name: paper-revision description: Academic paper revision assistant for processing reviewer comments, generating response letters, and modifying manuscripts. Use when user mentions "revision", "reviewer comments", "response letter", "返修", "审稿意见", "修改稿", or when working with academic journal submissions requiring systematic comment-by-comment responses. allowed-tools: Read, Write, Edit, Glob, Grep, Bash(python:*), TodoWrite

Paper Revision Skill

Academic paper revision workflow assistant for systematic processing of reviewer comments.

When to Use This Skill

Activate when user:

• Receives reviewer comments or decision letters from journals
• Needs to process manuscript revision requirements
• Wants to generate response letters
• Asks about "返修", "审稿意见", "response letter", "reviewer comments"
• Works on journal resubmission workflow

Quick Reference

• Workflow details: WORKFLOW.md
• Response templates: TEMPLATES.md
• Quality standards: QUALITY.md

Core Workflow Overview

Phase 1: PARSE     → Structure reviewer comments
Phase 2: ANALYZE   → Diagnose each comment's requirements
Phase 3: STRATEGIZE → Plan response approach
Phase 4: EXECUTE   → Modify manuscript + write responses
Phase 5: VERIFY    → Check consistency
Phase 6: PACKAGE   → Prepare submission materials

Project Structure

When initiating a revision project, create:

revision_project/
├── CLAUDE.md                    # Project context (revision-specific)
├── reviewer_comments.txt        # Original comments input
├── tracking/
│   └── revision_tracking.md     # Progress matrix
├── responses/
│   ├── response_letter.md       # Main response document
│   └── comment_[id]/            # Per-comment analysis
│       ├── analysis.md
│       └── strategy.md
├── manuscript/
│   ├── revised_marked.docx      # Yellow highlighted version
│   └── revised_clean.docx       # Clean version
└── submission/                  # Final package

Comment Classification System

Response Letter Format

Four-Part Framework

### Comment [ID]: [Topic]

**Original Comment**:
> "[Exact reviewer text]"

**RESPONSE**:

**Part 1 - Acknowledgment**:
Thank you for this [insightful/constructive] observation regarding [topic].

**Part 2 - Modification Summary**:
We have addressed this through [N] modifications:
1. [Category]: [Brief description]
2. [Category]: [Brief description]

**Part 3 - Evidence & Location**:
**Revised text ([Section], Paragraph [N])**:
> "[Exact revised text as it appears in manuscript]"

**Supporting data**: [Table/Figure reference if applicable]

**Part 4 - Conclusion** (if needed):
[How this addresses the reviewer's concern]

Quality Assurance

Four-Layer Check System

Verification Checklist

For each comment, verify:

• Response cites correct paragraph numbers
• All modified text has yellow highlight in manuscript
• No conflicting statements with other comments
• Data/statistics are accurate
• Claims are verifiable in manuscript

Decision Tree for Comment Handling

Comment received
├─ Data/Statistics issue?
│   ├─ YES → Reanalysis needed
│   │   ├─ Existing data suffices? → Script + New Table
│   │   └─ Need new data? → Document limitation
│   └─ NO → Text modification
│       ├─ New section needed? → Draft 500-2000 words
│       └─ Paragraph edit? → Targeted modification
├─ Format/Terminology?
│   └─ Global find-replace + Review
└─ Multiple comments overlap?
    └─ Integrated handling with cross-references

Reviewer Psychology Patterns

Journal-Specific Formats

JMIR:
  p_value: "P<.05"
  citation: "[1]"
  supplementary: "Multimedia Appendix"

Nature:
  p_value: "P < 0.05"
  citation: superscript
  supplementary: "Extended Data"

Lancet:
  p_value: "p<0·05"
  citation: superscript
  supplementary: "Appendix"

Instructions for Claude

When user initiates a revision task:

Step 1: Understand Context

• Ask for reviewer comments if not provided
• Identify journal name and revision round
• Check if project structure exists

Step 2: Parse Comments

• Extract each comment with reviewer ID
• Classify priority (🔴🟡🟢⚪)
• Estimate complexity and time

Step 3: Process Each Comment

For each comment:

1. Analyze: What is the root issue?
2. Strategize: What approach to take?
3. Execute: Generate response + manuscript changes
4. Verify: Check consistency

Step 4: Quality Assurance

• Run four-layer checks
• Flag any inconsistencies
• Generate verification report

Step 5: Package for Submission

• Compile response letter
• Prepare marked and clean manuscripts
• Create submission checklist

Key Principles

1. Systematic Processing: Handle comments one by one, track progress
2. Evidence-Based Responses: Every claim must be verifiable
3. Precise Locations: Always cite "Section, Paragraph N"
4. Highlight All Changes: Yellow highlight in manuscript
5. Professional Tone: Respectful, grateful, transparent
6. Consistency First: Response must match manuscript exactly

Common Pitfalls to Avoid

• Missing highlights for claimed modifications
• Paragraph numbers shifting after edits
• Conflicting statements across comments
• Vague locations ("we revised the Methods section")
• Unsupported claims about data or results

Example Session Flow

User: "I received revision comments from JMIR, help me process them"

Claude:
1. Request reviewer comments file
2. Create project structure
3. Parse and prioritize comments
4. For each comment:
   - Show analysis
   - Propose strategy
   - Generate response draft
   - Suggest manuscript modifications
5. Run quality checks
6. Prepare submission package

Integration with Other Skills

• Use with /sc:research for literature support
• Use with /sc:analyze for data verification
• Use with /sc:document for response formatting
• Use session management (/sc:save, /sc:load) for multi-session work