djinson / laravel-mcp
The ultimate Laravel integration for Model-Centric Processing (MCP). Seamlessly connect Azure OpenAI, OpenAI, and Google Gemini to build powerful AI agents with tool use, conversation orchestration, and robust error handling.
Installs: 0
Dependents: 0
Suggesters: 0
Security: 0
Stars: 2
Watchers: 0
Forks: 0
Open Issues: 0
pkg:composer/djinson/laravel-mcp
Requires
- php: ^8.2
- illuminate/support: ^10.0|^11.0|^12.0
- spatie/laravel-package-tools: ^1.16
Requires (Dev)
- orchestra/testbench: ^9.0
- pestphp/pest: ^2.0
- pestphp/pest-plugin-laravel: ^2.0
README
Connect your Laravel app to Claude Desktop, OpenAI, and Gemini in minutes. Expose your database and logic as AI tools with zero friction.
Quick Start
composer require djinson/laravel-mcp
Requirements
- PHP: ^8.2
- Laravel: ^10.0 or ^11.0
Features
- Multi-Provider Support: Supports Azure OpenAI, standard OpenAI, and Google Gemini.
- Centralized Configuration: Publishes
config/openai-mcp.phpfor managing credentials, endpoints, model options, retries, and timeouts. - Robust Error Handling: Custom
LlmExceptionfor specific error handling and structured logging for better observability. - Conversation Orchestration: Manages multi-turn conversations, tool execution, and cycle detection.
- Prompt Management: Publishes prompt stubs to
resources/ai-prompts/. - Tool Auto-Discovery: Automatically discovers and registers
ToolInterfaceimplementations underapp/MCP/Tools. - Testing Support: Includes a test suite using
orchestra/testbenchfor verification.
Installation
composer require djinson/laravel-mcp
Publish Configuration
Publish the configuration file to config/openai-mcp.php:
php artisan vendor:publish --provider="Djinson\OpenAiMcp\LaravelMcpServiceProvider" --tag="config"
Publish Prompts
Publish the default AI prompts to resources/ai-prompts/:
php artisan vendor:publish --provider="Djinson\OpenAiMcp\LaravelMcpServiceProvider" --tag="openai-mcp-prompts"
Configuration
Configure your Azure OpenAI credentials and settings in your .env file:
AZURE_OPENAI_KEY=your-api-key AZURE_OPENAI_ENDPOINT=https://your-resource-name.openai.azure.com AZURE_OPENAI_DEPLOYMENT=gpt-4 AZURE_OPENAI_API_VERSION=2024-02-15-preview
You can also customize retries, timeouts, and model parameters in config/openai-mcp.php.
Testing
To run the package tests, you can use phpunit or pest. If you are on Windows, it is recommended to run tests via WSL.
composer install vendor/bin/phpunit