Research Paper Figure Skill Factory
ReviewAudited by ClawScan on May 5, 2026.
Overview
This instruction-only skill is coherent for making research-paper figure workflows, but users should notice that it can process local research PDFs, create reusable skill packages, and use image-generation services.
This skill appears safe to install as an instruction-only workflow, but use it with a scoped research corpus, verify any generated skill ZIP before installing it elsewhere, and confirm image-generation steps before sending confidential paper content to external image tools.
Findings (5)
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.
Using the skill may invoke image-generation tooling and may create several images per decision point, which can affect time, quota, or cost.
The skill intentionally defaults to generating multiple candidate images when the user asks to generate without specifying a count.
"if_user_says_generate_without_count": "default_to_6_images"
Before generation, confirm the number of images and avoid putting confidential paper details into prompts unless you are comfortable sending them to the image-generation route being used.
A generated skill can influence future agent behavior, even if this factory skill itself contains no executable code.
The skill is meant to produce new reusable skill ZIP packages that can later be installed into agent environments.
生成完成后，你会得到一个新的专项制图 skill zip。之后把这个 zip 放入 ChatGPT Sources，或安装到 Codex skills 中
Review any generated skill package, especially its instructions and requested capabilities, before installing it into ChatGPT Sources, Codex skills, or another agent environment.
Private drafts, PDFs, captions, or figure observations may be stored in local artifacts and reused as context for later taxonomy or skill-generation steps.
The workflow creates structured extraction artifacts from papers and figures, which may include summaries of user-provided or local research materials.
<local_corpus.root_dir>/extracted/ paper_cards.csv ... figure_inventory.csv ... caption_inventory.csv ... evidence_map.json ... extraction_report.md
Use a dedicated corpus folder, include only lawful and intended materials, and review generated extraction artifacts before reusing or sharing them.
Earlier context may shape later figure decisions, which is useful for continuity but can also carry forward stale or unintended information.
The skill tells generated workflows to continue state from active session history by default.
默认根据当前 session/history 自动延续状态；用户不需要手动上传或粘贴状态。
If the conversation changes scope or contains sensitive material, ask the agent to reset or summarize the current state before proceeding.
Users may place extra trust in the skill based on the claimed author even though provenance is not verified by the provided source information.
The skill front matter claims an OpenAI author, while the supplied registry summary lists the source as unknown and no homepage is provided.
author: OpenAI
Treat the author claim as unverified unless you can confirm publisher provenance through the marketplace or an official source.