/rebuttal-response

name: rebuttal-response
description: Analyze, draft, revise, or audit scientific peer-review response letters and point-by-point rebuttals. Use for reviewer comments, response letters, rebuttals, revision summaries, quoted manuscript changes, replies that must align with a manuscript or Supplementary Information, and submission-ready checks of direct question-answer alignment, evidence, figures, tables, notes, panels, terminology, numbers, and internal citations. Especially useful when replies must reuse the reviewer's terms, map every sub-question to an explicit answer, report new analyses or experiments, identify exact revision locations, or distinguish mandatory fixes from optional polishing.

Revise Reviewer Responses

Build reviewer replies for a busy reader. Begin every formal response with a brief expression of thanks to the reviewer. Then make every answer immediately traceable to the reviewer's exact concern: lead with what was done and found, state what it means, and identify where the manuscript changed. Do not make the reviewer infer the mapping, search across paragraphs, or read defensive prose after the answer is already clear.

Start from the authoritative materials

1. Identify the newest response letter, manuscript, Supplementary Information, figures, tables, and result files relevant to the comment. 2. Treat the user's stated version hierarchy as binding. Do not revive text or conclusions from an older response when the user says the manuscript or SI is authoritative. 3. Read the complete reviewer comment and the complete current response before editing either. 4. Split the comment into independently answerable requests. Preserve the reviewer's own nouns and distinctions, including datasets, methods, comparison units, requested analyses, and claimed outcomes. 5. Build a private alignment matrix with one row per request:

• reviewer's exact concern or phrase;
• direct answer;
• action or evidence;
• result;
• supported conclusion;
• revision location;
• essential evidence boundary, if one is genuinely required.

6. Use the matrix as an acceptance test. Every row must appear explicitly in the response; no paragraph may substitute an adjacent analysis for the requested object. 7. Never invent a result, statistic, experiment, citation, revision location, or placeholder value. Mark unavailable values explicitly or ask for the missing source.

If the task involves direct document editing, also use the relevant document skill and render the finished file for visual verification.

Analyze before drafting

Unless the user explicitly requests only final prose, begin with a short diagnosis:

• What is the reviewer actually asking for?
• Is the concern methodological, evidential, interpretive, presentational, or terminological?
• Does it require a new experiment, a new analysis of existing results, a clarification, or only narrower wording?
• What is the strongest conclusion the available evidence supports?
• What statement in the manuscript, SI, or response must change for the reply to be credible?

Do not assume that adding an experiment is automatically the best response. Prefer the smallest action that genuinely resolves the concern, but do not use wording changes to conceal a real evidential gap.

Read [triage-and-stance.md](references/triage-and-stance.md) to classify the comment by cause and to decide whether the reply should concede, clarify, or push back. Make that decision before drafting; claim calibration applies to the stance you chose, it does not substitute for choosing one.

Choose the response architecture by complexity

Do not impose one template on every comment.

• **Simple comment:** direct answer or action, corrected text if useful, and revision location in one compact paragraph.
• **Moderate substantive comment:** one stand-alone overview followed by two or three sections named after the reviewer's actual sub-questions, then a precise revision-location paragraph.
• **Complex multi-part major comment:** one stand-alone overview, an optional `Specifically, we:` action list, matching evidence sections in exactly the same order, and a precise revision-location paragraph.

Use an action list only when it materially helps navigation. If used, make it a true table of contents: its item count, order, concepts, and labels must match the detailed sections. Do not place a top-level action beside a lower-level result such as gene and drug analyses if both belong under one experiment.

The overview must stand alone for a reviewer who reads only the first paragraph. Its first sentence must briefly thank the reviewer. Vary the wording naturally according to the comment, for example `We thank the reviewer for this important comment`, `We appreciate the reviewer's helpful suggestion`, or `We thank the reviewer for raising this point`. Do not use the same phrase mechanically throughout the letter, add excessive praise, agree with a premise that is not accepted, or paraphrase the full comment in the thank-you sentence.

Immediately after the thank-you sentence, open the substantive response with the action, result, or direct answer that resolves the concern, not a sentence merely announcing that the concern was addressed. Follow this order whenever the evidence permits:

1. brief thanks; 2. action or experiment; 3. central result; 4. direct answer to the reviewer; 5. manuscript change.

Name detailed sections with the reviewer's question objects rather than the author's internal reasoning categories. Reuse the reviewer's exact technical terms when accurate. If the reviewer separately asks about two representations, datasets, baselines, settings, or outcomes, answer them separately.

Read [response-patterns.md](references/response-patterns.md) when drafting or substantially restructuring a reply.

Make every paragraph reviewer-navigable

• Put the paragraph's conclusion or function in its first sentence.
• Make the first substantive sente