Archive Reprocessing System

Auto-activates when: Working with tools/scripts/lib/, reprocessing scripts, transform versions, or discussing archive transformations.

Quick Reference

When to use:

• New transformations on existing archives (flattening, normalization)
• After version bumps (v1.0 → v1.1)
• Vocabulary/model updates
• Regenerating derived outputs

When NOT to use:

• Adding new videos → ingest_youtube.py
• Rebuilding Qdrant cache → reingest_from_archive.py

Scripts:

• reprocess_qdrant_metadata.py - Flattening + weights (fast, no LLM)
• reprocess_normalized_tags.py - Tag normalization (slow, LLM calls)