🚀 Symfony MCP Server 1.4.0 - Unleashing Agentic Potential with Sampling

[klapaudius]

We're thrilled to announce the release of Symfony MCP Server 1.4.0, featuring a groundbreaking addition that transforms how your MCP tools can interact with Large Language Models. This release introduces comprehensive Sampling Support, enabling true agentic behavior in your Symfony applications.

🧠 What is Sampling and Why It Matters

Sampling is a revolutionary feature that allows your MCP tools to request LLM assistance during execution, creating a powerful feedback loop between your application logic and AI reasoning. This opens up a new paradigm of agentic applications where tools can:

• Think and reason about complex problems by consulting LLMs
• Make intelligent decisions based on dynamic context
• Generate sophisticated responses that adapt to user needs
• Collaborate with AI to solve problems beyond traditional tool capabilities

🔥 Key Features in 1.4.0

🎯 Comprehensive Sampling Architecture

• SamplingAwareToolInterface: Transform your tools into intelligent agents
• SamplingClient: Powerful service for managing LLM interactions
• Model Preferences: Fine-tune LLM selection based on cost, speed, and intelligence priorities
• Multi-Message Conversations: Enable complex reasoning with context preservation

🛠️ Smart Tool Integration

Tools implementing SamplingAwareToolInterface automatically receive the SamplingClient service, making it effortless to add AI-powered reasoning to your existing tools.

🎨 Flexible Sampling Types

• Text Sampling: Simple text-based AI requests
• Multi-Message Sampling: Complex conversations with preserved context
• Dynamic Model Selection: Choose the right LLM for each task

🌟 Real-World Agentic Examples

Intelligent Code Analysis

class CodeAnalyzerTool implements SamplingAwareToolInterface
{
    private SamplingClient $samplingClient;

    public function execute(array $arguments): ToolResultInterface
    {
        // Check if sampling is available before using it
        if ($this->samplingClient->canSample()) {
            try {
                // Use AI to analyze code for security vulnerabilities
                $response = $this->samplingClient->createTextRequest(
                    'Analyze this code for security issues: ' . $arguments['code'],
                    new ModelPreferences(
                        [['name' => 'claude-3-sonnet']],
                        0.2, // costPriority
                        0.8, // intelligencePriority 
                        0.2  // speedPriority
                    ),
                    'You are a security expert analyzing code for vulnerabilities.',
                    2000
                );
                
                // Return the AI analysis result
                return new TextToolResult($response->getContent()->getText());
            } catch (\Exception $e) {
                // Fallback if sampling fails
                return new TextToolResult('Static analysis: Please review code manually for security issues.');
            }
        }
        
        // Fallback when sampling is not available
        return new TextToolResult('Code analysis requires AI sampling capability.');
    }
    
    public function setSamplingClient(SamplingClient $samplingClient): void
    {
        $this->samplingClient = $samplingClient;
    }
}

Dynamic Content Generation

class ContentGeneratorTool implements SamplingAwareToolInterface
{
    private SamplingClient $samplingClient;

    public function execute(array $arguments): ToolResultInterface
    {
        if ($this->samplingClient->canSample()) {
            try {
                // Generate personalized content using AI
                $response = $this->samplingClient->createTextRequest(
                    "Generate a blog post about {$arguments['topic']} " .
                    "targeting {$arguments['audience']} with tone: {$arguments['tone']}",
                    new ModelPreferences(
                        [['name' => 'claude-3-sonnet']],
                        0.3, // costPriority
                        0.7, // intelligencePriority
                        0.4  // speedPriority
                    ),
                    'You are a professional content writer creating engaging blog posts.',
                    1500
                );
                
                return new TextToolResult($response->getContent()->getText());
            } catch (\Exception $e) {
                return new TextToolResult('Content generation failed. Please try again.');
            }
        }
        
        return new TextToolResult('Content generation requires AI sampling capability.');
    }
    
    public function setSamplingClient(SamplingClient $samplingClient): void
    {
        $this->samplingClient = $samplingClient;
    }
}

🚀 Agentic Potential Unlocked

With sampling, your MCP tools become true AI agents capable of:

🧩 Complex Problem Solving

Tools can break down complex tasks, consult AI for reasoning, and provide sophisticated solutions that adapt to context.

🤖 Intelligent Decision Making

Make smart choices based on real-time analysis rather than static rules, enabling dynamic behavior that scales with problem complexity.

🔄 Iterative Refinement

Tools can engage in back-and-forth conversations with LLMs, refining responses and improving output quality through multiple iterations.

📊 Context-Aware Processing

Leverage AI's understanding of context to provide more relevant and personalized responses to user requests.

🛡️ Enterprise-Ready Features

• Automatic Capability Detection: Graceful fallback when sampling isn't available
• Secure Transport: Maintains the package's commitment to secure StreamableHTTP and SSE transport
• Model Preference Controls: Fine-tune costs and performance based on business requirements
• Comprehensive Documentation: Full guides and examples for immediate implementation

🎯 Getting Started with Sampling

1. Update your tools to implement SamplingAwareToolInterface
2. Inject the SamplingClient (automatically provided)
3. Create sampling requests with your desired prompts
4. Set model preferences for optimal performance
5. Deploy agentic behavior in your applications

🔧 Additional Improvements

• Bug Fixes: Resolved Use same option name on tests commands as input #45 - Renamed "arguments" to "input" in TestMcpPromptCommand for consistency
• Enhanced Documentation: Comprehensive sampling guides and updated README
• Better Testing: Improved test coverage for sampling components

📚 Resources

• Sampling Documentation: Complete guide with examples
• Building Tools: Updated with sampling integration
• Example Tools: See CodeAnalyzerTool for sampling in action

🌈 What's Next?

This release represents a significant leap toward truly agentic applications. With sampling, your Symfony MCP Server becomes a platform for building intelligent, adaptive tools that can reason, learn, and evolve with your users' needs.

The future of web applications is agentic, and Symfony MCP Server 1.4.0 puts that power in your hands.

---

Ready to build the next generation of intelligent applications?

composer require klapaudius/symfony-mcp-server:^1.4.0

Join the discussion in our GitHub Discussions and share your agentic creations!

---

Built with ❤️ by Boris AUBE and the amazing contributors