Journal Peer Review Skill — Development skill for Claude Code

journal-peer-review: a Claude Code skill

This skill turns a reviewer's own draft into a polished journal peer-review report (comments to the authors).

Scope: it polishes your review, it does not write it

**This skill does not review the manuscript for you.** It starts from an initial draft that you, the human reviewer, have written. That draft must already:

• list the strengths and weaknesses you want to raise, and say which concerns are major or minor;
• reflect your own recommendation (accept, minor revision, major revision or reject). You submit it separately, in the confidential comments to the editor.

From that draft, the skill only helps you:

• paraphrase your points into clear, kind, conventional reviewer language;
• reorganize them into the standard journal structure;
• check the structure and phrasing against conventions measured on published reviews;
• verify that every quote, number and page/line reference in your draft is correct.

It does not add criticisms of its own, decide which issues matter, or make the recommendation decision. This follows publisher policies on AI in peer review, such as Springer Nature's, which keep the evaluation and the recommendation with the human reviewer. Check your journal's policy, and declare any AI assistance it requires.

**Required input:** the manuscript PDF and your initial draft, in any format, even rough bullet points.

What it does

1. Setup. Builds a folder layout, extracts the manuscript text, and dumps every line with its page and margin line number.
2. Background notes (optional, for your own reading). Fact-finding notes on internal consistency, the journal's conventions and the topic literature (from full texts). You may consult them while writing your draft. They don't replace it, and nothing goes into the review unless you put it in your draft.
3. Conversion. Paraphrases and reorganizes your draft into the conventional journal format:
   • an unheaded summary;
   • an