# Sigmie

> A modern, developer-friendly Elasticsearch and OpenSearch library for PHP and Laravel. Fluent search, semantic and hybrid retrieval, AI-ready, no boilerplate.

## Getting Started
- [Introduction](https://sigmie.com/docs/v2/introduction): Sigmie is a Laravel-inspired PHP library for Elasticsearch and OpenSearch with fluent search, typo tolerance, faceting, and semantic vectors.
- [Installation](https://sigmie.com/docs/v2/installation): Install Sigmie via Composer and connect to a local, cloud, or multi-node Elasticsearch or OpenSearch cluster with basic auth, API keys, or SSL.
- [Quick Start](https://sigmie.com/docs/v2/quick-start): Build your first Sigmie search in five minutes — define a schema, index documents, and search with typo tolerance, filters, sorting, and weighting.

## Core Concepts
- [Core Concepts](https://sigmie.com/docs/v2/core-concepts): The four pieces of Sigmie that everything else composes from — the client, indices, documents, and properties — and how they fit together.
- [Indices](https://sigmie.com/docs/v2/index): Create, configure, and update Elasticsearch indices with Sigmie — schema, analysis, sharding, replication, zero-downtime updates, and deletion.
- [Documents](https://sigmie.com/docs/v2/document): Index documents into Elasticsearch with Sigmie collections — add, merge, update, iterate with each() and lazy(), and stream large result sets.
- [Mappings & Properties](https://sigmie.com/docs/v2/mappings): Define your Elasticsearch schema with Sigmie's high-level and native field types — title, name, category, price, semantic vectors, nested, and more.
- [Search](https://sigmie.com/docs/v2/search): Build user-facing Elasticsearch searches with Sigmie — typo tolerance, faceted navigation, highlighting, semantic search, and filter-parser syntax.
- [Advanced Queries](https://sigmie.com/docs/v2/query): Build raw Elasticsearch queries with Sigmie's low-level newQuery() API — boolean must/should/filter, term, match, range, and script scoring.

## Features
- [Semantic Search](https://sigmie.com/docs/v2/semantic-search): Build semantic vector search with Sigmie — semantic fields, accuracy levels, similarity metrics, and embeddings from OpenAI, Cohere, or Infinity.
- [Aggregations](https://sigmie.com/docs/v2/aggregations): Run Elasticsearch metric and bucket aggregations with Sigmie — sum, avg, stats, terms, histogram, date histogram, and pipeline aggregations.
- [Magic Tags](https://sigmie.com/docs/v2/magic-tags): LLM-generated taxonomy tags for Elasticsearch documents — classify with embeddings, dedup with vector similarity, share a tag registry across indices.
- [Facets](https://sigmie.com/docs/v2/facets): Build faceted navigation with Sigmie — term, price histogram, and stats facets, with disjunctive and conjunctive logic for e-commerce sidebars.
- [Recommendations](https://sigmie.com/docs/v2/recommendations): Build similar-item recommendations with Sigmie's newRecommend() API — multi-seed retrieval, field weighting, RRF fusion, and MMR diversification.

## Text Analysis
- [Text Analysis](https://sigmie.com/docs/v2/analysis): How Elasticsearch transforms text at index and query time — the char filter, tokenizer, and token filter pipeline explained with worked examples.
- [Tokenizers](https://sigmie.com/docs/v2/tokenizers): Split text into tokens with Elasticsearch tokenizers in Sigmie — word boundaries, whitespace, pattern, simple pattern, path hierarchy, non-letter.
- [Token Filters](https://sigmie.com/docs/v2/token-filters): Transform Elasticsearch tokens with Sigmie — stemming, synonyms, stopwords, lowercase, ASCII folding, decimal digit, truncate, and keyword guards.
- [Character Filters](https://sigmie.com/docs/v2/char-filters): Pre-process text before Elasticsearch tokenization with Sigmie's character filters — HTML strip, character mapping, and regex pattern replace.
- [Languages](https://sigmie.com/docs/v2/language): Language-specific Elasticsearch analyzers in Sigmie — English, German, and Greek stemmers, stopwords, lowercase, and umlaut normalizers.

## Utilities
- [Filter Parser](https://sigmie.com/docs/v2/filter-parser): Filter Elasticsearch results with human-readable expressions — ranges, wildcards, AND/OR/NOT, nested fields, geo proximity. Full syntax reference.
- [Sort Parser](https://sigmie.com/docs/v2/sort-parser): Sort Elasticsearch results with space-separated expressions like "_score rating:desc name:asc". Auto-routes text fields to .keyword sub-fields.

## Configuration
- [Connection Setup](https://sigmie.com/docs/v2/connection): Configure Elasticsearch and OpenSearch connections in Sigmie — basic auth, API keys, bearer tokens, SSL certificates, and multi-node clusters.
- [Docker Setup](https://sigmie.com/docs/v2/docker): Run Elasticsearch, OpenSearch, and local AI services — embeddings, reranker, CLIP — via Docker Compose for semantic search development with Sigmie.
- [OpenSearch](https://sigmie.com/docs/v2/opensearch): Use OpenSearch 2.x or 3.x with Sigmie — connect to local clusters or AWS OpenSearch Service. Every Sigmie feature works identically to Elasticsearch.

## Integrations
- [Laravel Scout](https://sigmie.com/docs/v2/laravel-scout): Use Sigmie as a Laravel Scout driver — index Eloquent models in Elasticsearch with typo tolerance, highlighting, facets, and custom analyzers.
- [Laravel AI SDK](https://sigmie.com/docs/v2/laravel-ai): Expose Sigmie indices as Laravel AI agent tools — auto-generated descriptions, base filters for multi-tenancy, and the full Sigmie filter syntax.
- [MCP Server](https://sigmie.com/docs/v2/mcp): Connect AI agents like Claude Code and Cursor to the Sigmie documentation via the Model Context Protocol — semantic search, page read, page list.

## Reference
- [Packages](https://sigmie.com/docs/v2/packages): The modular Sigmie Composer packages — base, http, index, document, mappings, query, search, parse, testing, plus English, German, and Greek.

## Advanced Features
- [Retrieval and Agents](https://sigmie.com/docs/v2/rag): Combine Sigmie retrieval and reranking with text generation in RAG workflows — search, rerank, then call your LLM client of choice from app code.
- [Extending Sigmie](https://sigmie.com/docs/v2/extending): Extend Sigmie with custom packages — register field-type macros on NewProperties and add CollectionHook implementations that fire during indexing.

## Blog
- [A different approach for Elasticsearch](https://sigmie.com/blog/a-different-approach): Why Sigmie takes a different approach to Elasticsearch in PHP — fluent API, no boilerplate, focus on relevance instead of low-level mappings.
- [Why are Search services expensive](https://sigmie.com/blog/why-are-search-services-expensive): Why hosted search-as-a-service is so expensive — what you actually pay for under the hood and when running your own Elasticsearch makes sense.
- [High level Elasticsearch properties](https://sigmie.com/blog/high-level-properties): A walkthrough of Sigmie's predefined Elasticsearch property types and the analysis decisions baked in to get the best search relevance out of the box.
- [Elasticsearch shards rules](https://sigmie.com/blog/calculating-index-shards): Practical rules of thumb for calculating Elasticsearch primary and replica shards — sizing, distribution, and the math behind cluster capacity.

## Optional
- [MCP Server](https://sigmie.com/mcp): Streamable HTTP MCP endpoint for AI agents — exposes search_docs, read_doc, and list_docs tools
- [Sitemap](https://sigmie.com/sitemap.xml): Machine-readable list of indexable URLs