| name | paper-illustration |
| title | Web-Safe Paper Illustration Designer |
| description | Design paper-ready method diagrams, architecture illustrations, visual abstracts, and image-generation prompts without relying on external image-generation APIs. |
| triggers | ["paper illustration","method diagram","architecture diagram","framework figure","visual abstract","image prompt","架构图","方法图","机制图","示意图","论文插图","AI绘图","画架构图"] |
Web-side execution adapter
- This skill is workflow guidance for the ChatGPT web-side connector.
- Loading this SKILL.md is only the setup step; it does not mean the task is complete.
- After loading, continue to execute the workflow, constraints, and output format below before answering.
- Mentions of local automation, local file operations, local command execution, or external integrations are descriptive only. Use capabilities available in the current ChatGPT session, or ask the user for needed files/links.
- For literature search, current facts, factual verification, source tracing, numeric values, material properties, legal/medical/financial/current information, or any evidence-heavy claim: use available search/browsing tools first and cite verifiable sources. Do not answer such tasks only from memory.
- Preserve the original workflow and scope unless the user explicitly asks for changes.
Web-Safe Paper Illustration Designer
Purpose
Use this skill when the user wants to design a paper illustration, method figure, architecture diagram, visual abstract, pipeline figure, or image-generation prompt for an academic paper.
This web-safe version does not call Gemini, image-generation APIs, local drawing tools, local command execution, local coding-assistant integration, or file-system operations. It produces a clear design specification, layout plan, diagram text, Mermaid/TikZ-style skeletons, and image-generation prompts that the user can copy into external tools.
Operating Rules
- Do not claim that an image has been generated unless the current ChatGPT environment actually generates it.
- Treat the output as a design brief, not a finished figure file.
- Prioritize scientific correctness over visual ornamentation.
- Every box, arrow, icon, and label must correspond to a real component or process.
- Avoid decorative elements that imply unsupported mechanisms.
- Make diagrams self-contained: title, components, flow, legend, and caption.
Workflow
Step 1: Understand the figure goal
Identify:
- paper topic and target venue
- figure type: method overview, architecture, training pipeline, inference pipeline, visual abstract, comparison diagram
- main contribution to communicate
- modules/components
- data flow
- training vs inference distinction
- inputs and outputs
- important equations or losses
- visual constraints: one-column, two-column, slide, poster
Step 2: Build a semantic diagram plan
Represent the method as:
Input → Encoder/Processor → Core method → Output → Supervision/Evaluation
For complex methods, split into zones:
- Data / input zone
- Model architecture zone
- Training objective zone
- Inference/output zone
- Evaluation/result zone
Step 3: Produce a layout specification
Use this template:
## Illustration Specification
**Figure purpose:**
**Canvas:** one-column / two-column / slide / poster
**Reading direction:** left-to-right / top-to-bottom / circular
**Main zones:**
**Key components:**
**Arrows:**
**Labels:**
**Color coding:**
**Legend:**
**What to emphasize:**
**What to avoid:**
Step 4: Provide alternative layouts
Give 2–3 layout options:
- Option A: clean pipeline
- Option B: modular architecture
- Option C: contribution-focused visual abstract
For each option, state:
- best use case
- strengths
- weaknesses
- estimated complexity
- whether it fits one-column or two-column format
Step 5: Generate diagram text
When useful, provide one or more of:
- ASCII block layout
- Mermaid flowchart
- TikZ-style pseudocode
- Figma/draw.io construction instructions
- image-generation prompt
- caption draft
For Mermaid, keep the graph simple and syntactically conservative.
Step 6: Review the design
Check:
- Are all arrows directionally correct?
- Are training-only components separated from inference-time components?
- Are losses and supervision signals visually distinct?
- Are all labels short enough?
- Is the main contribution visually prominent?
- Could a reader understand the method without reading the full paper?
Output Structure
- Figure goal
- Core message
- Recommended layout
- Detailed visual specification
- Alternative layouts
- Mermaid / TikZ / prompt draft
- Caption
- Self-review checklist
Image Prompt Template
When the user wants an image-generation prompt, use:
Create a clean academic paper method diagram on a white background. The diagram shows [method]. Use [layout]. Include the following labeled components: [components]. Show arrows from [flow]. Emphasize [contribution]. Use minimal colors, thin arrows, readable sans-serif labels, and no decorative background. Avoid photorealism, icons unrelated to the method, and unreadable tiny text.
Style Guidance
- Use neutral academic style.
- For CV/ML papers, prefer clean vector-like diagrams.
- For systems papers, show modules and data movement clearly.
- For theory papers, prefer conceptual diagrams with minimal components.
- Do not overuse colors; reserve accent color for the main contribution.
