內容簡介
RiTriever 外掛為 WordPress 搜尋功能增添 RAG 風格的向量檢索，將已發佈內容嵌入並儲存在本地向量資料表中，並與標準 WordPress 搜尋結果結合，提升搜尋精準度。
【主要功能】
• RAG 風格搜尋與標準 WordPress 搜尋結合
• 本地向量儲存於 WordPress 資料庫
• 支援 OpenAI 和 Azure OpenAI 外部嵌入 API
• 支援 Ollama、LM Studio 等本地嵌入
• 支援 WordPress 支援的語言選擇
• 管理員診斷資料庫支援及索引進度
外掛標籤
開發者團隊
原文外掛簡介
RiTriever adds RAG-style vector retrieval to WordPress search. Published content is embedded, stored in a local vector table, and blended with standard WordPress search results.
The plugin is intended for WordPress sites that can use native database vector support, primarily MariaDB 11.7 or later. It can index posts, pages, and configured public post types, then use those vectors during front-end search.
Main features:
RAG-style search blending with standard WordPress search.
Native vector storage in the WordPress database.
OpenAI and Azure OpenAI support for external embedding APIs.
Local or self-hosted embedding support through Ollama, LM Studio, Infinity, TEI, or Custom HTTP endpoints.
RAG target language selection from WordPress-supported locales.
Admin diagnostics for database support, embedding providers, indexing progress, and live vector queries.
External API use is optional. WordPress 7.0 AI Client does not provide embedding generation, so RiTriever uses direct embedding APIs when external embeddings are configured. Post content and configured custom field values are sent to the selected provider to create embeddings.
External services
RiTriever connects to external services only when you select an external embedding provider. It sends post titles, post excerpts, post content chunks, selected taxonomy terms, selected custom field values, and short admin test strings when indexing or testing embeddings. The vectors returned by the provider are stored in your WordPress database.
WordPress 7.0 includes the WordPress AI Client for supported AI capabilities, but the current AI Client announcement and implementation do not include an embeddings API. Because RiTriever requires text embeddings for vector search, it calls embedding providers directly instead of using WordPress AI Client. Reference: https://make.wordpress.org/core/2026/03/24/introducing-the-ai-client-in-wordpress-7-0/
OpenAI sends embedding requests to OpenAI for text-embedding-3-small or text-embedding-3-large. This service is provided by OpenAI: Terms of use https://openai.com/policies/terms-of-use/ and Privacy policy https://openai.com/policies/privacy-policy/.
Azure OpenAI sends embedding requests to your configured Azure OpenAI endpoint. This service is provided by Microsoft Azure: Terms of use https://www.microsoft.com/licensing/terms/productoffering/MicrosoftAzure/MCA and Privacy statement https://privacy.microsoft.com/privacystatement.
Local or self-hosted endpoints, including Ollama, LM Studio, Infinity, TEI, and Custom HTTP, receive the same embedding inputs when selected. Review the terms, privacy policy, and network location for the endpoint you operate or configure.
