Generate publication-ready method, architecture, and system diagrams as SVG from a research proposal using the HappyFigure pipeline. Use this skill whenever the user wants to create an architecture diagram, method figure, system overview, pipeline visualization, flowchart, block diagram, or any structural illustration for a paper — including phrases like "draw my model architecture", "I need a method figure", "create a system diagram", "make a pipeline figure", "generate an SVG of my approach", "illustrate how the system works", or "I need a figure showing the components". Two modes: full pipeline (high-fidelity, uses image generation + SAM3 segmentation + SVG conversion + team review, requires GPU microservices) and sketch (lightweight, agent writes SVG directly from text, no services needed).
Generate publication-ready statistical plots and charts from experiment data using the HappyFigure multi-agent pipeline. Use this skill whenever the user wants to create figures, charts, plots, or statistical visualizations from experimental results — including phrases like "plot my results", "make a bar chart", "generate figures for my paper", "visualize the data", "I need results figures", "show performance comparison", or any request to turn experiment data into publication-quality graphics. Also triggers for requests about rerunning or improving previously generated figures.
Given a project directory, produce a detailed figure description document (paper_summary.md) with precise per-panel plot-type specifications for every figure a paper needs. This is the planning step before figure generation — it scans data files, reads paper text, and outputs a structured markdown spec where each panel has an explicit visual form (e.g., "paired-line plot", "heatmap matrix", "violin distribution"). Use this skill when the user says "plan my figures", "describe what figures I need", "create a figure spec", "write a paper_summary", "what plots should I make", or points you at a project directory and wants to know what figures to generate before actually generating them. Also use this skill when preparing inputs for the figures skill, or when the user has data and a draft but hasn't decided on figure layouts yet.