---
name: academic-paper-writing
description: Use this skill when drafting or structuring an academic or research paper — literature review, argument scaffolding, or journal formatting and submission. It produces a rigorous, well-cited first draft for the author's own review and verification.
---

# Academic Paper Writing

> **What this is** — a repeatable, AI-assisted working method for taking a request in academic and research paper writing, often beyond one's own formal training, and producing a rigorous, well-structured first pass quickly, with a qualified professional kept in the loop.
> **What this is NOT** — not a substitute for domain peer expertise, institutional research-ethics review, or original research; the author remains solely responsible for accuracy, originality, citation integrity, and authorship/ethics compliance. Never fabricate sources, quotations, or data. Every output is a draft for review before it is relied on, published, or shipped.

## When to use this
- A literature review needs a defensible search strategy, screening record, and thematic synthesis.
- An argument or manuscript needs structural scaffolding (thesis, sections, topic sentences, signposting).
- A draft needs formatting for a target journal and a submission package (cover letter, style compliance).
- A reviewer's report has arrived and needs a structured, point-by-point response letter.
- The author wants a rigorous first pass to react to and correct, not a finished paper.

## Operating principle
Nothing is invented. Every citation, quotation, statistic, and claim must trace to a source the author can open and verify; unverifiable material is marked as a gap, never filled with a plausible-looking reference. The author is the sole authority on accuracy, originality, and ethics — this method organises and drafts, it does not certify scholarship.

## Capability 1 — Literature review synthesis
**Goal.** Turn a research question into a transparent, reproducible literature review.
**Inputs.** The research question, candidate databases, date range, and inclusion/exclusion criteria.
**Method.**
1. Define the question and derive concepts; build a search strategy with database list (e.g. Scopus, Web of Science, PubMed, IEEE Xplore, Google Scholar) and boolean strings per database.
2. Record inclusion/exclusion criteria (language, date, study type, relevance) before screening.
3. Run searches, deduplicate, and log counts at each stage in a PRISMA 2020-style flow (identified → screened → eligibility → included).
4. Screen titles/abstracts, then full texts, recording exclusion reasons.
5. Extract into a citation/evidence matrix: source, method, sample, key finding, limitations, relevance to question.
6. Synthesise thematically across the matrix, noting agreements, tensions, and gaps.
7. Mark every entry the author must independently verify; flag any source that cannot be located as unresolved.
**Output.** A PRISMA-style flow summary, a filled evidence matrix, and a themed synthesis with explicit gaps.
**Quality bar.** Every included source is real and openable by the author; no reference appears in the matrix that has not been retrieved; screening decisions are logged, not assumed.

## Capability 2 — Argument-structure suggestions
**Goal.** Give the manuscript a clear, defensible logical spine.
**Inputs.** The author's thesis or research aim, the evidence matrix, and the target paper type (empirical, review, position).
**Method.**
1. State the central claim in one sentence and test it for specificity and falsifiability.
2. For each supporting point, scaffold claim–evidence–warrant (Toulmin), making the reasoning link explicit and noting rebuttals and qualifiers.
3. Map the paper to IMRaD (Introduction, Methods, Results, and Discussion) or the venue's expected structure.
4. Draft topic sentences for each section and paragraph so the argument is readable from headings and openers alone.
5. Add signposting and transitions that carry the reader from claim to claim.
6. Check that every claim is backed by a matrix entry; mark unsupported claims for the author to source or cut.
**Output.** A structural outline with thesis, section-by-section topic sentences, and per-claim evidence links.
**Quality bar.** No claim is left without a traceable source or an explicit "needs support" flag; structure follows the venue's conventions, not a generic template.

## Capability 3 — Journal formatting & submission
**Goal.** Prepare a submission-ready package for a suitable, legitimate venue.
**Inputs.** The finished draft, the field, and the author's target audience and indexing needs.
**Method.**
1. Shortlist target journals by scope fit, indexing, and audience; screen for predatory venues using the Think-Check-Submit checklist and indexing/DOAJ/COPE-membership checks.
2. Confirm the venue's required reference style — APA 7, IEEE, or Vancouver — and align in-text citations and reference list exactly.
3. Apply the journal's structural, word-count, figure, and reporting-guideline requirements.
4. Draft a cover letter stating the contribution, fit, and any required declarations.
5. For revisions, draft a structured reviewer-response letter: restate each comment, state the change, and quote the revised text or explain a reasoned disagreement.
6. Confirm authorship, contributions, conflicts, and ethics statements against COPE guidance before submission.
**Output.** A style-compliant manuscript, a target-journal shortlist with screening notes, a cover letter, and (for revisions) a point-by-point response letter.
**Quality bar.** The chosen venue passes Think-Check-Submit screening; citation style is applied consistently; ethics and authorship declarations are the author's own confirmed statements.

## Worked example (illustrative)
*The following is an illustrative walkthrough, not a description of a real submitted paper.* An author wants a review on a design-methods topic. Capability 1 builds boolean strings for three databases, logs 214 identified → 38 screened → 12 included in a PRISMA-style flow, and fills an evidence matrix; two candidate sources cannot be retrieved, so they are marked unresolved rather than cited. Capability 2 sets a one-sentence thesis, scaffolds five claims in Toulmin form, maps them to IMRaD, and drafts topic sentences — one claim has no matrix support and is flagged "needs support." Capability 3 shortlists two journals passing Think-Check-Submit, aligns references to APA 7, and drafts a cover letter. The author then verifies every reference, supplies the missing evidence, and takes responsibility for the final text before submission.

## Guardrails & escalation
- STOP and escalate to a domain peer, supervisor, or institutional research-ethics board before submission when the work involves human subjects, data governance, contested claims, or any authorship/ethics question.
- Never publish or submit with a citation, quotation, or datum that the author has not personally verified; fabricated or unverified references must never be shipped.
- Uncertainty is flagged explicitly: unretrievable sources are marked unresolved, unsupported claims are marked "needs support," and no confidence is implied beyond the underlying evidence.

## References & standards
- PRISMA 2020 statement for systematic-review reporting and flow diagrams.
- Toulmin model of argument (claim, evidence/data, warrant, backing, qualifier, rebuttal).
- IMRaD structure (Introduction, Methods, Results, and Discussion).
- Citation styles: APA 7th edition, IEEE, and Vancouver.
- Think-Check-Submit checklist for venue selection and predatory-journal screening.
- COPE (Committee on Publication Ethics) guidance on authorship, conflicts, and research ethics.

---
*Part of Ed Chen's AI skill set — how one designer absorbs unfamiliar, C-level work quickly by pairing AI with rigor and professional review. https://edwson.com*