paper-illustration
Generate publication-quality AI illustrations for academic papers using Gemini image generation. Creates architecture diagrams, method illustrations with Claude-supervised iterative refinement loop. Use when user says \"生成图表\", \"画架构图\", \"AI绘图\", \"paper illustration\", \"generate diagram\", or needs visual figures for papers.
Profile →npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill paper-illustration --agent claude-code
Same command for any agent — swap --agent for codex, cursor, copilot.
Weekly change comes from our own snapshots, not the repository page — it measures attention, not adoption.
What it does
Generates publication-quality AI illustrations for academic papers. It uses a multi-stage workflow where Claude plans the figure, Gemini optimizes the layout, and Gemini verifies style compliance before final rendering. It targets architecture diagrams and method illustrations and emphasizes strict visual standards and iterative refinement. The process includes an iterative loop with a final strict visual review scored by Claude, requiring a score of 9 or higher to accept, otherwise generating targeted improvement feedback.
How it works
- User request is parsed by Claude (Planner) to understand the desired figure and components.
- Gemini (Layout) receives an initial detailed prompt and outputs an optimized layout description, including exact component positions, spacing, grouping, arrow routing, and visual hierarchy.
- Gemini (Style) reviews the optimized layout against CVPR/NeurIPS style guidelines, enhancing the specification to ensure color usage, fonts, arrow requirements, and overall visual appeal meet top-tier conference standards. Output includes explicit style instructions for rendering.
- Step 4 renders the final image according to the validated specification (Paperbanana / Nano Banana Pro for rendering). A strict visual review and score (1-10) by Claude decides acceptance.
- If Score ≥ 9, the figure is accepted and output; otherwise, the process loops back to Step 2 for improvement (up to MAX_ITERATIONS). Output directory is figures/ai_generated/.
When to use it
Use when a user requests publication-quality figures for papers (e.g., “生成图表”, “画架构图”, “AI绘图”, “paper illustration”, “generate diagram”) or needs visual figures for papers requiring CVPR/NeurIPS-style standards.
What it can touch
- Uses the Gemini models gemini-3-pro-preview and Claude for planning and reviewing. Tool interactions include Bash scripts, curl to Gemini endpoints, and Python JSON payload generation to drive the requests. Output files are written to figures/ai_generated/ (layout_description.txt, final render, etc.).
Caveats
- The workflow relies on external Gemini and Claude services; API keys are required (GEMINI_API_KEY). The process is iterative with a maximum of MAX_ITERATIONS refinement rounds and a TARGET_SCORE of 9 for acceptance. The approach emphasizes CVPR/NeurIPS style constraints and may require manual adjustments if automated prompts do not capture all user specifics.
# Paper Illustration: Multi-Stage Claude-Supervised Figure Generation Generate publication-quality illustrations using a **multi-stage workflow** with **Claude as the STRICT supervisor/reviewer**. ## Core Design Philosophy ``` ┌──────────────────────────────────────────────────────────────────────────┐ │ MULTI-STAGE ITERATIVE WORKFLOW │ ├────────────────────────────────────────────────────────────
What does the paper-illustration skill do?
Generate publication-quality AI illustrations for academic papers using Gemini image generation. Creates architecture diagrams, method illustrations with Claude-supervised iterative refinement loop. Use when user says \"生成图表\", \"画架构图\", \"AI绘图\", \"paper illustration\", \"generate diagram\", or needs visual figures for papers.
How do I install it?
Run `npx skills add brycewang-stanford/Auto-Empirical-Research-Skills --skill paper-illustration --agent claude-code` — it drops the skill into your project so the agent can pick it up. Swap the --agent value for codex, cursor or copilot if you use one of those.
Where does this skill come from?
From brycewang-stanford/Auto-Empirical-Research-Skills, a repository with 3,244 stars. We read it straight from the repository tree rather than a submitted listing, so what you see here is what is actually published.
Is a popular skill a good skill?
Not necessarily. Stars measure attention, not adoption — a repository can trend for a week and be abandoned. That is why we show the weekly change from our own snapshots next to the total, instead of a single flattering number.