AI Dev Overview

This document describes how to integrate and use the AiDev module to connect your Spryker application to AI development tools through the Model Context Protocol (MCP).

Overview

The AiDev module provides an MCP server that enables AI assistants to interact with your Spryker application. It exposes Spryker-specific information through MCP tools and prompts, allowing AI assistants to better understand and work with your codebase.

The module includes:

• MCP Server: A console command that runs an MCP server to communicate with AI assistants
• Extension Points: Plugin interfaces for adding custom MCP tools and prompts
• Built-in Tools: Pre-configured tools for accessing Spryker transfers, interfaces, and OMS information
• Prompt Generation: Automatic generation of context-aware prompts from documentation

Install the AiDev module

1. Require the package:

composer require spryker-sdk/ai-dev --dev

2. Generate transfers:

console transfer:generate

Console commands

The AiDev module provides two console commands:

MCP Server Command

The ai-dev:mcp-server command starts an MCP server that allows AI assistants to interact with your Spryker application.

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

This command:

• Starts an MCP server using stdio transport
• Registers all configured MCP tool and prompt plugins
• Automatically generates prompts if they don’t exist
• Listens for requests from AI assistants

Usage: This command is typically configured in AI assistant tools (like Claude Desktop) to enable them to access Spryker-specific information.

Generate Prompts Command

The ai-dev:generate-prompts command generates MCP prompts from a configured Prompt Library.

docker/sdk console ai-dev:generate-prompts

This command:

• Fetches prompts
• Generates PHP-based prompt classes
• Stores generated prompts in the configured directory

Usage: Use this command when you need to regenerate prompts from updated documentation or when initializing the module for the first time.

Extension points

The AiDev module provides plugin interfaces for extending the MCP server with custom functionality:

AiDevMcpToolPluginInterface

Implement this interface to add custom MCP tools that AI assistants can use to query or interact with your application.

Interface location: SprykerSdk\Zed\AiDev\Dependency\AiDevMcpToolPluginInterface

Integration: Register your tool plugins in AiDevDependencyProvider::getMcpToolPlugins():

<?php

namespace Pyz\Zed\AiDev;

use SprykerSdk\Zed\AiDev\AiDevDependencyProvider as SprykerAiDevDependencyProvider;
use Pyz\Zed\AiDev\Communication\Plugins\CustomAiDevMcpToolPlugin;

class AiDevDependencyProvider extends SprykerAiDevDependencyProvider
{
    /**
     * @return array<\SprykerSdk\Zed\AiDev\Dependency\AiDevMcpToolPluginInterface>
     */
    protected function getMcpToolPlugins(): array
    {
        return array_merge(parent::getMcpToolPlugins(), [
            new CustomAiDevMcpToolPlugin(),
        ]);
    }
}

AiDevMcpPromptPluginInterface

Implement this interface to add custom MCP prompts that provide context or instructions to AI assistants.

Interface location: SprykerSdk\Zed\AiDev\Dependency\AiDevMcpPromptPluginInterface

Integration: Register your prompt plugins in AiDevDependencyProvider::getMcpPromptPlugins():

<?php

namespace Pyz\Zed\AiDev;

use SprykerSdk\Zed\AiDev\AiDevDependencyProvider as SprykerAiDevDependencyProvider;
use Pyz\Zed\AiDev\Communication\Plugins\CustomAiDevMcpPromptPlugin;

class AiDevDependencyProvider extends SprykerAiDevDependencyProvider
{
    /**
     * @return array<\SprykerSdk\Zed\AiDev\Dependency\AiDevMcpPromptPluginInterface>
     */
    protected function getMcpPromptPlugins(): array
    {
        return [
            new CustomAiDevMcpPromptPlugin(),
        ];
    }
}

Configuration

The AiDev module can be configured through the AiDevConfig class. Key configuration options include:

• Prompt Directory: Set the directory where generated prompts are stored

Refer to the module’s configuration class for available options and their default values.

Debugging the MCP server

Before connecting your MCP server to AI assistants, you can test and debug it using the MCP Inspector tool. The inspector provides a web interface to interact with your MCP server, test tools, and verify that everything works correctly.

Using the MCP Inspector

Navigate to your Spryker project directory and run:

npx @modelcontextprotocol/inspector docker/sdk console ai-dev:mcp-server -q

This command will:

• Start the MCP Inspector in your browser
• Connect to your local MCP server
• Display all available tools and prompts
• Allow you to test tool calls interactively

With Xdebug

npx @modelcontextprotocol/inspector docker/sdk cli -x console ai-dev:mcp-server