Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Systematic review response strategy guide for RebuttalStudio. Use when developing response strategies for reviewer comments, deciding how to classify concerns, or choosing between Accept/Defend/Clarify/Experiment approaches at any stage of the rebuttal pipeline.
.claude/skills/runtsang-review-response/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 28% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 87% | 0% |
| case-09 | ✗→✓ | ▲ Improved | 48% | 0% |
| case-21 | ✗→✓ | ▲ Improved | 95% | 0% |
| case-22 | ✗→✓ | ▲ Improved | 86% | 0% |
> RebuttalStudio Utility — Stage-General > This skill applies across the full 5-stage rebuttal pipeline. It provides strategic guidance for classifying reviewer comments and choosing response approaches, complementing the stage-specific breakdown and refinement skills.
A systematic framework for analyzing reviewer comments and developing high-quality, professionally grounded rebuttal responses.
Every reviewer comment belongs to one of four types. Identify the type before selecting a strategy.
| Type | Description | Examples | |------|-------------|---------| | Major | Core scientific or methodological concerns that affect the paper's conclusions | "The baseline comparison is unfair", "The claims are unsupported by experiments" | | Minor | Peripheral concerns about presentation, wording, or supplementary experiments | "Figure 3 is hard to read", "Missing ablation on X" | | Misunderstanding | Reviewer has misread or misinterpreted the paper | "The authors do not compare with Y" (but Y is in Table 2) | | Typo / Clarity | Factual errors, grammar, or unclear exposition | "Definition 2 appears to be incorrect", "Section 3 is confusing" |
After classification, select the appropriate response strategy:
| Strategy | When to Use | Core Approach | |----------|-------------|---------------| | Accept | Reviewer is correct; the concern is valid | Acknowledge explicitly, describe the change made, point to location | | Defend | Reviewer's concern is based on a valid scientific disagreement | Provide evidence, cite prior work, offer additional data if available | | Clarify | Reviewer has a misunderstanding that the paper actually addresses | Quote the relevant section, explain why the concern does not apply | | Experiment | Reviewer requests additional empirical validation | Describe the experiment, provide preliminary results if possible |
These apply to every response at every stage, from Stage 2 draft outlines to Stage 4 follow-up refinement:
Key lessons distilled from analyzing accepted papers' rebuttals:
| Stage | How to Use This Skill | |-------|-----------------------| | Stage 1 — Breakdown | After receiving reviewer text, apply the classification table to categorize each atomic issue before writing titles | | Stage 2 — Reply | Use the Accept/Defend/Clarify/Experiment framework when drafting outlines; choose the strategy that matches each issue type | | Stage 4 — Multi-Round | Re-apply classification to follow-up comments; use "Clarify" or "Experiment" strategy for reviewer persistence | | Stage 5 — Final Remarks | Summarize the response strategies used for each reviewer in the final remarks |
For a well-structured per-reviewer rebuttal block:
We thank Reviewer [X] for their thoughtful and constructive feedback.
**Response to [Issue Title]**
> "[quoted reviewer comment]"
[Strategy: Accept / Defend / Clarify / Experiment]
[Response body: acknowledgment → direct answer → evidence → forward-looking note]
**Changes made**: [Section X, paragraph Y; or "no change needed because…"]Adapted from Claude Scholar's review-response skill by gaoruizhang. Original: https://github.com/Galaxy-Dawn/claude-scholar/blob/main/skills/review-response/SKILL.md
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 22,206 | 20,559 | -7% | 1 | 1 | 0% | 3,094 | 3,958 | +28% | 0 | 0 | — |
case-02 | fail→fail | 15,998 | 17,422 | +9% | 1 | 1 | 0% | 2,635 | 3,791 | +44% | 0 | 0 | — |
case-03 | fail→pass | 8,797 | 9,455 | +7% | 1 | 1 | 0% | 1,588 | 2,970 | +87% | 0 | 0 | — |
case-04 | pass→pass | 14,051 | 12,517 | -11% | 1 | 1 | 0% | 2,265 | 3,290 | +45% | 0 | 0 | — |
case-05 | pass→pass | 11,869 | 13,530 | +14% | 1 | 1 | 0% | 2,026 | 3,470 | +71% | 0 | 0 | — |
case-06 | pass→pass | 13,717 | 9,720 | -29% | 1 | 1 | 0% | 2,200 | 2,689 | +22% | 0 | 0 | — |
case-07 | fail→fail | 12,826 | 10,939 | -15% | 1 | 1 | 0% | 2,051 | 2,850 | +39% | 0 | 0 | — |
case-08 | pass→pass | 9,180 | 7,846 | -15% | 1 | 1 | 0% | 1,502 | 2,397 | +60% | 0 | 0 | — |
case-09 | fail→pass | 11,689 | 9,138 | -22% | 1 | 1 | 0% | 2,000 | 2,959 | +48% | 0 | 0 | — |
case-10 | pass→pass | 14,179 | 10,635 | -25% | 1 | 1 | 0% | 2,089 | 3,260 | +56% | 0 | 0 | — |
case-11 | pass→pass | 10,657 | 11,242 | +5% | 1 | 1 | 0% | 1,626 | 2,846 | +75% | 0 | 0 | — |
case-12 | pass→pass | 10,691 | 6,291 | -41% | 1 | 1 | 0% | 1,535 | 2,151 | +40% | 0 | 0 | — |
case-13 | pass→pass | 10,987 | 9,978 | -9% | 1 | 1 | 0% | 1,705 | 2,856 | +68% | 0 | 0 | — |
case-14 | pass→pass | 11,687 | 9,118 | -22% | 1 | 1 | 0% | 1,884 | 2,692 | +43% | 0 | 0 | — |
case-15 | fail→fail | 13,998 | 8,991 | -36% | 1 | 1 | 0% | 2,131 | 2,692 | +26% | 0 | 0 | — |
case-16 | pass→pass | 15,909 | 11,410 | -28% | 1 | 1 | 0% | 2,255 | 2,809 | +25% | 0 | 0 | — |
case-17 | pass→pass | 14,056 | 12,447 | -11% | 1 | 1 | 0% | 2,048 | 3,046 | +49% | 0 | 0 | — |
case-18 | fail→fail | 9,050 | 5,268 | -42% | 1 | 1 | 0% | 1,572 | 2,217 | +41% | 0 | 0 | — |
case-19 | pass→pass | 8,442 | 9,281 | +10% | 1 | 1 | 0% | 1,566 | 2,712 | +73% | 0 | 0 | — |
case-20 | pass→pass | 19,808 | 15,488 | -22% | 1 | 1 | 0% | 2,767 | 3,756 | +36% | 0 | 0 | — |
case-21 | fail→pass | 8,806 | 7,718 | -12% | 1 | 1 | 0% | 1,371 | 2,668 | +95% | 0 | 0 | — |
case-22 | fail→pass | 7,506 | 6,594 | -12% | 1 | 1 | 0% | 1,323 | 2,456 | +86% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +23 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/28/2026 | +18% |
Other measured skills in the registry, with their headline benchmark lift.