MCP: connecting Pricing Fox with an AI assistant#

The MCP connection lets you link Pricing Fox with an AI client, such as Claude Desktop, Claude Code, Cursor, or other tools supporting the Model Context Protocol.

Thanks to MCP, an AI assistant can securely access your project’s data and help you manage it using ordinary instructions in natural language. So you don’t have to search for everything manually in the app’s interface – you can simply ask the AI assistant about selected things.

You can ask it, for example:

  1. “How many products are currently being processed in the project?”

  2. “Find me products that have a low margin.”

  3. “List the rules used for repricing.”

  4. “Check whether there are products in the project without a purchase price.”

  5. “Summarize what changed in the project over the last 30 days.”

  6. “Find products that are cheaper than the competition.”

The available options always depend on which MCP tools Pricing Fox has available for the given project and what permissions the access token used has.

What MCP is#

MCP stands for Model Context Protocol. It’s a communication standard that lets AI tools securely work with external applications, services, and data.

In practice, this means an AI assistant doesn’t have to answer based only on general knowledge. After connecting via MCP, it can work with specific data from your project in Pricing Fox and perform the available actions on it.

Simply put:

MCP is a bridge between Pricing Fox and your AI client. Thanks to it, the AI assistant gains access to selected Pricing Fox tools and can use them according to your instructions.

What MCP is useful for#

MCP is useful mainly when you want to work with the project faster, more naturally, or at a larger scale.

Instead of clicking manually in the app, you can ask the AI assistant a question or give it a task.

Typically it can come in handy, for example, for:

  1. a quick check of products,

  2. finding problematic items,

  3. checking margins,

  4. working with rules,

  5. getting your bearings in repricing,

  6. summarizing project performance,

  7. analyzing the assortment,

  8. checking the competition,

  9. preparing materials for further work.

Example:

Instead of manually filtering products by margin, you can ask the AI assistant:

Find me products with a margin lower than 10 % and sort them by turnover.

The AI client asks Pricing Fox for the corresponding data via MCP and returns the result to you directly in the conversation.

Where to find MCP in Pricing Fox#

You’ll find MCP in Pricing Fox in several places.

1. In Project settings#

MCP is displayed in the Settings part, in the Active connections block.

Among the other connections, you’ll find the MCP card here. By clicking the question-mark icon, you open a side panel with a connection guide. In it you’ll find basic information about MCP, the address of the Pricing Fox MCP server, and the procedure for connecting to selected AI clients.

2. On the Connections page#

You’ll also find MCP on the Connections page, specifically in the Input data sources part.

It’s displayed here among the other input connections, such as Mergado, Heureka, Zboží.cz, Google Merchant, Google Analytics 4, Shoptet, or a custom CSV source.

MCP is classified here as an input connection, because it lets the AI client access the project’s data via the Model Context Protocol.

So it’s not a classic product or statistics data source, but a technical link between Pricing Fox and the AI assistant.

By clicking the question-mark icon on the MCP card, you open the same information panel with the connection guide.

3. On the Overview page#

MCP is also displayed on the home Overview page, in the Active connections block.

Thanks to this, you can immediately see that MCP is part of the project’s active connections. As with the other connections, the card shows the connection status here and lets you quickly get your bearings on which sources and outputs the project currently has available.

If you see an information icon next to MCP, you can open the guide for setting up the connection with an AI client through it.

In the interface, MCP behaves like a separate connection. In the overviews it’s displayed alongside the other sources, but its purpose is different from that of ordinary data connectors. It serves to let an external AI client securely connect to Pricing Fox and work with the available tools and data of a specific project.

What you’ll find in the MCP panel?#

In the panel you’ll find:

  1. a brief explanation of what MCP is for,

  2. the address of the Pricing Fox MCP server,

  3. the procedure for obtaining an access token,

  4. a sample configuration for VSCode,

  5. a sample connection for Claude Code

Technically:

The MCP server communicates at the address:

https://app-pf.bdfx.cz/mcp

You paste this address into the configuration of the AI client you want to connect with Pricing Fox.

How the connection works#

The connection is made via the Pricing Fox MCP server and an access token from Mergado.

In simple terms:

  1. In Mergado, you create a Personal Access Token, or PAT for short.

  2. You paste the token into your AI client’s configuration.

  3. Using the token, the AI client connects to the Pricing Fox MCP server.

  4. Pricing Fox verifies the permissions and makes the available tools for working with the project accessible.

Technically:

Authorization is done via an HTTP header:

Authorization: Bearer mergado_pat_token

The token serves to verify that the AI client has permission to work with the project’s data. The scope of available actions depends on the permissions assigned to the token.

How to obtain a PAT access token#

To connect Pricing Fox with an AI client, you need to create an access token in Mergado.

Procedure:

  1. Log in to your account in Mergado.

  2. Go to your profile settings.

  3. Open the API and access section.

  4. In the Personal access tokens part, click Add.

  5. Fill in the required items.

  6. Set the token’s validity.

  7. Select the permissions.

  8. Click Create token.

  9. Copy the generated token and store it securely.

Important:

The generated token won’t be displayed again after you close the page. If you lose it, you’ll need to create a new one.

In the current guide in the app, we recommend setting:

  • the token validity to the maximum available period,

  • the permissions according to what the AI client is supposed to do in Pricing Fox.

If the AI assistant is to help with full-fledged management of the project, it will need the corresponding permissions. If you want to use it only to check and read data, set read-only permissions where it makes sense.

Security recommendation:

Work with the token similarly to a password. Don’t send it to anyone, don’t insert it into public repositories, and don’t share it in unprotected documents.

Connecting to AI clients#

After creating the token, you need to add the Pricing Fox MCP server to the AI client’s configuration.

The specific procedure may differ depending on which tool you use. In the app you’ll find a sample for VSCode and Claude Code.

Connecting in VSCode#

Insert the following configuration and replace the mergado_pat_token value with your token:

{
  "servers": {
    "pricing-fox": {
      "url": "https://app-pf.bdfx.cz/mcp",
      "type": "http",
      "headers": {
        "Content-Type": "application/json",
        "Accept": "application/json, text/event-stream",
        "Authorization": "Bearer mergado_pat_token"
      }
    }
  },
  "inputs": []
}

After saving the configuration, the AI client should recognize the Pricing Fox MCP server and make its tools available.

Connecting in Claude Code#

Insert the command into the console:

claude mcp add --transport http pricing-fox https://app-pf.bdfx.cz/mcp \
  --header "Authorization: Bearer mergado_pat_token"

Replace the mergado_pat_token value with your actual access token.

After connecting, you can verify that the MCP server is available directly in Claude Code. You can then start entering queries or tasks related to your project in Pricing Fox.

What the AI assistant can do via MCP#

The AI assistant can work only with the data and functions that Pricing Fox makes available to it via MCP.

It can help you with these areas, for example:

Products:

The AI can help find and check products by the available parameters.

For example:

  1. products with a low or negative margin,

  2. products without a purchase price,

  3. products without a rule,

  4. products with a significant price change,

  5. products by reference price source,

  6. products by position relative to the competition.

Rules:

The AI can help you get your bearings in the configured rules and explain how prices are adjusted in the project.

For example:

  1. list the active rules,

  2. summarize the logic of a specific rule,

  3. find the products a rule applies to,

  4. suggest how to check the rules.

Statistics:

If the project has performance data available, the AI can help interpret it.

For example:

  1. summarize orders, turnover, or clicks,

  2. find high-performing products,

  3. compare the performance of selected groups,

  4. alert you to significant changes.

Competition and reference:

If the project has competitor data available, the AI can help evaluate it.

For example:

  1. find products where your store is more expensive than the competition,

  2. find products where you’re the cheapest,

  3. check prices according to the reference source,

  4. evaluate products by position among offers.

An important limitation: MCP works within a specific project#

The Pricing Fox MCP currently works with tools inside a specific project. This means the AI client can’t generally list all the Pricing Fox projects you have access to.

In other words:

  • MCP can work with the data of a specific project,

  • it can’t automatically list all your Pricing Fox projects,

  • to work, you need to specify a specific project or work in its context.

If you want to deal with a specific project with the AI assistant, it’s advisable to clearly tell it in the instruction which project it is, or to work in an environment where the given project is already selected.

Security and working with data#

The MCP connection lets the AI client access the project’s data, so it’s important to work with it carefully.

We recommend:

  • using only AI clients you trust,

  • storing the token securely,

  • not sharing the token in chat, email, or public files,

  • regularly checking where you use the token,

  • revoking a token you no longer need,

  • not entering sensitive data into AI instructions unless necessary.

Technically:

The PAT serves to authorize requests between the AI client and Pricing Fox. If you revoke the token or change its permissions, this can also affect the functioning of the connection.

How to tell that MCP is connected correctly#

After setup, the AI client should display the available Pricing Fox MCP tools or be able to answer a query concerning the project.

You can try a simple query, for example:

Check the basic information about the current project in Pricing Fox.

Or:

List the available Pricing Fox MCP tools.

If the AI client returns an authorization error, check mainly:

  • whether the token is copied correctly,

  • whether you’re using the Bearer format in the header,

  • whether the token hasn’t been revoked or expired,

  • whether the token has the necessary permissions,

  • whether the MCP server address is entered correctly.

The most common problems#

The AI client reports a 401 error#

A 401 error usually means an authorization problem.

Check:

  • whether the token is valid,

  • whether you pasted it in full,

  • whether it’s stated in the correct format in the configuration,

  • whether you’re using the Authorization: Bearer ... header.

The AI client doesn’t see the Pricing Fox tools#

Check:

  • whether the server address is entered correctly,

  • whether the AI client supports MCP over HTTP,

  • whether you saved the configuration,

  • whether the client needs to be restarted.

The AI can’t find all projects#

This is expected behavior. The Pricing Fox MCP currently doesn’t work with a list of all projects, but with tools within a specific project.

The AI doesn’t have access to some data#

The reason may be:

  • missing token permissions,

  • data that isn’t available in the project,

  • an action not supported by the current set of MCP tools,

  • working outside the context of a specific project.

Practical usage examples#

A quick project check#

Summarize the current state of the project and alert me to things I should check.

Checking margins#

Find products with a margin under 10 % and sort them from the lowest margin.

Checking rules#

Explain to me which rules the project currently reprices by.

Checking products without the necessary data#

Find products that are missing a purchase price or a reference price.

Competitor analysis#

Find products where we're significantly more expensive than the main competition.

A summary for the team#

Prepare a short summary of the project's state for my colleagues: what's working, what needs checking, and where you see the biggest risks.

Technical notes#

  • The Pricing Fox MCP server is available at the address https://app-pf.bdfx.cz/mcp.

  • The connection is made over HTTP.

  • Authorization is done using a Mergado PAT token.

  • The token is passed in the Authorization HTTP header.

  • The AI client works only with the tools that Pricing Fox makes available via MCP.

  • The scope of available actions depends on the token’s permissions.

  • The Pricing Fox MCP currently works in the context of a specific project, not over a list of all projects.

  • Simply opening the guide in the app doesn’t change any data.

  • An MCP connection doesn’t mean the AI automatically makes changes without your instruction – actions always stem from the available tools, permissions, and the specific request in the AI client.

Summary#

MCP in Pricing Fox lets you connect your project with an AI client and work with data using natural language. Thanks to this, you can check products, rules, margins, performance, or the competitive situation faster, without having to go through the individual parts of the app manually.

To connect, you just need to:

  1. create a PAT token in Mergado,

  2. paste it into the AI client’s configuration,

  3. connect the Pricing Fox MCP server,

  4. start working with the project through the AI assistant.

MCP is suitable mainly for users who want to control Pricing Fox more efficiently, analyze data faster, and use AI as a helper in day-to-day pricing management.

Backward Zboží.cz Output Forward