Skip to main content
Free Trial Book a Demo
Product · MCP Server

Your retail reviews, inside Claude and ChatGPT.

The Indellia MCP Server exposes your feedback corpus over the Model Context Protocol, so your team asks questions in the assistant they already use and gets answers from every review you have — each one resolved to a SKU. Read-only by default.

Start Free Trial Book a Demo
Shipped · in production

The short answer

The Indellia MCP Server is a Model Context Protocol connector that lets Claude Desktop, ChatGPT or Cursor query your retail review corpus directly. Access is scoped per workspace and read-only by default; write actions need an explicit grant. An MCP server is not by itself a reason to choose a feedback platform — several vendors ship one as of Q4 2026. What differs here is the corpus underneath: retail marketplace reviews, resolved to individual SKUs.

Why this page is not a sales pitch for MCP

Everyone has one now. That is the point.

Eighteen months ago an MCP server was a differentiator. It is not any more, and a page pretending otherwise would be the sort of claim you could disprove in one browser tab. Of the twenty-nine vendors we surveyed in this category, nineteen publish an llms.txt, and the closest two to us publish MCP servers with more documentation than this page has:

  • Chattermill names Claude, Claude Code, ChatGPT, Codex, Glean, Cursor, Notion Agent, Gemini and Microsoft Copilot among its MCP clients, and governs access through existing user permissions.
  • Enterpret lists its MCP server in the Claude Connector Directory and names Claude, ChatGPT, Cursor, Slack, Glean, Notion and LibreChat.
  • UserVoice advertises “ask Claude or ChatGPT what customers want, answered from your live UserVoice data.”

Those are their own published claims, and they are fair ones. So the honest question is not whether a platform speaks MCP. It is what answers when you ask.

What answers

A corpus nobody else in this category publishes.

Ask an MCP server about customer feedback and you get back whatever that platform holds. Across every vendor we checked, none publishes direct ingestion from retail marketplace listings — Amazon, Walmart, Best Buy, Costco, Lowe’s, Target, Home Depot — and none publishes a product-, SKU-, UPC- or model-number unit of analysis. Their units are the ticket, the account, the theme, the interaction, the location.

So the same question returns a different kind of answer. “Why are returns spiking?” against a ticket corpus gives you themes in support conversations. The same question here resolves to a variant: which model number, at which retailer, since when, with the reviews cited. That is the SKU-level method doing the work, not the protocol.

Scopes

Read-only until you say otherwise.

Access is scoped per workspace and the default scope is read-only. Write actions — posting a review response, for instance — require an explicit scope grant.

This is not caution for its own sake. An assistant calling tools on your behalf is acting with your authority, and the failure mode for a brand is publishing something in your own name that nobody approved. Reading is safe and immediately useful; writing is a decision, so it is a decision you have to make rather than one you inherit.

What to ask it

Questions worth a connector.

The queries that justify this are the ones that are tedious in a dashboard and impossible in a chat window without the data: which SKU’s sentiment moved most this month and why; whether a complaint on Walmart also appears on Best Buy for the same model; which variant of one product line carries the defect; what changed in the week after a packaging revision. Each needs the whole corpus and the SKU resolution underneath it.

If your question is “summarise these two hundred reviews I already have,” you do not need this or us — paste them into the assistant. We wrote that down properly on Indellia vs ChatGPT.

Where to go next

Related pages.

The agent roster

Theme, Anomaly, SKU, Defect, Response and indelliaGPT™ — what each one owns.

See the platform →

FAQ

Frequently asked questions

What does the Indellia MCP Server actually do?

It exposes Indellia’s feedback intelligence as a Model Context Protocol server, so an MCP-capable client — Claude Desktop, ChatGPT or Cursor — can query your own review corpus directly. You ask in the assistant you already use; the answer comes from your complete, SKU-linked retail feedback rather than from whatever you managed to paste into the window.

Is it read-only?

By default, yes. Access is scoped per workspace and the default scope is read-only. Write actions — posting a review response, for example — require an explicit scope grant. That is deliberate: a tool an assistant can call on your behalf should not be able to publish something in your brand’s name because a prompt went sideways.

Which clients work with it?

Any MCP-capable client. Claude Desktop, ChatGPT and Cursor are the ones we test against and name. The protocol is open, so a client we have never heard of will generally work; we only put our name behind the three we actually check.

Is an MCP server a reason to choose Indellia?

On its own, no, and we would rather tell you than let you find out. As of Q4 2026 Chattermill publishes an MCP server naming Claude, Claude Code, ChatGPT, Codex, Glean, Cursor and others; Enterpret publishes one in the Claude Connector Directory; UserVoice advertises assistant access to its feedback data. This is table stakes. The question worth asking is what sits behind the server — whose reviews, from which channels, resolved to what.

What makes the corpus behind it different?

It is retail. Amazon, Walmart, Best Buy, Costco, Lowe’s, Target, Home Depot and Bazaarvoice-powered retailer pages, with every review resolved by the SKU Agent to one physical product across retailer-specific identifiers — ASIN, Item ID, Model#, UPC. No other MCP server in this category publishes retail marketplace ingestion or a SKU-level unit of analysis. That is the difference, not the protocol.

Does it need engineering work to set up?

No code. It is a connector configured per workspace, and the read-only default means the first thing you can do is ask questions. If you want write scopes, that is a deliberate grant rather than a default.

Ask Indellia

Have a specific question?

Indellia's AI agents answer with citations from real customer feedback across Amazon, Walmart, Best Buy, and 20+ retail channels.

Try Indellia

Connect it, then ask it something.

Start the free trial, connect a retail channel, and point Claude or ChatGPT at your own reviews. Read-only by default — nothing to undo.

Start Free Trial Book a Demo