/paper-reviewer

name: paper-reviewer
description: "Referee-side skill for Nature Portfolio manuscripts: write a referee report, inventory every ask and sub-ask in received comments, grade drafted replies, and test reply readability. Check methodology, statistics, reporting, reproducibility and data/code availability. Preserve referee wording, numbering and order; require one reply per ask with a direct opening answer and a manuscript location or explicit no-change rationale. Use for review this manuscript, what will reviewers attack, turn comments into a checklist, answer each comment, did I miss a point, would this reply satisfy the reviewer, 审稿、写审稿意见、模拟审稿、拆解审稿意见、逐个回复、逐条对照、有没有漏回、审稿人会不会接受、以审稿人视角检查回复、回复够不够直白. Use rebuttal-response to choose response stances, calibrate claims and assemble or audit the whole letter."

Referee Reports and the Replies That Answer Them

Two jobs, one standpoint. Writing a referee report and answering one are the same competence seen from two sides: you can only prove you answered a question if you can reconstruct the question that was asked.

Pick the mode before starting.

| Mode | Input | Output | |---|---|---| | `report` | a manuscript | a referee report, or a prediction of what referees will attack | | `inventory` | a referee report | every ask and sub-ask, in the referee's own words and order | | `reply` | a report plus a manuscript | one reply per ask, plus the coverage audit | | `grade` | a drafted reply | what an impatient referee would send back |

The contract

A referee report is a list of questions. A reply is a list of answers. The job is to make the two lists line up so exactly that a referee can verify it by scrolling their own report alongside your letter.

**Same count. Same identifiers. Same order. Same words. Depth matched to weight.**

The alignment has to be visible on the page, not asserted in a summary. Three consequences, and every rule in this skill follows from them:

1. **The referee's comment is reproduced verbatim, immediately above its reply.** Not paraphrased, not summarised, not trimmed. Nature Methods names rephrasing a referee's comment as a manipulative tactic, and Nature Chemistry asks authors to list the referees' individual points. 2. **The reply uses the referee's own words.** If they wrote `data leak`, the reply says `data leak`. Never coin a replacement term, a framework name, or a category label that the referee would have to learn in order to read your answer. Plain language comes from glossing their term, never from inventing a better one. 3. **The referee is impatient.** Nature Neuroscience puts it as `keeping in mind that referees are busy`. The first sentence carries the answer. Reply length is the minimum that settles the point. Over-answering a light comment is as much a defect as under-answering a heavy one, and Nature Chemistry warns specifically that a long-winded essay style makes a revision harder to follow.

What this skill does not own

• **`rebuttal-response`** decides what to say: the stance on each ask, the strongest claim the evidence

supports, honest handling of negative results, and keeping letter, manuscript, and Supplementary Information consistent. Build the inventory here, draft there, bring the draft back here for the impatient-referee pass.

• **`submission-audit`** runs a single referee-style rejection pass over a manuscript nobody has

reviewed yet. Use it before submitting; use this skill to simulate a referee beforehand and to work real reports afterwards.

• **`stats-reporting-audit`** owns statistical substance: the independent unit behind `n`, replication,

multiple-comparison correction, figure-legend statistics. When an ask is statistical, take the substance from there and only the reply shape from here.

• **`nature-portfolio-playbook`** owns venue and article-type choice. This skill assumes the venue is

fixed.

Mode `report`

Read [review_checklist.md](references/review_checklist.md) for the stage-by-stage criteria, the referee fields Nature Portfolio actually asks about, the report shape, and the rules for writing comments that can be answered one at a time. [common_issues.md](references/common_issues.md) is the catalogue of what to look for; [reporting_standards.md](references/reporting_standards.md) covers guideline compliance and the portfolio-wide Reporting Summary.

Mode `inventory`

Read [comment_splitting.md](references/comment_splitting.md). The rule in one line:

> Split wherever a different action would be needed to satisfy the referee.

Two clauses are one ask only if the same action settles both. The test: could the referee write in round 2, about this span alone, `the authors did not do this`?

Split on any enumerator the referee supplied, any request verb, any question, any sentence asserting a defect as fact, any separate target object, and any change of deliverable. Subtract a lead sentence that the sub-items enumerate, a request restated in different words, and the sentences that merely justify a request. An ask naming a set expands to one entry per element in the ledger, which is what stops a half-done set from slipping through. In the letter those entries may share one reply, provided it names every element and says what happened to each, including the ones needing no change. A range reply that names none of them reads the same whether you fixed four panels or two.

**Numbering is the referee's.** Never renumber, never reorder, never regroup by theme. Preserve gaps: if they commented on panels `a)` and `c)` and skipped `b)`, so does the reply. Preserve their typos inside the quote. If the referee numbered nothing, do not invent a scheme that looks like theirs; split at their paragraph breaks, use visibly author-added labels such as `R1-P1`, and say once that you added them.

Record the result in `notes/review-ledger.md`. It is a working artifact and it never enters the letter: a referee should not have to consult a table t