What problem does it solve? Research papers and theses need figures that support specific claims with traceable data, but producing consistent, publication-ready plots and architecture diagrams manually is slow and error-prone. This Skill ties every figure to a claim, source data, and an audit record. ## Core Features & Use Cases - Nature-Style Plot Templates: Render grouped bars, line trends, heatmaps, scatter, radar, distribution, forest, ablation, threshold, confusion-matrix, and multi-panel figures from JSON specs via scripts/nature_plot_templates.py, exporting SVG/PDF/PNG with QA reports. - Network Architecture Diagrams: Generate CNN, ResNet, U-Net, and Transformer architecture figures from .network.json structure specs using scripts/render_network_architecture.py, with presets (thesis-clean, nature-minimal, ppt-template-rich) and PPTX export modes. - Figure Audit & Caption Safety: Apply the figure audit standard to verify claim clarity, evidence traceability, caption safety, typography, and export quality before marking figures as final. - Use Case: A graduate student needs a ResNet-18 architecture figure and an ablation bar chart for their thesis. They write a .network.json spec and a figure spec JSON, run the renderers, and receive SVG/PDF/PNG outputs plus QA reports linked to their figure plan. ## Quick Start Use the research-paper-figures skill to plan and generate a Nature-style ablation bar chart and a ResNet-18 architecture diagram for my thesis results section.
