README

AI-powered Symfony bundle to inspect, analyze, and interact with logs using Large Language Models (LLMs) and Model Context Protocol (MCP) agents.

Features

🤖 AI-Powered Log Analysis: Leverage LLMs to understand and analyze your application logs
🔍 Intelligent Log Search: Vector-based semantic search through log entries
📊 Log Summarization: Automatically summarize and extract insights from logs
🛠️ Extensible Tool System: Add custom analysis tools using Symfony's #[AsTool] attribute
⚙️ Multiple AI Providers: Support for OpenAI, Anthropic, Ollama, and more
💾 Vector Storage: Multiple vector store backends (Chroma, Pinecone, Weaviate)
📱 Console Commands: CLI interface for log analysis and indexing
🔄 Auto-Indexing: Automatic log indexing via event listeners

Installation

Install the bundle using Composer:

composer require hakam/ai-log-inspector-bundle

Configuration

Create a configuration file at config/packages/hakam_ai_log_inspector.yaml:

hakam_ai_log_inspector:
    ai_platform:
        provider: 'openai'  # or 'anthropic', 'ollama', etc.
        api_key: '%env(AI_API_KEY)%'
        model:
            name: 'gpt-4'
            capabilities: ['text_generation', 'embeddings']
            options:
                temperature: 0.7

    vector_store:
        provider: 'chroma'  # or 'pinecone', 'weaviate', etc.
        connection_string: '%env(VECTOR_STORE_URL)%'

    log_sources:
        - path: '%kernel.logs_dir%'
          pattern: '*.log'
        - path: '/var/log/application'
          pattern: '*.log'

Environment Variables

Set the required environment variables in your .env file:

AI_API_KEY=your_ai_provider_api_key
VECTOR_STORE_URL=your_vector_store_connection_string

Usage

Console Commands

Ask Questions About Your Logs

php bin/console ai:log-inspector:ask "What errors occurred in the last hour?"
php bin/console ai:log-inspector:ask "Show me all database connection errors"
php bin/console ai:log-inspector:ask "Summarize today's critical issues"

Index Logs for Analysis

php bin/console ai:log-inspector:index

Service Integration

Using the Agent Factory (Recommended)

use Hakam\AiLogInspectorBundle\Factory\LogInspectorAgentFactory;
use Symfony\Component\DependencyInjection\Attribute\Autowire;

class LogAnalysisService
{
    public function __construct(
        #[Autowire(service: 'hakam_ai_log_inspector.agent_factory')]
        private LogInspectorAgentFactory $agentFactory
    ) {}

    public function analyzeError(string $errorMessage): string
    {
        $agent = $this->agentFactory->create();
        return $agent->interact("Analyze this error: {$errorMessage}");
    }

    public function getLogSummary(): string
    {
        $agent = $this->agentFactory->create();
        return $agent->interact("Provide a summary of recent log activity");
    }
}

Service Configuration

# config/services.yaml
services:
    App\Service\LogAnalysisService:
        arguments:
            $agentFactory: '@hakam_ai_log_inspector.agent_factory'

Creating Custom Tools

Create custom analysis tools by implementing LogInspectorToolInterface and using the #[AsTool] attribute:

<?php

namespace App\Tool;

use Hakam\AiLogInspector\Tool\LogInspectorToolInterface;
use Symfony\AI\Agent\Toolbox\Attribute\AsTool;

#[AsTool(
    name: 'error_pattern_detector',
    description: 'Detects specific error patterns in logs'
)]
class ErrorPatternDetectorTool implements LogInspectorToolInterface
{
    public function __invoke(string $pattern, int $hours = 24): array
    {
        // Search for error patterns in the last N hours
        return [
            'pattern' => $pattern,
            'matches' => $this->searchPattern($pattern, $hours),
            'severity' => $this->calculateSeverity($matches)
        ];
    }

    private function searchPattern(string $pattern, int $hours): array
    {
        // Your pattern detection logic
        return [];
    }

    private function calculateSeverity(array $matches): string
    {
        // Calculate severity based on matches
        return 'medium';
    }
}

Tools are automatically discovered and injected into the agent when they implement LogInspectorToolInterface and are tagged with #[AsTool].

Event System

The bundle includes automatic log indexing via event listeners:

kernel.terminate: Indexes logs after request completion
console.terminate: Indexes logs after console command execution

Configuration Examples

OpenAI with Chroma

hakam_ai_log_inspector:
    ai_platform:
        provider: 'openai'
        api_key: '%env(OPENAI_API_KEY)%'
        model:
            name: 'gpt-4'
            options:
                temperature: 0.3
    vector_store:
        provider: 'chroma'
        connection_string: 'http://localhost:8000'

Anthropic with Pinecone

hakam_ai_log_inspector:
    ai_platform:
        provider: 'anthropic'
        api_key: '%env(ANTHROPIC_API_KEY)%'
        model:
            name: 'claude-3-sonnet'
            options:
                temperature: 0.5
    vector_store:
        provider: 'pinecone'
        connection_string: '%env(PINECONE_API_KEY)%'

Local Ollama Setup

hakam_ai_log_inspector:
    ai_platform:
        provider: 'ollama'
        api_key: ''  # Not required for local Ollama
        model:
            name: 'llama2'
            options:
                temperature: 0.1
    vector_store:
        provider: 'chroma'
        connection_string: 'http://localhost:8000'

Security & Privacy

Data Sanitization: All logs are automatically sanitized to remove sensitive information
Local Processing: Can run entirely offline with local models (Ollama)
Configurable Privacy: Choose your AI provider and data handling preferences
No Data Persistence: Logs are processed in memory by default

Contributing

Fork the repository
Create a feature branch
Make your changes
Add tests for new functionality
Ensure all tests pass
Submit a pull request

License

This bundle is released under the MIT License. See the LICENSE file for details.

Author

Ramy Hakam - pencilsoft1@gmail.com

Support

For issues and feature requests, please use the GitHub issue tracker.

hakam / ai-log-inspector-bundle

Maintainers

Details