AgentSkillsCN

scientific-writing

按照 IMRAD 结构、APA/AMA/温哥华引用规范、图表/表格,以及 CONSORT/STROBE/PRISMA 报告指南,撰写科学论文。在起草任何论文章节、提升写作清晰度,或为期刊投稿做准备时使用。

SKILL.md
--- frontmatter
name: scientific-writing
description: Write scientific manuscripts with proper structure (IMRAD), citations (APA/AMA/Vancouver), figures/tables, and reporting guidelines (CONSORT/STROBE/PRISMA). Use when drafting any manuscript section, improving writing clarity, or preparing for journal submission.

Scientific Writing

Write clear, precise, and publication-ready scientific manuscripts.

When to Use

  • Drafting manuscript sections (abstract, intro, methods, results, discussion)
  • Structuring a research paper using IMRAD format
  • Formatting citations and references
  • Creating or improving figures and tables
  • Applying reporting guidelines (CONSORT, STROBE, PRISMA)
  • Preparing manuscripts for journal submission
  • During the WRITING or REVIEW phases

Manuscript Structure (IMRAD)

code
┌─────────────────────────────────────────────────────────────┐
│ TITLE                                                       │
│ Concise, specific, informative (12-15 words)               │
├─────────────────────────────────────────────────────────────┤
│ ABSTRACT (150-250 words)                                    │
│ Background → Objective → Methods → Results → Conclusion     │
├─────────────────────────────────────────────────────────────┤
│ INTRODUCTION                                                │
│ Context → Gap → Objective → Approach                        │
│ Funnel: Broad → Narrow → Your question                     │
├─────────────────────────────────────────────────────────────┤
│ METHODS                                                     │
│ Study design → Participants → Procedures → Analysis         │
│ Enough detail for replication                              │
├─────────────────────────────────────────────────────────────┤
│ RESULTS                                                     │
│ Objective findings, no interpretation                       │
│ Text + Figures + Tables                                     │
├─────────────────────────────────────────────────────────────┤
│ DISCUSSION                                                  │
│ Key findings → Context → Limitations → Implications         │
│ Reverse funnel: Specific → Broad                           │
├─────────────────────────────────────────────────────────────┤
│ REFERENCES                                                  │
│ Consistent style, verified DOIs                            │
└─────────────────────────────────────────────────────────────┘

Section-by-Section Guidance

Abstract

Purpose: Standalone summary of the entire paper

Structure (for structured abstracts):

  • Background: Why this matters (1-2 sentences)
  • Objective: What you did (1 sentence)
  • Methods: How you did it (2-3 sentences)
  • Results: Key findings with numbers (3-4 sentences)
  • Conclusion: Main takeaway (1-2 sentences)

Tips:

  • Write LAST after all other sections
  • Include specific numbers/results
  • Avoid abbreviations (or define them)
  • Stay within word limit (usually 150-250)

Introduction

Purpose: Establish context, gap, and rationale

Structure (Funnel):

  1. Broad context (1-2 paragraphs): Why does this field matter?
  2. Current knowledge (2-3 paragraphs): What's known? What approaches exist?
  3. Gap/Problem (1 paragraph): What's missing? What's the controversy?
  4. Your study (1 paragraph): What did you do? Why this approach?

Tips:

  • End with clear objectives or hypotheses
  • Cite 20-40 references typically
  • Use present tense for established facts
  • Be specific about what you're studying

Methods

Purpose: Enable replication

Key Sections:

  1. Study Design: Type of study, setting, dates
  2. Participants/Samples: Selection, criteria, sample size, ethics
  3. Procedures: What was done, in order
  4. Measurements: What and how measured
  5. Statistical Analysis: Tests, software, significance criteria

Common Mistakes:

MistakeProblemFix
Vague methods"Standard methods"Specify exact protocol
Missing stats"Data were analyzed"Name specific tests
No software versionsNot reproducibleInclude version numbers
Missing sample size justificationWhy this n?Add power analysis

Tips:

  • Use past tense
  • Be specific: model numbers, concentrations, durations
  • Reference published protocols if applicable
  • Include ethical approvals

Results

Purpose: Present findings objectively

Organization:

  1. Order by importance or by methods flow
  2. Each paragraph: finding + evidence (figure/table reference)
  3. Stats: test, statistic, df, p-value, effect size

Structure Pattern:

code
[What was found] (Figure X).
[Statistical support] (t(df) = X.XX, p = .XXX, d = X.XX).
[Additional detail or subgroup analysis].

Tips:

  • NO interpretation (save for Discussion)
  • Include negative/null results
  • Reference every figure and table
  • Use past tense
  • Include exact p-values (not just p < 0.05)

Discussion

Purpose: Interpret findings in context

Structure (Reverse Funnel):

  1. Key findings (1-2 paragraphs): Main results, directly address objectives
  2. Comparison to literature (2-3 paragraphs): How do findings fit with prior work?
  3. Mechanisms (1-2 paragraphs): Why might this happen?
  4. Limitations (1 paragraph): Be honest and specific
  5. Implications (1-2 paragraphs): Clinical, practical, theoretical significance
  6. Future directions (optional): What next?
  7. Conclusion (1 paragraph): Main takeaway

Tips:

  • Start with your results, not literature
  • Acknowledge limitations honestly
  • Don't overstate conclusions
  • Distinguish correlation from causation

Citation Styles

APA (7th Edition)

In-text: (Author, Year) or Author (Year)

code
Previous research found significant effects (Smith, 2023).
Smith (2023) reported significant effects.

Reference list:

code
Smith, J. D., Johnson, M. L., & Williams, K. R. (2023). Title of 
    article. Journal Name, 22(4), 301-318. https://doi.org/10.xxx/yyy

Vancouver/ICMJE

In-text: Superscript or bracketed numbers¹ or [1]

code
Previous research found significant effects.¹
Multiple studies support this finding.¹⁻³

Reference list (numbered):

code
1. Smith JD, Johnson ML, Williams KR. Title of article. J Name. 
   2023;22(4):301-18.

Nature

In-text: Superscript numbers¹

code
Previous research found significant effects¹.

Reference list:

code
1. Smith, J. D., Johnson, M. L. & Williams, K. R. Title of article. 
   Nat. Rev. Drug Discov. 22, 301-318 (2023).

Figures and Tables

When to Use Which

Use Tables ForUse Figures For
Exact values neededTrends and patterns
Many variablesComparisons
Summary statisticsRelationships
Participant characteristicsProcesses

Figure Checklist

  • Self-explanatory with caption
  • Axes labeled with units
  • Error bars defined (SEM, SD, CI)
  • Significance markers explained
  • Colorblind-safe
  • Resolution ≥300 DPI

Table Checklist

  • Clear, descriptive title
  • Column headers with units
  • Appropriate precision (not too many decimals)
  • Notes for abbreviations
  • n values included

Caption Template

markdown
**Figure 1. Brief descriptive title.**

(A) Description of panel A. (B) Description of panel B. 
Data shown as mean ± SEM (n = X per group). Statistical 
comparisons by [test name]. *p < 0.05, **p < 0.01, ***p < 0.001.

Reporting Guidelines

Which Guideline to Use

Study TypeGuidelineURL
Randomized trialCONSORTconsort-statement.org
Observational (cohort, case-control)STROBEstrobe-statement.org
Systematic reviewPRISMAprisma-statement.org
Diagnostic accuracySTARDstard-statement.org
Prediction modelsTRIPODtripod-statement.org
Animal researchARRIVEarriveguidelines.org
Case reportsCAREcare-statement.org
Quality improvementSQUIREsquire-statement.org

Using Checklists

  1. Download checklist from guideline website
  2. Complete each item during writing
  3. Include page/line numbers
  4. Submit with manuscript (often required)

Writing Principles

Clarity

  • Use precise, unambiguous language
  • One idea per sentence
  • Define technical terms at first use
  • Use active voice when possible

Example:

code
❌ "The samples were subjected to analysis"
✓ "We analyzed the samples using..."

❌ "It has been shown that..."
✓ "Smith et al. (2023) showed that..."

Conciseness

WordyConcise
"Due to the fact that""Because"
"In order to""To"
"A large number of""Many"
"At the present time""Now" / "Currently"
"In the event that""If"
"Has the ability to""Can"

Accuracy

  • Report exact values with appropriate precision
  • Use consistent terminology
  • Distinguish observation from interpretation
  • Acknowledge uncertainty

Objectivity

  • Present results without bias
  • Don't overstate findings
  • Acknowledge contradictory evidence
  • Maintain professional, neutral tone

Field-Specific Terminology

General Principles

  • Match terminology to the target journal
  • Use established nomenclature systems
  • Define abbreviations at first use
  • Be consistent throughout

Quick Reference

FieldConvention
GenesItalics (BRCA1)
ProteinsRoman (BRCA1)
SpeciesItalics, full at first (Escherichia coli), then abbreviated (E. coli)
StatisticsItalics (p, n, t, F, r)
DrugsGeneric name first, brand in parentheses

Common Mistakes to Avoid

Top Rejection Reasons

  1. Incomplete or inappropriate statistics
  2. Over-interpretation of results
  3. Poor methods description
  4. Inadequate sample size
  5. Poor writing quality
  6. Inadequate literature review
  7. Unclear figures
  8. Failure to follow guidelines

Writing Issues

IssueExampleFix
Tense mixing"We collected... and analyze"Past for methods/results
Excessive jargonToo many undefined termsDefine or simplify
Paragraph breaksRandom breaksOne topic per paragraph
Missing transitionsAbrupt section changesAdd linking sentences

Manuscript Development Workflow

Recommended Order

  1. Figures/Tables first (core data story)
  2. Methods (often easiest to draft)
  3. Results (describe figures/tables)
  4. Discussion (interpret findings)
  5. Introduction (set up the question)
  6. Abstract (synthesize everything)
  7. Title (last refinement)

Revision Checklist

  • Logical flow throughout
  • Consistent terminology
  • All figures/tables referenced
  • All citations verified
  • Word counts met
  • Reporting checklist complete
  • Journal format requirements met

Integration with RA Workflow

WRITING Phase Files

FilePurpose
manuscript/background.mdIntroduction content
manuscript/methods.mdMethods section
manuscript/results.mdResults + figure refs
manuscript/discussion.mdDiscussion section
manuscript/figures/figN/caption.mdFigure captions

Connected Skills

  • /write_background: Drafts introduction
  • /write_methods: Generates methods from scripts
  • /write_results: Drafts results from figures
  • /peer_review: Self-review before submission