WordPress Vector Database (WPVDB)

WPVDB stores and queries vector embeddings inside a WordPress site's own database, using MySQL's or MariaDB's native vector column types where available and falling back to JSON storage on older database versions, and extends WP_Query with a vector similarity query. It can automatically chunk and embed site content, optionally summarize chunks with AI, and generate embeddings using either OpenAI's API or Automattic's own API.

It exposes REST API endpoints for triggering embedding generation and running vector search, an admin settings page for configuring API keys and the vector database, block-editor integration, and background processing through Action Scheduler when it is available. The repository includes a Docker-based testing setup that runs parallel WordPress instances against MySQL 8.0.32 and MariaDB 11.7 to test both native vector implementations. It requires WordPress 6.0 and PHP 7.4 or newer.