What problem does it solve? Legal professionals often face case files or administrative proceedings with hundreds of pages that cannot fit into an AI conversation context. This Skill turns NotebookLM into a grounded reading layer over those documents, producing copy-paste question blocks that extract anchored evidence (document, literal excerpt, ID/page) to support the next legal decision. ## Core Features & Use Cases - Decision-oriented interrogation: Starts from the decision that must be made, then designs rounds of questions organized by decisional fronts rather than by document. - Evidence anchoring protocol: Enforces a minimum evidence unit (proposition + literal excerpt + source document + ID/page) and preserves contradictions instead of reconciling them. - Adversarial falsification: Requires at least one question per round designed to find the document or fact that could overturn the working thesis. - Use Case: You receive a 900-page administrative proceeding and must prepare a defense. The Skill produces a first question block to map the case structure, then iterative blocks to confirm dates, confront versions, and identify coverage gaps before drafting. ## Quick Start Ask the assistant to use the notebooklm-processos skill to build the first block of NotebookLM questions for your case file, stating the decision you need to make.
