--- name: figure-table-planner description: Plan the smallest set of diagnostic, comparison, paper, and appendix figures or tables needed to support verified mathematical-modeling decisions and claims. --- # Purpose Make every visual evidence-bearing. Prefer fewer useful visuals over a decorative inventory. # Inputs - method card and decision ledger; - run summaries and final result analysis; - robustness evidence; - solution package and frozen numbers in submission mode; - existing figures/tables. # Figure Types - Type 1 diagnostic: internal debugging; never in the paper. - Type 2 comparison: main vs usable baseline or a genuinely tested alternative; optional in paper. - Type 3 paper: directly supports a main claim; required only when the claim benefits materially from a visual. - Type 4 appendix: supplementary evidence referenced from the main text. # Workflow 1. List verified claims that need visual or exact tabular support. 2. Reuse an existing artifact when it already communicates the claim. 3. For each proposed visual record: - ID and Qx; - type; - source artifact and frozen claim IDs when applicable; - one core claim; - chart/table form; - target section; - status and render needs. 4. Ask the human to confirm judgment-bearing Type 3 claims through one compact choice card when they are not already in the decision ledger. 5. Save `methods/Qx/qx_figure_table_plan.md` only when durable planning is needed. In lean exploration, a compact in-conversation plan is sufficient. # Planning Heuristics - Use tables for exact values, parameters, and small comparisons. - Use plots for trends, distributions, sensitivity, or many-item comparisons. - Use diagrams for mechanisms, dependencies, and workflows. - A main-vs-baseline figure needs compatible metrics and the same evaluation setup. - Do not create a multi-method comparison merely to imply breadth. # Rules - Type 1 never enters the paper. - Type 3 uses final validated sources and a human-confirmed core claim. - Do not use unresolved exploratory figures as paper evidence. - Do not fabricate data, captions, or claims. - Do not fill plans with placeholder sentinels; pause for one human choice instead. - Every visual must have a source and purpose. # Verification - Each planned visual supports a verified claim. - Types, sources, sections, and statuses are explicit. - Type 3 claims trace to human decisions and frozen evidence. - No unnecessary or decorative visual remains.