What problem does it solve? LLM outputs often ship as first drafts with unexamined assumptions and blind spots. This Skill provides a shared refinement checkpoint that pressure-tests any recent output — a plan, draft, section, or decision — using a catalog of proven elicitation methods before the work moves forward. ## Core Features & Use Cases - Method Catalog Served On-Demand: A Python script serves categories, listings, and random draws from a CSV catalog of critique methods (socratic, first principles, pre-mortem, red team, and more) without flooding the context window. - Interactive Refinement Menu: Offers five targeted methods per round, with reshuffle, full-catalog listing, and proceed options; every proposed change requires explicit user approval before being applied. - Customizable Workflow: Team and personal TOML overrides can add custom methods, swap the catalog file, and set persistent preferences that shape every session. - Use Case: After drafting a project plan, invoke this Skill to run a pre-mortem and red-team pass on the plan, review the weaknesses each method reveals, and accept only the improvements you want before handing the refined plan back to the invoking workflow. ## Quick Start Ask the assistant to run advanced elicitation on the plan it just produced and pick a method from the menu it offers.
