Getting Started with GemPages MCP

Over the past several years, AI has changed how ecommerce teams research markets, develop campaign ideas, and produce landing pages. Tasks that once required several days can now be completed in hours.

Yet faster output has not removed the operational gaps surrounding it.

Research often happens in one conversation. Product data sits in Shopify. Customer reviews and competitor insights are spread across several sources. Finished HTML may be stored in a local folder while the final page still needs to be transferred into the platform where the ecommerce team will manage it.

Teams continue to move information between tools, repeat instructions, reconcile different versions, and involve developers in work that has already been completed elsewhere.

We also found that many AI solutions concentrated on generating more content or assembling pages faster. Speed matters, but it cannot answer two important questions:

• Is the content based on a clear understanding of the customer?
• Can the finished output move into the team’s actual ecommerce workflow?

These questions shaped the development of GemPages MCP.

GemPages MCP connects compatible AI agents with Shopify and GemPages, giving them access to specialized workflows for researching page content and turning finished HTML into editable GemPages pages.

The goal is not to ask AI to produce more work in isolation. It is to give the AI clearer ecommerce context and the ability to complete useful actions within the tools merchants and marketers already use.

What Is GemPages MCP?

MCP is a standardized framework that allows an AI agent to connect with external data sources and tools. Instead of only generating text inside a chat window, the agent can access approved information and perform defined actions through the connected platforms.

With GemPages MCP, a compatible AI agent can support tasks such as:

• Accessing relevant Shopify product information
• Researching competitors and customers
• Analyzing reviews, pain points, and objections
• Developing structured content directions
• Reading finished HTML files
• Building and validating page sections
• Creating editable pages in GemPages
• Returning preview and editor URLs

The user still directs the task through natural-language instructions. GemPages MCP gives the AI the tools required to carry out that instruction within an ecommerce context.

Why Landing Page Creation Should Start With Customer Research

Before deciding what the page should say, ecommerce teams need answers to questions such as:

• Who is most likely to need this product?
• What problem are they trying to solve?
• How do they describe that problem?
• Which outcomes matter most to them?
• What alternatives have they already considered?
• What concerns could prevent them from purchasing?
• Which product claims require more explanation or proof?
• How much information do they need before they are ready to act?

When this information is missing, AI can still generate polished copy. However, the message may rely on broad assumptions rather than the reasons customers actually consider or reject the product.

Consider a sleep supplement.

A generic page might lead with a promise such as “Sleep better and wake up refreshed.” The claim sounds relevant, but it says little about the specific problem the product addresses or why someone should choose it over another supplement.

Research may reveal several distinct customer concerns:

• Some customers fall asleep easily but wake repeatedly during the night.
• Some have tried other supplements but disliked the next-day grogginess.
• Some question whether another sleep product will work after previous failures.
• Some want support without becoming dependent on a nightly sleep aid.

Each finding points to a different content direction.

The page may need to address interrupted sleep rather than difficulty falling asleep. It may need to explain the morning experience, compare the product with alternatives, or respond to skepticism earlier in the customer journey.

These insights can affect:

• The main message
• The order of product benefits
• The angle used to introduce the problem
• The proof included on the page
• The alternatives used in a comparison
• The objections covered in FAQs
• The information placed before a CTA

Faster content creation is useful. It becomes more valuable when the content begins with a clearer view of the customer and market.

Two Specialized Workflows Available in GemPages MCP

GemPages MCP launches with two workflows that teams can use based on the work they need to complete.

1. Content Research: Turn Product and Market Data Into Content Direction

Useful page insights are often spread across product information, competitor pages, customer reviews, discussion threads, and the language buyers use to describe their experiences.

The Content Research workflow helps the AI collect and analyze these inputs before developing content.

Depending on the product and available sources, the research may examine:

• Shopify product information
• Product features and differentiators
• Competitor positioning
• Competitor value propositions
• Recurring customer review themes
• Common frustrations and pain points
• Desired outcomes
• Purchase motivations
• Questions customers ask before buying
• Objections within the category
• Language customers use to describe their problems
• Claims and content angles used by competing products

The goal is not to collect a large amount of unrelated information. The workflow looks for patterns that can shape how the product is presented.

For example:

• A repeated customer concern may need a direct response in the content.
• A product benefit valued by buyers may deserve greater priority.
• A gap in competitor messaging may provide a clearer point of difference.
• A common failed alternative may support a comparison angle.
• Customer language may help the message sound closer to the way buyers describe the problem themselves.

The findings can support several content formats, including advertorials, listicles, and comparison articles.

Research does more than improve individual headlines. It helps the team decide:

• Which customer problem should lead the narrative
• Which benefits deserve the most attention
• Which objections must be answered
• What proof the audience may expect
• How the product should be positioned against alternatives
• Which content angle best fits the available evidence

The result is a stronger starting point for content development than a generic prompt based only on the product name or description.

2. HTML to GemPages: Turn Finished HTML Into an Editable Page

AI coding tools and developers can create complete HTML landing pages, but the finished code often introduces another handoff.

The HTML still needs to be moved into the page-building environment, checked against the storefront, reviewed across different screen sizes, and revised when the campaign changes.

When the people responsible for the page do not work with code, even a small content update may require further developer support.

The HTML to GemPages workflow addresses this final production gap.

It accepts a finished HTML page and turns it into a page inside GemPages, where the ecommerce team can review, edit, and prepare it for publication.

The HTML can come from:

• An internal developer
• An ecommerce or creative agency
• Claude Code or another AI coding tool
• A Figma-to-HTML process
• An existing campaign workflow
• Another source that produces completed HTML

The workflow does not require teams to change how they research, write, design, or code their pages. It works with the approved HTML they already have.

After receiving the file, the AI:

1. Reads the HTML.
2. Divides the page into semantic sections such as Hero, Features, Testimonials, FAQ, and CTA.
3. Retrieves GemPages’ generation rules.
4. Builds each section separately.
5. Validates the section before adding it to the page.
6. Repeats the process until the page is complete.
7. Returns the preview and editor URLs.

The user can then open the preview, inspect the finished page, continue working in the editor, and publish it manually when ready.

Instead of ending with an isolated code file, the finished work becomes part of the page-building environment the ecommerce team uses.

Start With the Workflow That Matches What You Have

Ecommerce teams rarely begin from the same point. One may have a product but no clear campaign direction. Another may already have a completed HTML page.

GemPages MCP allows each team to start with the input available to them.

Start With a Shopify Product

Use the Content Research workflow when you have a product but need a clearer understanding of:

• The target customer
• Customer pain points
• Purchase motivations
• Common objections
• Competitor messaging
• Potential content angles

The product and market findings can then guide the content your team develops.

This starting point can be useful for:

• New product launches
• Campaign planning
• Products entering a crowded category
• Offers with an unfamiliar mechanism
• Pages that need stronger customer language
• Existing content that feels too broad or generic

Start With Finished HTML

Use the HTML to GemPages workflow when the page has already been written, designed, and coded.

Provide the finished HTML file and ask the AI to build it as an editable GemPages page.

This starting point can be useful when:

• An agency has delivered the final HTML
• A developer has completed the page
• An AI coding tool has generated an approved design
• A Figma design has already been implemented as HTML
• Several HTML page variations need to be brought into GemPages
• The marketing team needs to manage the page after development

These workflows solve different parts of the production process and do not depend on one another. Use the one that matches the work your team needs to complete.

What Makes GemPages MCP Different?

1. It Gives AI Access to Ecommerce Context

A general AI tool can generate ideas based on the information included in a prompt. GemPages MCP allows compatible agents to work with relevant Shopify and GemPages context through defined tools.

This means the workflow can begin with the actual product rather than a brief description copied into a new conversation.

For content research, access to product information provides a stronger foundation for examining the market, customer, and offer.

For page creation, the connection allows the AI to move beyond generating code and create the page where the team will continue working.

2. It Uses Specialized Workflows Rather Than Generic Prompts

A general request such as “research this product” may produce a broad collection of facts. It does not tell the AI which customer signals matter for landing page content or how the findings should affect the message.

The Content Research workflow guides the AI through information that can shape purchase decisions, including:

• Pain points
• Objections
• Purchase motivations
• Competitor gaps
• Customer language
• Content angles

The HTML to GemPages workflow also follows a defined process. It divides the page, retrieves the relevant rules, builds each section, and validates the result before proceeding.

The AI is not left to invent the complete process from a single prompt.

3. It Supports Existing Team Workflows

GemPages MCP does not require every user to follow the same production process.

Marketers can use Content Research before developing a campaign. Agencies and developers can bring finished HTML into the HTML to GemPages workflow. Teams can use either capability without repeating work already completed elsewhere.

This makes the product relevant to different roles:

• Ecommerce marketers researching a new campaign
• Content teams developing product angles
• Solo merchants looking for clearer customer insight
• Agencies producing multiple landing pages
• Developers handing finished pages to marketing teams
• Ecommerce managers responsible for final review and launch

Example Use Cases

1. Research a Product Before Developing Campaign Content

A marketer preparing a campaign can ask the AI to use GemCommerce MCP to research a Shopify product.

The workflow can examine the product, competitors, customer feedback, pain points, objections, and potential angles. The resulting report gives the team a clearer basis for developing an advertorial, listicle, comparison article, or landing page brief.

2. Bring an AI-Coded Page Into GemPages

A marketer or developer may use Claude Code to create a finished HTML landing page.

Instead of leaving the file in the coding environment, the team can ask GemCommerce MCP to build it as a GemPages page. Once complete, the workflow returns the preview and editor URLs.

The team can review the result, make final changes, and publish it manually.

3. Transfer an Agency-Delivered HTML Page

An agency may deliver a page as HTML while the client’s ecommerce team manages campaigns in GemPages.

The HTML to GemPages workflow can handle the transfer. The client receives an editable page rather than a code file that requires continued agency or developer support for every change.

Human Review Remains Necessary

GemPages MCP can reduce repetitive research and production work, but every output should be reviewed before it informs a campaign or becomes customer-facing.

Review Content Research

Before using the findings, check:

• Whether the sources are relevant
• Whether customer statements are represented accurately
• Whether conclusions are supported by the available evidence
• Whether competitor information is current
• Whether the suggested claims can be substantiated
• Whether the recommended angle fits the brand and product

Research may reveal a useful pattern, but the team still needs to decide whether that pattern applies to the campaign.

Review Converted Pages

Before publishing an HTML page created through GemPages MCP, verify:

• Product specifications
• Prices and discounts
• Offer conditions
• Brand language
• Product images
• Customer reviews and testimonials
• Claims and supporting evidence
• Links and CTA destinations
• Desktop, tablet, and mobile presentation
• Forms and interactive elements
• Shopify and GemPages settings

The HTML to GemPages workflow creates the page but does not publish it. Open the preview and editor links, make any necessary changes, and publish only after the page has passed review.

GemPages MCP is designed to bring teams closer to a strong working result. It does not remove the need for accuracy, compliance, brand judgment, or final approval.

GemPages MCP Is Currently in Beta

The available workflows, supported elements, setup process, and related guides may change as the product develops. New capabilities and fixes will be released based on technical progress and feedback from early users.

Access is being introduced gradually so that we can work closely with beta users, identify issues, and improve the experience before broader availability.

During the beta, users may encounter limitations with certain research sources, HTML structures, page formats, or components. The level of editability may also vary depending on the HTML supplied and the elements currently supported.

Feedback from early users will play an important role in shaping future development.

How to Connect to GemPages MCP

After joining the GemPages MCP waitlist, you will receive an email when your access is ready. The email will include the instructions you need to connect GemCommerce MCP to a supported AI tool.

The setup process differs slightly depending on whether you use a personal Claude account or an account managed by an organization.

1. Connect With a Claude Personal Account

Step 1: Add a Custom Connector

Open Claude and go to: Settings → Connectors

Select Add custom connector.

Step 2: Enter the GemCommerce MCP Details

Complete the connector fields with the following information:

• Name: GemCommerce MCP
• Remote MCP URL: https://mcp.gemcommerce.com/

Click Add, then select Connect.

Step 3: Open GemPages Preferences

Claude will redirect you to the GemCommerce connection page.

Select Open GemPages Preferences to continue.

Step 4: Select Your Shopify Store

Choose the Shopify store you want to connect to GemPages MCP.

Check the store name and domain before continuing, especially when your account has access to several stores.

Step 5: Copy the Connection Prompt

Find the MCP Connection section in GemPages Preferences.

Copy the prompt provided in this section.

The prompt contains the information Claude needs to complete the connection with your selected store.

Step 6: Paste the Prompt Into Claude

Return to Claude and paste the copied prompt into the Paste prompt here field.

Click Connect.

Your selected Shopify store should now be connected to GemCommerce MCP in Claude.

Step 7: Allow GemCommerce MCP Tool Calls

Claude requests permission before using an MCP tool by default. Since GemCommerce MCP workflows can involve many tool calls, Claude may repeatedly pause and ask for approval.

To prevent these interruptions:

• Open the GemCommerce MCP connector settings.
• Go to Tool permissions.
• Change all required permissions to Always allow.

For the complete connection process, watch the video guide below:

2. Connect With a Claude Organization Account 

First, check whether GemCommerce MCP already appears in your organization’s available connectors.

If the connector is unavailable, ask your Claude administrator to add it with the following URL:

https://mcp.gemcommerce.com/

Your administrator may need to approve the connector before organization members can access it.

Once the connector is available, follow the same setup process used for a personal account.

3. Connect GemPages MCP to Codex

Step 1: Turn On Developer Mode

Open the following page: ChatGPT Security Settings

Turn on Developer mode.

Developer mode allows you to add a custom plugin through an MCP connection.

Step 2: Add a New Plugin

Open Plugins and click the plus icon (+) to add a plugin.

Step 3: Enter the GemCommerce MCP Details

Complete the plugin fields with the following information:

• Name: GemCommerce MCP
• Connection: https://mcp.gemcommerce.com/

Step 4: Confirm the Custom Plugin Connection

Select: I understand and want to continue

Then click Create.

Step 5: Sign In With GemCommerce MCP

Choose Sign in with GemCommerce MCP. You will be redirected to the connection page.

Step 6: Open GemPages Preferences

Select Open GemPages Preferences.

Step 7: Select Your Shopify Store

Choose the Shopify store you want to connect.

When your account has access to several stores, confirm that you select the correct store before copying the connection prompt.

Step 8: Copy the MCP Connection Prompt

In GemPages Preferences, find the MCP Connection section.

Copy the prompt shown in this section.

Step 9: Paste and Run the Prompt in Codex

Return to Codex and paste the complete connection prompt into the conversation.

Run the prompt to complete the connection.

For the complete connection process, watch the video guide below:

FAQs About Connecting to GemPages MCP

1. What ChatGPT Configuration Is Recommended for GemCommerce MCP?

For better results, use ChatGPT version 5.5 or later and set Reasoning effort to High.

Go to Settings → General and turn on Full access.

When Full access is disabled, ChatGPT may not be able to process an attached file directly. In that case, paste the required content into the conversation instead.

2. What Is Required to Run a GemPages MCP Workflow in ChatGPT?

Two conditions must be met:

1. The instruction must use the correct prompt syntax.
2. GemCommerce MCP must be attached to the current session.

Both conditions are required. The workflow will not run when the MCP is not attached, even when the prompt is written correctly.

Before submitting the prompt, confirm that GemCommerce MCP appears among the tools or plugins available in the current session.

3. Which Claude Configuration Should I Use for GemCommerce MCP Tasks?

The recommended configuration depends on the task:

• Content Research: Claude Opus 5.0 or later
• HTML to GemPages: Claude Opus 5.0 with Reasoning effort set to High

HTML conversion involves multiple section-building and validation steps, so it requires a model that can maintain the page structure and workflow state throughout the run.

4. Can I Connect GemCommerce MCP to a Free Claude or ChatGPT Account?

A free account may be able to connect to GemCommerce MCP.

However, GemCommerce MCP workflows involve multiple tool calls, research steps, and file-processing tasks. These tasks may exceed the processing capacity or usage limits available on a free account.

As a result, free-tier users may experience:

• Incomplete workflow runs
• Inconsistent output
• Earlier usage-limit interruptions
• Difficulty processing long pages or large files

For more consistent results, use the model and configuration recommended for the selected workflow.

5. Does GemPages MCP Have Separate Usage Limits From the GemPages App?

No. Actions performed through GemPages MCP follow the limits of the connected GemPages plan.

Using MCP does not create a separate publishing allowance outside the GemPages app.

6. What Does This Mean for a Store on the GemPages Free Plan?

A store on the GemPages Free plan can publish one page.

This limit applies across the account, regardless of whether the page is created through GemPages MCP or directly through the GemPages interface.

Creating or publishing a page through MCP does not bypass the plan limit.

7. Why Does the Workflow Not Start After I Paste the Prompt?

Check the following:

• GemCommerce MCP is attached to the current session.
• The prompt was copied completely.
• You selected the correct Shopify store.
• The MCP connection is still active.
• The AI tool has permission to call the required tools.
• Your account has not reached its usage limit.
• Your GemPages plan allows the requested action.

In Claude, repeated permission requests may also make the workflow appear to stop. Open Tool permissions and set the required GemCommerce MCP actions to Always allow.

8. Can I Connect More Than One Shopify Store?

You must choose the target Shopify store during the connection process.

When your account has access to several stores, check the selected store before copying and running the MCP connection prompt. This helps prevent the AI from researching or creating a page for the wrong store.