AI Dev MCP Server

Last updated: December 9, 2025

Edit on GitHub

This document describes how to configure and use the AiDev MCP server to connect AI assistants to your Spryker application.

About Model Context Protocol (MCP)

The Model Context Protocol (MCP) is an open-source standard for connecting AI applications to external systems. For Spryker developers, MCP allows AI assistants like Claude or Copilot to understand your project improving the quality of AI-generated code and recommendations.

Learn more about MCP at modelcontextprotocol.io.

Configure the MCP server

The AiDev module provides an MCP server through the ai-dev:mcp-server console command. To configure it for use with AI assistants, you need the full path to your project and the Docker SDK command.

docker/sdk console ai-dev:mcp-server -q

The -q flag (quiet mode) suppresses unnecessary output, which is important for the MCP stdio transport.

Integration with AI assistants

Claude Code

For Claude Code CLI, add the MCP server using the command line.

Navigate to your Spryker project directory and run:

claude mcp add spryker-project "$(pwd)/docker/sdk console ai-dev:mcp-server -q"

This command will:

Add the MCP server configuration to Claude Code
Use the current project directory path automatically
Configure the server to run in quiet mode

Claude Code will now have access to Spryker-specific tools and prompts through the MCP server. MCP prompts MCP claude code

Claude Desktop

For Claude Desktop application, configure the MCP server in the application settings.

Open Claude Desktop settings
Navigate to the Developer section
Add the following configuration to claude_desktop_config.json:

{
  "mcpServers": {
    "spryker-ai-dev": {
      "command": "/Users/username/projects/spryker-project/docker/sdk",
      "args": [
        "console", 
        "ai-dev:mcp-server", 
        "-q"
      ]
    }
  }
}

Restart Claude Desktop

GitHub Copilot in PHPStorm

For GitHub Copilot Chat in PHPStorm with MCP support (requires PHPStorm 2024.3+):

Open PHPStorm Settings/Preferences
Navigate to Tools � GitHub Copilot � MCP Servers
Add a new server configuration with the following JSON:

{
  "servers": {
    "spryker-mcp": {
      "type": "stdio",
      "command": "/Users/username/projects/spryker-project/docker/sdk",
      "args": [
        "console",
        "ai-dev:mcp-server",
        "-q"
      ]
    }
  },
  "inputs": []
}

Restart PHPStorm

PHPStorm version

MCP support in GitHub Copilot for PHPStorm requires PHPStorm version 2024.3 or later. Check your IDE version and update if necessary.

MCP Copilot

Available MCP tools

The AiDev module provides the following built-in tools that AI assistants can use:

Tool name	Description
`getTransferStructureByName`	Retrieves the structure of a Spryker transfer object by its name. Returns all properties with their types and metadata.
`getTransferStructureByNamespace`	Retrieves the structure of a Spryker transfer object by its fully qualified namespace.
`getInterfaceMethodsByNamespace`	Retrieves all method signatures, parameters, return types, and PHPDoc for a given interface FQN (Fully Qualified Name).
`getOmsTransitionsByState`	Retrieves OMS state machine transitions for a specific state. Returns all transitions that start from the given state, optionally filtered by process name.
`getOrderOmsTransitions`	Retrieves OMS state machine transitions for a specified order from the order’s current state. Helps identify the current state and possible transitions.

AI assistants can automatically discover and use these tools when connected to the MCP server.

MCP prompts

The AiDev module provides contextual prompts to help AI assistants understand Spryker-specific concepts and patterns.

Default prompts

By default, prompts are automatically downloaded from the Spryker Prompt Library when you run the MCP server for the first time. These prompts include:

Spryker architecture patterns
Module structure guidelines
Development best practices
Common workflows and conventions

The prompts are generated and stored in: src/Generated/Shared/Prompts/

Custom prompts

You can extend the default prompts by adding custom prompt files to the prompts directory:

PROJECT_ROOT/data/prompts/

To add custom prompts:

Create the prompts directory if it doesn’t exist:

mkdir -p data/prompts

Add your custom prompt markdown file data/prompts/custom-prompt.md to this directory with the following structure:

---
title: prompt name
description: short description
---

prompt content ...

Regenerate prompts to include your custom ones:

docker/sdk console ai-dev:generate-prompts

Custom prompts should be written in Markdown format and follow the same structure as the default prompts from the Prompt Library. They will be converted into PHP classes and made available to AI assistants through the MCP server.

Managing prompts

To manually regenerate prompts from the Prompt Library and your custom prompts:

docker/sdk console ai-dev:generate-prompts