Scientific Manuscript Skill

Craft and refine manuscripts for high-impact journals through structured feedback, comparative analysis with exemplar papers, and sentence-level prose craft.

Core Focus Areas

This skill emphasizes four dimensions of high-impact writing:

1. Narrative arc — how the story builds across sections and paragraphs
2. Paragraph architecture — how individual paragraphs create momentum and impact
3. Sentence-level craft — concise, punchy prose with word economy
4. Strategic rhetoric — calculated use of adjectives and flourishes

Reference Files

Section-Specific Guides

Craft-Level Guides (Load for any section)

Annotated Examples

Feedback Modes

Mode 1: Narrative Analysis

Analyze story arc and momentum. Use when:

• Draft feels flat or disjointed
• User asks about "flow" or "structure"
• Sections don't build toward a conclusion

Load: references/narrative-structure.md

Format:

**Arc diagnosis**: [Where tension is established, where it resolves, what's missing]
**Momentum check**: [Which paragraphs build vs. stall]
**Pivot analysis**: [How you transition from known → unknown → your contribution]
**Suggested restructuring**: [Concrete reordering or additions]

Mode 2: Prose Craft Review

Sentence-level style analysis. Use when:

• User asks about "style" or "writing quality"
• Prose feels bloated or academic
• User wants text to be "punchier" or "more concise"

Load: references/prose-craft.md

Format:

**Bloat patterns**: [Specific phrases to cut with rewrites]
**Verb weakness**: [Weak verbs → strong verb suggestions]
**Sentence rhythm**: [Where variation is needed]
**Adjective audit**: [Which add meaning vs. clutter]
**Power positions**: [Sentences where key info is buried]

Mode 3: Comparative Critique

Compare to exemplar papers. Use when:

• User wants to see "how good papers do it"
• Draft has rhetorical issues best shown by contrast

Load: Relevant example file from references/examples/

Format:

**Your draft**: [quote passage]
**Exemplar**: [quote from example paper]
**Key difference**: [What the exemplar does that yours doesn't]
**Principle**: [The generalizable technique]
**Suggested revision**: [Concrete rewrite applying the principle]

Mode 4: Rubric Evaluation

Systematic scoring. Use when:

• User asks "Is this ready for submission?"
• Comprehensive assessment needed

Load: references/rubrics.md

Mode 5: Iterative Dialogue

Socratic questioning. Use when:

• Logic or argument needs development
• User refining specific claims

Quick Diagnostics

The Arc Check

For any section, ask:

1. Where is the tension established?
2. Where does it resolve?
3. Does each paragraph advance toward resolution?

The First/Last Test

• First sentence of each paragraph: Does it state the main point?
• Last sentence of the paper: Is it memorable and quotable?
• Last sentence of each section: Does it transition or conclude powerfully?

The Economy Test

Flag these bloat patterns:

• "It is important to note that" → [delete]
• "plays a role in" → "regulates" / "drives"
• "is involved in" → [specific verb]
• "in order to" → "to"
• Stacked adjectives → choose one

The Flourish Audit

Flourishes should appear at:

• Paper opening (hook)
• Key conceptual pivot
• Paper closing (take-home)

NOT in: routine findings, methods, figure descriptions

Handling Requests

"Review my introduction" → Load introduction.md + narrative-structure.md, analyze arc and pivot, check prose craft

"Make this punchier" → Load prose-craft.md, do sentence-level audit, provide rewrites

"How does this compare to good examples?" → Load relevant example file, do comparative critique

"Is the narrative working?" → Load narrative-structure.md, analyze tension-resolution arc, check paragraph momentum

"Help me cut this down" → Load prose-craft.md, identify bloat patterns, suggest cuts with preserved meaning

"Is this ready for submission?" → Load rubrics.md + all relevant references, comprehensive evaluation

Output Principles

1. Quote exactly: Reference specific passages from the draft
2. Rewrite concretely: Don't just diagnose—show the fix
3. Be honest: User wants critical feedback, not encouragement
4. Prioritize: Identify 2-3 highest-impact changes
5. Model excellence: Pull from exemplar papers to show what good looks like