Paper Deep Reading Teaching Explainer
PassAudited by ClawScan on May 4, 2026.
Overview
This skill is coherent for paper reading and teaching workflows, with disclosed external image-generation and optional local PDF-assembly steps that users should run only intentionally.
This skill appears safe to use for its intended paper-reading and teaching workflow. Before installing, check whether your papers or LaTeX sources contain confidential content, use image-generation stages only with providers you approve, and run the optional PDF assembly helper only on the intended project image folder.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
If used, the helper script will read local images and write a PDF output, so incorrect paths could process the wrong files.
The skill documents a local Python helper for assembling generated images into a PDF. This is purpose-aligned, but it is still local file-processing code that should be run only on intended image folders.
Use `scripts/assemble_storyboard_pdf.py` or an equivalent local implementation with PIL/Pillow or ReportLab.
Run the PDF assembly step only after reviewing the input directory and output path, and keep it inside the project workspace.
Paper details, summaries, claims, or visual prompts may be shared with the selected image-generation provider during visual steps.
The skill may route paper-derived storyboard prompts to an image-generation skill or external image API. This is disclosed and central to the visual workflow, but it creates a data-sharing boundary users should understand.
Codex / Claude Code / coding-agent environments: prefer the `imagegen` skill when available. If it is unavailable or insufficient, use ChatGPT Images 2.0 API or another user-approved image-generation API.
Use visual generation only for papers you are allowed to share with the chosen provider, and prefer explicitly approved tools or private/local options for confidential work.
Private draft comments or accidental instructions inside LaTeX sources could influence outputs or be summarized into downstream artifacts.
The skill may treat uploaded LaTeX source comments as usable context. This can improve paper understanding, but comments may contain private notes or irrelevant instructions if the source bundle is not sanitized.
LaTeX-only input: use `figure`/`table` environments, captions, labels, `\includegraphics` paths, equations, section text, appendix content, and source comments when useful.
Before uploading LaTeX source for this skill, remove private comments, credentials, internal notes, or prompt-like instructions that should not affect the report or visual prompts.