Skill Claude CodeCodex
Detects whether a piece of writing is AI-generated and provides specific, actionable rewrite suggestions to make it sound more human. Use this skill whenever the user wants to check if text looks AI-written, wants to 'humanize' AI text, asks whether their writing will pass an AI detector, or wants feedback on making…
Skill Claude CodeCodex
Use when a paper draft is complete and needs a 'zero-context layperson' review. The idiot-reviewer reads the paper paragraph by paragraph as a complete outsider with no prior knowledge of the research domain, asking every question a non-expert would have. This catches readability and accessibility issues that expert…
Skill Claude CodeCodex
Use when creating academic paper figures, architecture diagrams, flowcharts, method illustrations, supplemental figures, or converting generated images to same-name PDFs. Reads IMAGEGEN model settings from .env and supports gpt-image-compatible providers.
Skill Claude CodeCodex
Use when the user provides an annotated academic paper PDF and asks to revise LaTeX/TeX source, address PDF comments, apply similar fixes across the paper, or preserve reviewer/reader annotation issues for future ai-detector checks.
Skill Claude CodeCodex
Use when reproducible paper collection is needed. Supports multi-provider crawling (OpenAlex, arXiv, Semantic Scholar, USENIX), deduplication, normalized metadata, and optional PDF download.
Skill Claude CodeCodex
Use when screening and summarizing collected academic papers. Supports two-round multi-agent screening (title+abstract pre-screening, full-text deep screening), score aggregation, structured summaries, and screening report generation.
Skill Claude CodeCodex
Use when the user asks to read PDF annotations, export highlights, organize paper comments, convert PDF notes/comments/annotations to Markdown or JSON, or extract annotated text and surrounding context from an academic-paper PDF.
Skill Claude CodeCodex
Use when generating research ideas from paper summaries. Analyzes research gaps, generates ideas via multi-agent brainstorming, evaluates and ranks ideas, performs adversarial review, and produces detailed research proposals.
Skill Claude CodeCodex
Use when drafting, revising, polishing, or reviewing computer science research papers, including Abstract, Introduction, Related Work, Technical Core, Evaluation, Conclusion, figures, tables, claim-evidence alignment, or submission self-review.
Skill Claude CodeCodex
Use for adversarial review of any research artifact — papers, experimental designs, or research ideas. Launches three independent, context-free reviewers (strict, constructive, newcomer) with a 0–5 scoring rubric. Each reviewer produces detailed scores and actionable feedback.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: