README

MCP Server Plugin

A Sylius plugin that exposes a Model Context Protocol (MCP) server for enabling interaction with your store via Large Language Models (LLMs) like ChatGPT.

Installation

Requirements

We work on stable, supported and up-to-date versions of packages. We recommend you to do the same.

Beware!

This installation instruction assumes that you're using Symfony Flex.

1. Require plugin with composer:

  composer require sylius/mcp-server-plugin

Clear application cache by using command:

  bin/console cache:clear

Documentation

What is MCP Server?

The Model Context Protocol (MCP) is a standardized way to connect language models (like ChatGPT) with external tools, APIs, and systems. It allows an AI model to make structured tool calls — similar to calling functions — during a conversation.

The MCP Server acts as a bridge between the language model and your application logic. It exposes a list of tools (like “search for a product” or “create an order”) and executes them in response to requests from the AI.

This plugin integrates Sylius with an MCP Server, enabling AI agents to interact with your store (e.g., search products, check prices, start checkout).

We use the official php-mcp/server package to provide the MCP server runtime.

┌────────────────┐                  ┌───────────────┐       ┌────────┐
│  MCP Client    │◄────────────────►│  MCP Server   │◄─────►│ Sylius │
│ (OpenAi, etc.) │ (Stdio/HTTP/SSE) │ (Tools, etc.) │ (API) │        │
└────────────────┘                  └───────────────┘       └────────┘

To learn more, see the official MCP introduction at modelcontextprotocol.io.

Running the MCP Server

You can run the server using the following command:

  bin/console sylius:mcp-server:start

By default, the server runs at: http://localhost:8080/mcp and it uses Streamable HTTP Transport.

MCP Server Configuration

Create a file config/packages/sylius_mcp_server.yaml and customize it if needed. Below is the default configuration:

sylius_mcp_server:
  server:
    name:    'Sylius MCP Server'
    version: '0.1.0'
    transport:
      host: 127.0.0.1
      port: 8080
      prefix: 'mcp'
      enable_json_response: false
      ssl:
        enabled: false
        context: []

    session:
      driver: cache
      ttl: 3600

    discovery:
      locations:
        - { base_path: '%sylius_mcp_server.plugin_root%', scan_dirs: ['src/Tool'] }

You can enable SSL support by setting ssl.enabled to true and configuring the appropriate SSL context. For details on how to configure the context, refer to the SSL section in the php-mcp/server documentation.

Transport

By default, the server uses the Streamable HTTP Transport. You may implement your own transport by creating a custom factory that implements Sylius\McpServerPlugin\Factory\ServerTransportFactoryInterface. Then, override the transport in the McpServerCommand service definition.

Sylius API

Tools communicate with the Sylius API through pre-configured HTTP clients:

sylius_mcp_server.http_client.api_shop:
    base_uri: '%sylius_mcp_server.api.shop_base_uri%'
    headers:
        - 'Accept: application/ld+json'
        - 'Content-Type: application/ld+json'
sylius_mcp_server.http_client.api_shop_merge_patch:
    base_uri: '%sylius_mcp_server.api.shop_base_uri%'
    headers:
        - 'Accept: application/ld+json'
        - 'Content-Type: application/merge-patch+json'

The default API base URI is http://localhost:8000/api/v2/shop/, and can be overridden using the environment variable SYLIUS_MCP_SERVER_API_SHOP_BASE_URI.

Tools

The following tools are available out of the box:

These tools currently operate in guest mode only — fetch_order returns data only if the order has no associated customer account.

Adding custom tools

You can extend the server with your own tools by creating PHP classes annotated with attributes like #[McpTool]. These tools will then be exposed to the MCP client just like the built-in ones.

To learn how to define and structure a tool, see the "Defining MCP Elements" section of the php-mcp/server documentation.

Once your tool is ready, make sure the plugin is configured to discover it:

sylius_mcp_server:
  server:
    discovery:
      locations:
        - { base_path: 'your_base_path', scan_dirs: ['your/custom/Tool/Directory'] }

Usage Example

You can use the server directly in OpenAI Playground, exposing it via ngrok or similar tunneling tools.

Example with OpenAI Playground

1. Add a new tool in the OpenAI Playground.

2. Configure the tool with the following settings:

   • URL: http://localhost:8080/mcp (or your ngrok URL)
   • Label: Sylius
   • Authentication: None

Example with API Call

  curl https://api.openai.com/v1/responses \
    -H "Content-Type: application/json" \
    -H "Authorization: Bearer $OPENAI_API_KEY" \
    -d '{
      "model": "gpt-4.1",
  
      "tools": [
        {
          "type": "mcp",
          "server_label": "Sylius",
          "server_url": "$YOUR_MCP_SERVER_URL",
          "require_approval": "never"
        }
      ],
  
      "input": [
        {
          "role": "user",
          "content": [
            {
              "type": "input_text",
              "text": "Find me red shoes"
            }
          ]
        }
      ],
      "max_output_tokens": 1024
    }'

Security issues

If you think that you have found a security issue, please do not use the issue tracker and do not post it publicly. Instead, all security issues must be sent to security@sylius.com

Community

For online communication, we invite you to chat with us & other users on Sylius Slack.

License

This plugin's source code is completely free and released under the terms of the MIT license.