What problem does it solve? Creating publication-quality scientific figures for papers, presentations, and reports requires coordinating layout, color semantics, data integrity, and quality checks, which is error-prone when done manually. ## Core Features & Use Cases - Multi-mode figure generation: Produces scientific data charts (bar, scatter, ROC, confusion matrix, heatmap), architecture/flow diagrams, Excalidraw hand-drawn flowcharts, Pillow promotional covers, paper-to-PPT conversion, and PDF reverse engineering. - Automated QA pipeline: Runs six geometric checks (text overflow, arrow endpoints, box overlap, boundary, color-blind safety) via scripts like figure-qa-check.py, blocking output until checks pass. - Data integrity enforcement: Requires every figure to have a runnable generation script that reads from JSON/CSV data sources rather than hardcoded values, with audit tooling to detect violations. - Use Case: Given experiment results in a JSON file, generate a multi-panel ROC comparison figure exported as SVG, PDF, and 300-DPI PNG that passes all QA checks for journal submission. ## Quick Start Ask the agent to generate a publication-ready ROC curve figure from your experiment results JSON file with SVG, PDF, and PNG outputs.
