Index fields

Title, filename, taxonomy, excerpt, Article content and extracted text receive configurable weights. Relevance presets, synonym groups and escaped contextual highlights shape results without exposing private records.

Extraction

Built-in bounded extraction supports DOCX, PPTX, XLSX and OpenDocument packages. PDF/OCR remains a provider gate and is not advertised as shipped.

Operations and privacy

Use Search Index diagnostics for queue, duration, provider, failure and stale-state checks. Zero-result analytics are salted aggregates with a configurable retention period; raw private queries are not stored as a marketing dataset.