10 SKUs lost the most net profit last week?"

explain_movement

"Why did this ASIN drop 30% in profit last week?"

list_cost_inputs

"Which SKUs are missing COGS, so their margin is overstated?"

Advertising and demand

Spend, the queries behind it, and your share of the funnel.

query_advertising

"Which keywords spent over $100 last month with zero attributed sales?"

query_search_terms

"Which search terms burned spend last month without a single sale?"

query_sqp

"Where am I losing share between click and purchase on my top queries?"

query_bsr

"How has BSR moved on this ASIN since the price change?"

Inventory and listings

The operational side that quietly caps revenue.

query_inventory

"Which ASINs run out of stock inside 21 days at the current sales rate?"

query_listing_health

"Which of my listings are suppressed or lost the buy box right now?"

Discovery

How the assistant learns your account without loading the whole schema.

list_metrics

"What can I actually measure here?"

describe_metric

"How exactly is net profit calculated?"

list_accounts

"Which accounts and marketplaces can I query?"

list_tags

"Which tags can I break this report down by?"

Inputs you can fill from the assistant

The costs and stock inputs behind the P&L, written back through the same connection. Every one of these previews the change first and writes only once you confirm.

import_cogs

"Here is my supplier price list, set the COGS for these 40 SKUs in the US."

import_fbm_zone_rates

"Set my UK shipping rate to 4.20 and packing to 0.35 per parcel."

import_vat_categories

"Apply the standard rate category to every UK SKU that has none."

import_stock_params

"Set a 45 day lead time on my EU stock for these ASINs."

import_supplier_orders

"Log a 2,000 unit PO landing in the US on the 14th."

import_external_stock

"Here is my 3PL stock report, load it against my ASINs."

import_product_tags

"Tag these 60 ASINs as the Winter line."

These 14 tools run against your own account the moment Seller Central is connected.

Try for Free

Why the tool list looks like this

19,000 tokens is what a full catalogue costs, every conversation

So discovery is shallow, detail is on demand, and every tool filters and ranks before it answers.

A modelled tool set against a raw API copy

01Token consumptionNova

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

02Speed of answersNova

Nova MCP

One question, one low-token call against a structured table

Typical Amazon MCP connector: Several round trips per question

03Data coverageNova

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

04Smart data enrichmentNova

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

05Data accuracyNova

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

Setup

Point your assistant at these tools, in four steps

Authorise Seller Central once, paste the endpoint, and the 14 tools are available in the chat.

  1. 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.

  2. 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.

    Claude connectors panel with the Add custom connector option highlighted
  3. 03

    Fill in the connector details

    Name

    Nova

    Remote MCP server URL

    https://mcp.novadata.io/api/mcp

    No authentication is required in the current beta. Click Add, then Save.

    Claude Add custom connector modal with Nova name and MCP server URL filled in
  4. 04

    Toggle Nova on and start asking questions

    Toggle Nova on, open a new conversation, and ask about your live Amazon business.

    Claude connectors toggle with Nova enabled next to Gmail, Mixpanel and Pappers
Full Connect to Claude guide

Screenshots, troubleshooting and example prompts.

Josh Grillo, Founder, Sell Through

"One of the blessings we've had with Nova is the MCP they offer."

Josh Grillo, Founder, Sell Through

Why this MCP and not another one

Coverage

SP-API, Ads, Search Query Performance, search terms, inventory, listing health and BSR, across 23 marketplaces. Vendor Central (1P) is live.

Freshness

Several refreshes a day, up to every 30 minutes on money metrics like Buy Box lost. Most connectors answer with yesterday's numbers.

Speed and tokens

Structured, pre-aggregated tables instead of raw dumps, so one question is usually one low-token call rather than several round trips.

Your costs, not just Amazon's

COGS per SKU, FBM cost, your own tags and 24 months of backfilled history, reconciled at 99.9% accuracy across 370 transaction lines.

The full row-by-row comparison lives on the Nova MCP page.

Every tool has a prompt already written for it

You never name a tool in the chat, you ask a question and the assistant picks. The prompt library has 32 of those questions, grouped by profit, PPC, keywords, inventory and listings, each one tagged with the tools it hits.

Three things to know before you point an assistant at your account.

Answers in seconds

Pre-aggregated responses, ranked server side, light on tokens.

Revocable

Per-account credentials you can pull back at any time.

Minutes, not a project

Authorise, paste one endpoint, 24 months of history backfills.

Longer read: choosing the best Amazon MCP server and why an MCP server is not a product.

Tool questions

How the tools are used, and why they are shaped this way.

No. You ask a question in plain language and the assistant picks the tool. The names here matter for understanding what is covered, not for writing calls by hand.
Tool definitions load on every conversation, so a large schema is a fixed cost per conversation. Returning the full metric catalogue in one call would be roughly 19,000 tokens, which is why discovery is shallow by default and detail is fetched on demand.
Because ranking a large catalogue inside the model means moving every row into context and discarding almost all of it. find_movers returns the ranked answer instead, which keeps the response fast and the token bill flat.
No. Your campaigns, budgets, prices and listings stay exactly as you set them, and the per-account credentials can be revoked at any time. The import tools write to your own Nova inputs, such as cost of goods and shipping rates, never to Seller Central.
Only the inputs you maintain yourself, and only after you approve it. An import runs in two steps: the assistant sends the rows, Nova validates them and returns a preview with the rows to import and a count of what is added, updated, unchanged and removed, plus a warning on anything that looks wrong. Nothing is saved until you confirm that exact preview.

Give your AI assistant every Amazon number

Connect Seller Central, add the endpoint to your assistant, ask the first question in minutes.

Try for Free

14-day free trial, no card required. Read-only MCP access.

The rest of the Amazon MCP guide

Every page in this cluster answers a different question. Start at the pillar if you want the full picture.