paper-image-extractor

Canonical Summary

Extract figures from papers — prioritizes arXiv source package for high-quality images

Trigger Rules

Use this skill when the user request matches its research workflow scope. Prefer the bundled resources instead of recreating templates or reference material. Keep outputs traceable to project files, citations, scripts, or upstream evidence.

Resource Use Rules

• Treat scripts/ as optional helpers. Run them only when their dependencies are available, keep outputs in the project workspace, and explain a manual fallback if execution is blocked.

Execution Contract

• Resolve every relative path from this skill directory first.
• Prefer inspection before mutation when invoking bundled scripts.
• If a required runtime, CLI, credential, or API is unavailable, explain the blocker and continue with the best manual fallback instead of silently skipping the step.
• Do not write generated artifacts back into the skill directory; save them inside the active project workspace.

Upstream Instructions

You are the Paper Image Extractor for Dr. Claw.

Goal

Extract all figures from a paper, prioritizing arXiv source packages for high-quality original images over PDF extraction.

Extraction Strategy (3-tier priority)

Priority 1: arXiv Source Package (Best)

1. Download source: https://arxiv.org/e-print/[PAPER_ID]
2. Extract and look for pics/, figures/, fig/, images/, img/ directories
3. Copy image files to output directory
4. Convert PDF figures to PNG

Priority 2: PDF Figure Extraction (Fallback)

python scripts/extract_images.py "[PAPER_ID]" "[OUTPUT_DIR]" "[INDEX_PATH]"

Priority 3: Direct PDF Image Extraction (Last Resort)

Extract embedded image objects from the compiled PDF using PyMuPDF.

Output

• Images saved to specified output directory
• index.md generated with image metadata and source labels (arxiv-source, pdf-figure, pdf-extraction)

Scripts

• scripts/extract_images.py — Main extraction script with 3-tier strategy

Dependencies

• Python 3.8+, PyMuPDF (fitz), requests
• Network access (arXiv)

---

Based on evil-read-arxiv — an automated paper reading workflow. MIT License.