Duplicate documents: Pipeline implementation and metadata

Defines duplicate detection and retrieval grouping while preserving each source document and version independently, preventing shared storage identities and invisible cross-project deletion dependencies.

How to write a block

Publié Sep 12, 2026