agentleFS
epicsagasepicsagas/Velith/skills/book-academic/SKILL.md

book-academic

Skill40 starsChanged 2 months agoApache-2.0View on GitHub

Academic writing craft reference: IMRAD and discipline structures, literature review, argument chains, citation integrity, academic register in en/ko/ja, and academic-writing AI tells. Read by all agents on theses, dissertations, monographs, and research reports.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Every citation refers to a real work the author has (or the fact-checker has verified exists) with the correct authors, year, title, venue, …

    line 28

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Academic

The reader is a peer or an examiner. They will check the citations, and they will notice when the literature review summarizes instead of argues. Academic AI text fails by being fluent about things it has not read.

Structure by discipline

  • Sciences: IMRAD (Introduction → Methods → Results → Discussion), plus Abstract, References, Appendices.
  • Social sciences: Introduction → Literature review → Theoretical framework → Methods → Findings → Discussion → Conclusion.
  • Humanities: Thesis → evidence chapters (each an argument, not a topic) → synthesis → conclusion.
  • Thesis/dissertation: as above with committee and program requirements from PRD.md overriding defaults (chapter count, word limits, citation style, formatting).
  • Monograph: like humanities but with a stronger narrative spine and less methodological apparatus.

Literature review

Organize by theme and argument, never chronologically and never author-by-author. Establish the field → identify the gap → position this work. For each source: claim, method, limitation, relevance to this work. Build a synthesis matrix (themes × sources) in sources/INDEX.md before drafting. The review argues that the gap exists; it does not list what people have said.

Argument

Toulmin: claim → evidence → warrant → backing → qualifier → rebuttal. Every chapter thesis supports the central thesis, and the chapter's opening states its claim, position, and scope. Paragraph: topic sentence (claim) → evidence and analysis → link. Hedging is calibrated, not stacked: "may suggest" is fine; "it could perhaps be argued that it may" is not.

Citation integrity

  • Every citation refers to a real work the author has (or the fact-checker has verified exists) with the correct authors, year, title, venue, and page range. A fabricated or misattributed citation is a Critical defect; the fact-checker removes it.
  • Drafts cite inline with source IDs ([S12, p. 45]); copy edit converts to the house style (APA 7, MLA 9, Chicago 17, IEEE, or the program's own).
  • Direct quotes ≤ 40 words inline, longer as block quotes, always with page numbers.
  • Paraphrase must change structure, not only vocabulary.
  • Citation density: heavy in the review, light in methods and results, moderate in discussion.
  • Self-citation disclosed; retracted works not cited; preprints marked.

Register

  • English: third person or disciplinary convention; present tense for established knowledge and for discussing texts, past tense for what was done; defined terms used consistently; no rhetorical questions; no contractions.
  • Korean (학술 문체): 한다체 fixed; 본고/본 연구 for self-reference; 무생물 주어 restrained (연구 결과 ~로 나타났다 rather than 결과는 보여준다); 한자어 register consistent with the field; 각주 vs 미주 per program; 인용 표기 (저자, 연도) or 각주 per style.
  • Japanese: だ・である体; 本稿/本研究; 引用 per discipline.

Figures, tables, cross-references

Numbered per chapter (Figure 3.1). Captions below figures, above tables. Every figure and table referenced in the text before it appears. Self-contained. Cross-references by section number (Section 4.2), never "above" or "below."

Academic-writing tells

  • A literature review that lists studies in order.
  • "Scholars have long debated" without naming any.
  • Citations clustered at the end of a paragraph supporting a claim none of them makes.
  • Perfectly balanced discussion sections that find every finding "interesting."
  • Limitations section that lists generic limitations.
  • Abstract that describes the paper instead of stating its findings.
  • Hedging on findings and confidence on speculation.
  • Keyword-stuffed sentences that restate the title.

Abstract formula

Context (1-2 sentences) → gap (1) → purpose (1) → method (1-2) → key finding with a number (1-2) → significance (1). 150-300 words. Written last.

Validation (architect scoring for academic)

Central thesis stated; each chapter's thesis supports it; review organized by theme with a stated gap; methods sufficient to replicate; every core claim in the results has evidence in the results; discussion addresses the strongest alternative explanation; synthesis matrix exists; citation style declared; program constraints in PRD.md satisfied.

Visuals

Figures and tables follow the discipline's conventions and the program's format. Every figure is code-rendered from data in sources/ with the exact numbers in the results; captions are self-contained; numbering per chapter; referenced in the text before they appear. No decorative images. Conceptual frameworks as diagrams ≤ 9 elements with a consistent grammar. figure-engineer produces; fact-checker reconciles every plotted value with the data file.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.