PHPackages cynchro/modux-ia - PHPackages - PHPackages [Skip to content](#main-content)[PHPackages](/)[Directory](/)[Categories](/categories)[Trending](/trending)[Leaderboard](/leaderboard)[Changelog](/changelog)[Analyze](/analyze)[Collections](/collections)[Log in](/login)[Sign up](/register) 1. [Directory](/) 2. / 3. [Utility & Helpers](/categories/utility) 4. / 5. cynchro/modux-ia ActiveLibrary[Utility & Helpers](/categories/utility) cynchro/modux-ia ================ SDK para integrar IA (LLM + RAG) directamente en aplicaciones PHP — sin Python, sin Node.js. v0.0.1(2mo ago)00MITPHPPHP >=8.1CI passing Since May 11Pushed 2mo agoCompare [ Source](https://github.com/cynchro/modux-ia)[ Packagist](https://packagist.org/packages/cynchro/modux-ia)[ Docs](https://github.com/cynchro/modux-ia)[ RSS](/packages/cynchro-modux-ia/feed)WikiDiscussions main Synced 1w ago READMEChangelog (1)Dependencies (3)Versions (2)Used By (0) cynchro/modux-ia ================ [](#cynchromodux-ia) SDK para integrar IA (LLM + RAG) directamente en aplicaciones PHP — sin Python, sin Node.js, sin microservicios. Funciona con **Modux**, Laravel, Symfony, Slim, Mezzio, CodeIgniter y PHP puro. --- Requisitos ---------- [](#requisitos) - PHP 8.1 o superior - Extensiones: `curl`, `pdo`, `json` - Composer 2 --- Instalación ----------- [](#instalación) ``` composer require cynchro/modux-ia ``` Eso es todo. Sin configuración adicional si usás variables de entorno. --- Integración por framework ------------------------- [](#integración-por-framework) ### Modux [](#modux) El módulo se auto-registra al boot. Solo agregá las variables de entorno y listo. **1. Instalar** ``` composer require cynchro/modux-ia ``` **2. Variables de entorno (`.env`)** ``` AI_DRIVER=cloud AI_CLOUD_PROVIDER=openai AI_CLOUD_API_KEY=sk-... AI_CLOUD_MODEL=gpt-4o-mini RAG_STORE=sqlite RAG_SQLITE_PATH=storage/ai/vectors.db ``` **3. Usar en cualquier servicio o controller** El container de Modux inyecta `DriverFactory` y `RAGEngine` automáticamente: ``` use PhpAI\DriverFactory; use PhpAI\RAG\RAGEngine; class SoporteService { public function __construct( private readonly DriverFactory $ai, private readonly RAGEngine $rag, ) {} public function responder(string $pregunta): string { return $this->rag->ask($pregunta); } } ``` Los endpoints listos para usar son: ``` POST /ia/chat — chat con el LLM POST /ia/ask — pregunta con contexto RAG POST /ia/ingest — carga un documento al vector store POST /ia/retrieve — búsqueda semántica pura ``` --- ### Laravel [](#laravel) ``` composer require cynchro/modux-ia ``` El ServiceProvider se autoregistra. Publicá la config si querés personalizarla: ``` php artisan vendor:publish --tag=modux-ia ``` ``` use PhpAI\DriverFactory; use PhpAI\RAG\RAGEngine; class SupportController extends Controller { public function __construct( private readonly DriverFactory $ai, private readonly RAGEngine $rag, ) {} public function ask(Request $request): JsonResponse { $answer = $this->rag->ask($request->input('question')); return response()->json(['answer' => $answer]); } } ``` --- ### Slim 4 / Mezzio / cualquier PSR-11 [](#slim-4--mezzio--cualquier-psr-11) ``` composer require cynchro/modux-ia cynchro/modux-ia-psr11 ``` ``` use DI\ContainerBuilder; use PhpAI\Psr11\AIDefinitions; $builder = new ContainerBuilder(); $builder->addDefinitions(AIDefinitions::get()); $container = $builder->build(); ``` --- ### PHP puro [](#php-puro) ``` require 'vendor/autoload.php'; use PhpAI\Bootstrap; // Zero config — lee todo desde variables de entorno $ai = Bootstrap::fromEnv(); // O configuración explícita $ai = Bootstrap::create([ 'driver' => 'cloud', 'drivers' => [ 'cloud' => [ 'provider' => 'openai', 'api_key' => 'sk-...', 'model' => 'gpt-4o-mini', ], ], 'rag' => ['sqlite_path' => __DIR__ . '/storage/vectors.db'], ]); ``` --- Variables de entorno -------------------- [](#variables-de-entorno) ``` # Driver: local | cloud | cluster AI_DRIVER=cloud # Cloud (OpenAI, Groq, Together, Ollama...) AI_CLOUD_PROVIDER=openai AI_CLOUD_API_KEY=sk-... AI_CLOUD_MODEL=gpt-4o-mini AI_CLOUD_EMBEDDING_MODEL=text-embedding-3-small # Local (llama.cpp) LLAMA_SERVER_URL=http://127.0.0.1:8080 LLAMA_EMBED_URL=http://127.0.0.1:8081 LLAMA_MODEL=llama3 LLAMA_TIMEOUT=120 # Cluster (múltiples nodos llama.cpp) AI_CLUSTER_NODES=http://node1:8080,http://node2:8080 # RAG — store: sqlite | qdrant RAG_STORE=sqlite RAG_SQLITE_PATH=storage/ai/vectors.db RAG_CHUNK_SIZE=512 RAG_CHUNK_OVERLAP=64 RAG_TOP_K=4 QDRANT_URL=http://127.0.0.1:6333 ``` Cambiar de driver es solo cambiar `AI_DRIVER` — el código de la aplicación no cambia. --- Drivers disponibles ------------------- [](#drivers-disponibles) DriverCuándo usarlo`cloud`OpenAI, Groq, Together, Ollama o cualquier API OpenAI-compatible`local`llama.cpp corriendo localmente — sin costo, sin internet`cluster`Múltiples nodos llama.cpp con round-robin### Providers cloud soportados [](#providers-cloud-soportados) Provider`AI_CLOUD_PROVIDER`OpenAI`openai`Groq`groq`Together AI`together`Ollama`ollama`Cualquier OpenAI-compatibleURL directa como provider--- Uso --- [](#uso) ### Chat [](#chat) ``` use PhpAI\Contracts\CompletionRequest; $response = $factory->llm()->complete(new CompletionRequest( messages: [ ['role' => 'system', 'content' => 'Sos un experto en PHP.'], ['role' => 'user', 'content' => '¿Qué es un trait?'], ], temperature: 0.7, maxTokens: 512, )); echo $response->content; ``` ### Streaming (SSE) [](#streaming-sse) ``` header('Content-Type: text/event-stream'); header('Cache-Control: no-cache'); foreach ($factory->llm()->stream($request) as $token) { echo "data: " . json_encode(['token' => $token]) . "\n\n"; ob_flush(); flush(); } echo "data: [DONE]\n\n"; ``` ### RAG — ingestar documentos [](#rag--ingestar-documentos) ``` $rag = $factory->rag(); $rag->ingest( document: file_get_contents('manual.txt'), id: 'manual-v1', metadata: ['tipo' => 'manual'], ); ``` ### RAG — hacer preguntas [](#rag--hacer-preguntas) ``` // Respuesta directa $answer = $rag->ask("¿Cuál es el proceso de devolución?"); // Con filtro de metadata $answer = $rag->ask( question: "¿Cómo configuro el sistema?", filter: ['tipo' => 'manual'], ); // Con system prompt personalizado $answer = $rag->ask( question: "¿Qué garantías tiene el producto?", systemPrompt: "Respondé de forma empática en no más de 3 oraciones.", ); ``` ### Búsqueda semántica pura [](#búsqueda-semántica-pura) ``` $chunks = $rag->retrieve("política de envíos internacionales"); foreach ($chunks as $chunk) { echo "[{$chunk['score']}] {$chunk['metadata']['content']}\n"; } ``` ### Embeddings [](#embeddings) ``` $vector = $factory->embeddings()->embed("texto de ejemplo"); $vectors = $factory->embeddings()->embedBatch(["texto 1", "texto 2"]); ``` --- Vector store ------------ [](#vector-store) StoreCuándo usarlo`sqlite`Desarrollo y producción pequeña (~50k chunks, zero infraestructura)`qdrant`Producción con millones de vectores``` # Levantar Qdrant con Docker docker run -d --name qdrant -p 6333:6333 qdrant/qdrant ``` --- Driver local con llama.cpp -------------------------- [](#driver-local-con-llamacpp) ``` # Instalar llama.cpp git clone https://github.com/ggml-org/llama.cpp cd llama.cpp && cmake -B build && cmake --build build --config Release # Descargar modelos wget https://huggingface.co/bartowski/Llama-3.2-3B-Instruct-GGUF/resolve/main/Llama-3.2-3B-Instruct-Q4_K_M.gguf -O models/llama3.gguf wget https://huggingface.co/nomic-ai/nomic-embed-text-v1.5-GGUF/resolve/main/nomic-embed-text-v1.5.Q4_K_M.gguf -O models/nomic-embed.gguf # Levantar servidores ./build/bin/llama-server --model models/llama3.gguf --host 127.0.0.1 --port 8080 --ctx-size 4096 ./build/bin/llama-server --model models/nomic-embed.gguf --host 127.0.0.1 --port 8081 --embedding --pooling mean ``` --- Estructura del SDK ------------------ [](#estructura-del-sdk) ``` src/ ├── Bootstrap.php Punto de entrada ├── Config.php Configuración con dot-notation y lectura de ENV ├── DriverFactory.php Resuelve y cachea drivers por nombre ├── Contracts/ │ ├── LLMDriver.php Interface de inferencia │ ├── EmbeddingDriver.php Interface de embeddings │ ├── VectorStoreContract.php Interface de almacenamiento vectorial │ ├── CompletionRequest.php DTO de entrada al LLM │ └── CompletionResponse.php DTO de salida del LLM ├── Http/ │ └── Client.php HTTP client con soporte SSE streaming ├── Drivers/ │ ├── Cloud/CloudDriver.php APIs OpenAI-compatible (OpenAI, Groq, etc.) │ ├── Local/LocalDriver.php llama.cpp HTTP server │ └── Cluster/ClusterDriver.php Round-robin sobre múltiples nodos ├── RAG/ │ ├── TextChunker.php Divide texto en chunks con overlap │ └── RAGEngine.php Orquesta ingest + retrieval + generación └── Storage/ ├── SqliteVectorStore.php SQLite + cosine similarity en PHP └── QdrantVectorStore.php Qdrant via HTTP API packages/ ├── psr11/ cynchro/modux-ia-psr11 └── laravel/ cynchro/modux-ia-laravel ``` --- Licencia -------- [](#licencia) MIT ### Health Score 32 — LowBetter than 69% of packages Maintenance86 Actively maintained with recent releases Popularity0 Limited adoption so far Community6 Small or concentrated contributor base Maturity32 Early-stage or recently created project Bus Factor1 Top contributor holds 100% of commits — single point of failure How is this calculated?**Maintenance (25%)** — Last commit recency, latest release date, and issue-to-star ratio. Uses a 2-year decay window. **Popularity (30%)** — Total and monthly downloads, GitHub stars, and forks. Logarithmic scaling prevents top-heavy scores. **Community (15%)** — Contributors, dependents, forks, watchers, and maintainers. Measures real ecosystem engagement. **Maturity (30%)** — Project age, version count, PHP version support, and release stability. ### Release Activity Cadence Unknown Total 1 Last Release 74d ago ### Community Maintainers ![](https://avatars.githubusercontent.com/u/22827205?v=4)[cynchro](/maintainers/cynchro)[@cynchro](https://github.com/cynchro) --- Top Contributors [![cynchro](https://avatars.githubusercontent.com/u/22827205?v=4)](https://github.com/cynchro "cynchro (4 commits)") --- Tags phpaillmllamaembeddingsrag ### Code Quality TestsPHPUnit Static AnalysisPHPStan Code StylePHP\_CodeSniffer Type Coverage Yes ### Embed Badge ![Health badge](/badges/cynchro-modux-ia/health.svg) ``` [![Health](https://phpackages.com/badges/cynchro-modux-ia/health.svg)](https://phpackages.com/packages/cynchro-modux-ia) ``` ### Alternatives [ardagnsrn/ollama-php This is a PHP library for Ollama. Ollama is an open-source project that serves as a powerful and user-friendly platform for running LLMs on your local machine. It acts as a bridge between the complexities of LLM technology and the desire for an accessible and customizable AI experience. 21284.7k](/packages/ardagnsrn-ollama-php) PHPackages © 2026 [Directory](/)[Categories](/categories)[Trending](/trending)[Changelog](/changelog)[Analyze](/analyze)