Skip to content
Get packs

Paper Illustration via Codex Image2

Draft publication-style academic figures by routing image generation through a local Codex bridge

Source Download skill
Repository
wanshuiyin/Auto-claude-code-research-in-sleep
Stars
18,482
Licence
MIT
Files
2 files

What it does

This skill guides an AI agent through making a scientific paper figure. Claude plans the layout and wording, reviews the draft against academic style rules, and asks a local Codex app-server bridge to render the image using Codex's native image generation. It loops through several review rounds, scoring each draft, and saves the output under figures/ai_generated/ in your project. It is an experimental alternative to the original paper-illustration skill.

Great for

  • Researchers who want a quick draft of a method or pipeline diagram for a paper
  • Claude Code users already running a local Codex MCP bridge who want to compare renderers
  • Students preparing figure drafts that they will then redraw or polish by hand

Skip it if

  • Final camera-ready figures that need exact data accuracy, since generated images can contain wrong labels or arrows and must be checked manually
  • Users without a local Codex app-server setup, because the skill cannot render images without that bridge
  • Sensitive or unpublished methods shared with external services, because prompts and drafts pass through the Codex and reviewer tools

Try asking

  • “Draw a system overview figure for my method described in method.md, in CVPR style”
  • “Create a pipeline diagram with three encoder stages and labelled arrows for my paper”
  • “Iterate on the existing figure in figures/ai_generated and tighten the labels”
Step-by-step guide: how to use Paper Illustration via Codex Image2Install, first run, examples, tips and FAQs

Install in 3 steps

  1. 1

    Download the skill

    The ZIP contains the 2 original files, the MIT licence and an attribution note.

    paper-illustration-image2.zip
  2. 2

    Add it to your agent

    Claude Code (all projects)

    mkdir -p ~/.claude/skills && unzip paper-illustration-image2.zip -d ~/.claude/skills/

    Claude Code / Agent SDK (this project only)

    mkdir -p .claude/skills && unzip paper-illustration-image2.zip -d .claude/skills/

    Claude apps (web/desktop): Settings → Capabilities → Skills → Upload skill, and choose the ZIP.

  3. 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 text description of components, labels and data flow so the planner has concrete input
  • Check every generated figure for correct text, arrows and numbers before using it in a paper
  • Keep the Codex bridge and reviewer backends configured locally; review what prompts are sent to external services

What's inside

  • scripts/paper_illustration_image2.py
  • 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.