MCP for Laravel

Laravel 的 MCP

Laravel 的 MCP（Model Context Protocol）— 一个用于构建智能应用的强大 AI 框架。基于 inspector-apm/neuron-ai

需求

• PHP: ^8.0
• Laravel: ^11.0

官方文档

前往官方文档

安装

安装最新版本的软件包：

composer require settled/mcp-laravel

创建 Agent

MCP for Laravel 为您提供了 Agent 类，您可以扩展它来继承该框架的主要功能， 并创建功能完善的 Agent。该类自动为您管理一些高级机制，例如内存、 工具和函数调用，直至 RAG 系统。让我们创建第一个 Agent，扩展 Settled\MCP\Agent 类：

use Settled\MCP\Agent;
use Settled\MCP\SystemPrompt;
use Settled\MCP\Providers\AIProviderInterface;
use Settled\MCP\Providers\Anthropic\Anthropic;

class YouTubeAgent extends Agent
{
    public function provider(): AIProviderInterface
    {
        return new Anthropic(
            key: 'ANTHROPIC_API_KEY',
            model: 'ANTHROPIC_MODEL',
        );
    }

    public function instructions(): string
    {
        return new SystemPrompt(
            background: ["You are an AI Agent specialized in writing YouTube video summaries."],
            steps: [
                "Get the url of a YouTube video, or ask the user to provide one.",
                "Use the tools you have available to retrieve the transcription of the video.",
                "Write the summary.",
            ],
            output: [
                "Write a summary in a paragraph without using lists. Use just fluent text.",
                "After the summary add a list of three sentences as the three most important take away from the video.",
            ]
        );
    }
}

SystemPrompt 类旨在接受您的基础指令并为底层模型构建一致的提示词 以减少提示词工程的工作量。

与 Agent 对话

向 Agent 发送提示词以从底层大模型获得响应：

$agent = YouTubeAgent::make();

$response = $agent->run(new UserMessage("Hi, I'm Valerio. Who are you?"));
echo $response->getContent();
// I'm a friendly YouTube assistant to help you summarize videos.


$response = $agent->run(
    new UserMessage("Do you know my name?")
);
echo $response->getContent();
// Your name is Valerio, as you said in your introduction.

如上例所示，Agent 自动拥有对正在进行的对话的内存。了解更多关于内存的信息，请参阅文档。

支持的大模型提供商

使用 Neuron，您可以仅用一行代码在大模型提供商之间切换，对您的 Agent 实现没有任何影响。 支持的提供商：

• Anthropic
• Ollama（也可用作嵌入提供商）
• OpenAI
• Mistral
• Deepseek

工具和函数调用

您可以通过工具数组向您的 Agent 添加执行具体任务的能力：

use NeuronAI\Agent;
use NeuronAI\SystemPrompt;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Anthropic\Anthropic;
use NeuronAI\Tools\Tool;
use NeuronAI\Tools\ToolProperty;

class YouTubeAgent extends Agent
{
    public function provider(): AIProviderInterface
    {
        return new Anthropic(
            key: 'ANTHROPIC_API_KEY',
            model: 'ANTHROPIC_MODEL',
        );
    }

    public function instructions(): string
    {
        return new SystemPrompt(
            background: ["You are an AI Agent specialized in writing YouTube video summaries."],
            steps: [
                "Get the url of a YouTube video, or ask the user to provide one.",
                "Use the tools you have available to retrieve the transcription of the video.",
                "Write the summary.",
            ],
            output: [
                "Write a summary in a paragraph without using lists. Use just fluent text.",
                "After the summary add a list of three sentences as the three most important take away from the video.",
            ]
        );
    }

    public function tools(): array
    {
        return [
            Tool::make(
                'get_transcription',
                'Retrieve the transcription of a youtube video.',
            )->addProperty(
                new ToolProperty(
                    name: 'video_url',
                    type: 'string',
                    description: 'The URL of the YouTube video.',
                    required: true
                )
            )->setCallable(function (string $video_url) {
                // ... retrieve the video transcription
            })
        ];
    }
}

了解更多关于工具的信息，请参阅文档。

MCP 服务连接器

您可以使用 McpConnector 组件连接由 MCP 服务公开的工具，而不是手动实现工具：

use NeuronAI\Agent;
use NeuronAI\MCP\McpConnector;
use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Anthropic\Anthropic;
use NeuronAI\Tools\Tool;
use NeuronAI\Tools\ToolProperty;

class SEOAgent extends Agent
{
    public function provider(): AIProviderInterface
    {
        return new Anthropic(
            key: 'ANTHROPIC_API_KEY',
            model: 'ANTHROPIC_MODEL',
        );
    }

    public function instructions(): string
    {
        return new SystemPrompt(
            background: ["Act as an expert of SEO (Search Engine Optimization)."]
            steps: [
                "Analyze a text of an article.",
                "Provide suggestions on how the content can be improved to get a better rank on Google search."
            ],
            output: ["Structure your analysis in sections. One for each suggestion."]
        );
    }

    public function tools(): array
    {
        return [
            // Connect an MCP server
            ...McpConnector::make([
                'command' => 'npx',
                'args' => ['-y', '@modelcontextprotocol/server-everything'],
            ])->tools(),

            // Implement your custom tools
            Tool::make(
                'get_transcription',
                'Retrieve the transcription of a youtube video.',
            )->addProperty(
                new ToolProperty(
                    name: 'video_url',
                    type: 'string',
                    description: 'The URL of the YouTube video.',
                    required: true
                )
            )->setCallable(function (string $video_url) {
                // ... retrieve the video transcription
            })
        ];
    }
}

了解更多关于 MCP 连接器的信息，请参阅文档。

实现 RAG 系统

对于 RAG 用例，您必须扩展 NeuronAI\RAG\RAG 类，而不是默认的 Agent 类。

要创建 RAG，您需要附加一些额外的组件，除了 AI 提供商，例如 向量存储 和 嵌入提供商。

以下是 RAG 实现的示例：

use NeuronAI\Providers\AIProviderInterface;
use NeuronAI\Providers\Anthropic\Anthropic;
use NeuronAI\RAG\Embeddings\EmbeddingsProviderInterface;
use NeuronAI\RAG\Embeddings\VoyageEmbeddingProvider;
use NeuronAI\RAG\RAG;
use NeuronAI\RAG\VectorStore\PineconeVectoreStore;
use NeuronAI\RAG\VectorStore\VectorStoreInterface;

class MyChatBot extends RAG
{
    public function provider(): AIProviderInterface
    {
        return new Anthropic(
            key: 'ANTHROPIC_API_KEY',
            model: 'ANTHROPIC_MODEL',
        );
    }

    public function embeddings(): EmbeddingsProviderInterface
    {
        return new VoyageEmbeddingProvider(
            key: 'VOYAGE_API_KEY',
            model: 'VOYAGE_MODEL'
        );
    }

    public function vectorStore(): VectorStoreInterface
    {
        return new PineconeVectoreStore(
            key: 'PINECONE_API_KEY',
            indexUrl: 'PINECONE_INDEX_URL'
        );
    }
}

了解更多关于 RAG 的信息，请参阅文档。

官方文档

前往官方文档

贡献

我们鼓励您为 Neuron AI 框架的开发做出贡献！ 请查看贡献指南了解如何进行。加入我们！

许可证

此软件包在 MIT 许可证下授权。