table-lab v2

Iteratively tune Camelot table extraction parameters for PDFs. Supports single-PDF tuning with convergence loop, batch corpus tuning with watchdog monitoring, Federated Taxonomy bridge tagging, and corpus-report integration.

Agent Workflow

When invoked, follow this sequence:

1. Memory Recall (MANDATORY FIRST STEP)

/memory recall "table extraction settings for <category> <preset>"

If memory returns a known-good configuration, use it as the starting baseline.

2. Single PDF Tuning