
Paper Illustration
Generates academic paper figures with Gemini and iteratively refines them under Claude review
- Stars
- 18,482
- Licence
- MIT
- Files
- 1 file
What it does
When installed, the agent can create publication-style diagrams and method illustrations for research papers. It drafts a layout plan, has Gemini image models render the figure, then has Claude score the result strictly (1-10) and loops with feedback until the score reaches 9 or 5 rounds pass. An optional style-reference flag can nudge structural conventions such as caption length or panel counts toward a reference paper.
Great for
- Researchers who need architecture or method diagrams for a paper draft
- Students preparing figures for a thesis or course report and who have a Gemini API key
- Developers building an automated paper-writing pipeline who want a figure step
Skip it if
- Figures that must be scientifically exact (generated diagrams can contain wrong labels, arrows or content; the skill's own review step is the only check and must be verified by a human)
- Users without a Gemini API key or who cannot send unpublished method descriptions to Google's image service (the workflow sends prompts and generated content to external models)
Try asking
- “Generate an architecture diagram for the method described in method.md”
- “生成图表 showing the encoder-decoder pipeline of my model”
- “Make a method illustration for section 3 with a CVPR-style colour palette”
- “Create a figure for my paper and save it to figures/ai_generated/”
Install in 3 steps
- 1
Download the skill
The ZIP contains the 1 original files, the MIT licence and an attribution note.
paper-illustration.zip - 2
Add it to your agent
Claude Code (all projects)
mkdir -p ~/.claude/skills && unzip paper-illustration.zip -d ~/.claude/skills/Claude Code / Agent SDK (this project only)
mkdir -p .claude/skills && unzip paper-illustration.zip -d .claude/skills/Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.
- 3
Ask for the task
No special command needed — the agent reads the skill's description and loads it when your request matches. Try one of the prompts above.
Get better results
- Give a clear method description or file path so the first layout plan is accurate, then check every arrow and label in the final image yourself
- Keep the GEMINI_API_KEY in your environment rather than in files, and review what text you send to external services
- Use the style-ref option only when you have a reference paper you are allowed to use, and expect only structural guidance from it
What's inside
- SKILL.md
Redistributed unmodified from wanshuiyin/Auto-claude-code-research-in-sleep under the MIT licence. The assessment was drafted with AI from the skill's files on 2026-10-29. Skills can instruct your agent to run code — read SKILL.md and any scripts before installing.