Use case

Get code diffs, not a dashboard screenshot

If you are the person who has to implement the change, a chart of what is wrong is only half a ticket. The other half is the diff.

  • MCP access
  • Reviewable diffs
  • Theme-aware changes
The MCP adoption gap

70%

Aware of MCP

It is no longer a niche protocol

10%

Using it regularly

Interest is well ahead of tooling

Seven in ten developers know what MCP is; one in ten actually works through it day to day. The gap is not scepticism — it is that very little of what a developer needs in context is exposed over MCP yet. Store behaviour data is a good candidate: it is exactly the context you do not have in your editor.

Source: Postman — State of the API Report 2025 — 5,700+ developers and architects

The handoff is the expensive part

A finding in a dashboard and a change in a theme are separated by a translation step, and that step is where storefront improvements die. Someone reads a chart, writes a description of what they think it implies, and hands it to someone who has to reconstruct the original observation from the description. Both people are competent; the format is what loses the information.

  • The observation was element-level; the ticket is page-level
  • The person who saw the pattern is not the person who changes the Liquid
  • By the time it ships, the theme has moved on
  • Nobody can attribute the result, so the next one is harder to justify

What changes when the data is in your editor

The same 5,700-developer survey found 69% spend ten or more hours a week on API-related work, and a quarter spend over twenty. Developer time is not the constraint people assume — attention-switching is. Being able to ask the store a question without leaving the editor, and get an answer shaped like a change, removes the round trip rather than the work.

Working against a store over MCP

The point is not chat. It is that the model has real store context, so what comes back is specific to your templates rather than to Shopify in general.

  1. 1

    Connect the store to your MCP client

    Claude, Cursor, or anything else that speaks MCP. Remote server, so there is nothing to run locally and nothing to keep updated.

  2. 2

    Ask in the terms you already think in

    Which template loses the most sessions before add-to-cart; what is being dead-clicked on mobile; what changed after last Thursday's deploy.

  3. 3

    Get the change as a diff, not a description

    Scoped to the theme file it belongs in, reviewable in the tool you already review code in.

  4. 4

    Ship it and let it be measured

    The change is attributed against a holdout, so the next one is an argument with a number behind it rather than a preference.

MCP access vs building the integration yourself
Over MCPRoll your own
Time to first answerConnect and askDesign the schema, build the pipeline, then ask
Who can use itAnyone with an MCP client, including non-developersWhoever knows the pipeline
MaintenanceNone on your sideYours, including every upstream change
FlexibilityThe questions the server exposesAny question you are willing to build

Where the other one wins: Rolling your own wins outright if you need the raw event stream in your own warehouse — joined against your ERP, your subscriptions, your margin data. An MCP server exposes a curated surface, and curation is a limit as much as a convenience. If you have a data team, take the raw feed.

The server, what it exposes, and how to point a client at it are documented here. Read on →

Questions

Shopify for Developers, answered

Does it edit my theme without asking?
No. Changes are proposed and previewed; applying one is an explicit approval. The agent does the drafting, you keep the merge button.
Which MCP clients are supported?
Claude, Cursor and ChatGPT. The integration exposes your store's behavioural data as tools those clients can call, so you can ask questions in context rather than exporting anything.
Do I need to install a tracking snippet?
No. DynoWeb installs as a Shopify app and collects on your existing theme — no snippet to paste, no tag manager, and nothing to keep in sync when the theme changes.
Can I use this alongside our own analytics stack?
Yes. Nothing here asks you to move off what you already run. The value is the behavioural layer and the proposed change, both of which sit alongside whatever reporting you already trust.

Stop guessing. Fix it and see the money.

Install DynoWeb on your Shopify store and see your first findings within a day of normal traffic.

Questions first? Talk to us or read the merchant stories.