agentleFS
bimajanuribimajanuri/academic-writing-skill/SKILL.md

academic-writing

Skill8 starsChanged 18 days agoMITView on GitHub

Plans, synthesizes, writes, and revises English-US academic manuscripts across systematic, scoping, meta-ethnographic, narrative, integrative, and critical review methodologies. Use when conducting literature reviews, building extraction and gap matrices, outlining or drafting paper sections, formatting citations, refining academic prose, targeting journals, or responding to peer reviewers.

28 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use scipilot-figure-skill and grill-me.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • HaojaeSkill

    scipilot-figure-skill

    SciPilot Skills 家族成员,负责科研数据可视化——但定位不是"画图工具", 而是"可视化顾问"。先做数据剖析(列类型/样本量/分布/异常值/分组结构/相关性), 再结合用户的论证目标推荐图型,主动拦截科研画图的经典错误(小样本画均值柱掩盖 分布、双 Y 轴、饼图、Y 轴不当截断、rainbow 色图、把分类点连成折线等),最后 产出 Nature / Science / IEEE / Elsevier / PNAS / 中文核心期刊级别的成图。 覆盖纯数据图:折线、柱状、散点、箱线 / 小提琴、热力图、误差棒、分布图(直方 图 / KDE)、相关性矩阵 / 散点矩阵、多面板组合。技术栈 matplotlib + seaborn + SciencePlots(静态)+ plotly(交互)。中英文双语,中文模式自动配置 Noto Sans CJK / Source Han Sans / SimHei 并修复负号方框,支持中文期刊"宋体正文 + Times New Roman 数字"混排。默认色盲安全配色 + 冗余编码 + 灰度预览。出图后做 "视觉自检闭环":渲染 PNG 预览→程序自检缺字/文字裁切/刻度重叠→AI 读图复核遮盖 与子图对齐→回改重渲,直到通过。 当用户的任务涉及以下任何情况时主动触发:论文配图、科研画图、数据可视化、 不知道用什么图、怎么展示数据、用什么图好、期刊投稿图、figure、出版级图表、 matplotlib、seaborn、plotly、误差棒、显著性标注、色盲安全配色、矢量图导出、 中文论文图表、多面板。**即使用户只是给一批数据问"这个怎么画"或"用什么图 好",也应使用本技能——本技能首要能力是"判断该用什么图",其次才是绘制。** 不做示意图、流程图、架构图。

    Also in 3 projects that use this

  • GoogleSkill

    grill-me

    A relentless interview to sharpen a plan or design.

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

7 rules. 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 20 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Search OpenAlex first, followed by Semantic Scholar, Crossref, DOAJ, PubMed, or arXiv.

    line 62
  2. Very specific:

    Check the mechanical gate (~21-word mean sentence length, max 40 words, active voice, evidence-calibrated modality, no em-dashes, …

    line 78
  3. Very specific:

    Run protocols/guardrails-and-evaluation.md before final output:

    line 79
  4. Very specific:

    Missing empirical statistics or study data: Never invent numbers, p-values, or outcomes.

    line 95
  5. Distinctive:

    Never invent references, DOIs, quartiles, data, quotations, or study findings.

    line 103
  6. Very specific:

    Keep papers.json and literature_matrix.md as the verified evidence base with explicit provenance markers.

    line 104
  7. Distinctive:

    Write all manuscript sections in English-US academic style following the required drafting sequence.

    line 105

“Common” is measured: kept by at least 6 in 10 of the 20 edited versions of this file we could read. “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 Writing

This skill provides a modular, multi-method pipeline for literature synthesis and academic manuscript production. All instructions in this file and its referenced modules are in English. Research input may be provided in Bahasa Indonesia or English, while the final manuscript output is always written in English (US) academic register.

Usage

The skill activates automatically for academic writing, literature synthesis, research gap analysis, manuscript drafting, citation formatting, prose refinement, journal targeting, and reviewer-response requests. Users may also invoke it explicitly by name:

Use academic-writing to plan and write a scoping review on [topic].
First lock Gate 0 (Scoping mode, PCC framework, and PRISMA-ScR reporting standard) before searching.
Preserve provenance, verify DOI metadata, build papers.json, and validate before delivery.

Select one of the six synthesis modes (SLR, Scoping Review, Meta-Ethnography, Narrative Review, Integrative Review, or Critical Review) before running literature acquisition or extraction. If the synthesis mode, scope, quartile filter, or citation style is ambiguous, ask once before searching.

Load order

  1. Load this SKILL.md.
  2. Load protocols/core-and-mode-protocols.md for universal guidelines, the selected synthesis mode protocol, and the assigned qualitative synthesis engine.
  3. Load references/stage1-explore-and-extraction.md for paper acquisition, papers.json provenance, mode extraction matrices, and research gap mapping.
  4. Load references/stage2-to-4-plan-write-cite.md for research frameworks, outlining, novelty framing, journal targeting, section drafting, and citation formatting.
  5. Load references/stage5-to-7-revise-output-post.md for prose style refinement, English-US grammar and text-overlap checks, quality gates, format conversion, and reviewer responses.
  6. Load templates/workflow-and-section-templates.md and templates/citation-style-templates.md for output contracts and citation styles.
  7. Load protocols/guardrails-and-evaluation.md and protocols/validation-and-reporting-checklists.md before final delivery.

Method selection router (Gate 0)

Choose and lock the analysis mode before searching for or extracting any paper:

ModeUse whenRQ frameworkReporting standardExtraction matrixSynthesis engine
SLRQuantitative or qualitative systematic review, meta-analysis, evidence of effectPICO, PICOS, PICo, or SPIDERPRISMA 2020Standard 7-field plus shared columnsMeta-analysis or thematic synthesis
Scoping ReviewBroad mapping of evidence, concepts, geographic scope, and methodology typesPCCPRISMA-ScR (JBI)PCC mapping matrixConceptual Mapping
Meta-EthnographyQualitative meta-synthesis, lived experience, and theory developmentPICo or SPIDEReMERGe and Noblit-Hare 7 steps1st- and 2nd-order constructsReciprocal Translation
Narrative ReviewNarrative overview, academic debate, and conceptual development5W+1H or CIMOSANRAArgument position matrixAcademic Debate
Integrative ReviewCombined qualitative and quantitative empirical or theoretical literatureAdapted frameworkWhittemore and KnaflMixed qualitative and quantitative matrix5-stage data reduction and comparison
Critical ReviewEpistemological, methodological, or bias critique across studies5W+1HCASP or JBI appraisalAppraisal lens matrixPer-study appraisal and ranking synthesis

Record the locked mode, reporting standard, RQ framework, publication-year range, and Scopus Quartile target in framework_selection.md before building search queries.

Pipeline & workflow checklist

  • Gate 0 — Method selection: Lock synthesis mode, reporting standard, RQ framework, quartile filter, and year range in framework_selection.md (protocols/core-and-mode-protocols.md).
  • Stage 1 — Explore: Run mode-calibrated acquisition (Search, Folder, or Hybrid), deduplicate and verify DOIs, build papers.json and literature_matrix.md with provenance, and generate gap_matrix.md and research_map.md (references/stage1-explore-and-extraction.md).
  • Stage 2 — Plan: Formulate FINER research questions, compare target journals, build paper_outline.md, score >= 28/35 on the 7-dimension reviewer rubric, and write contribution_statement.md (references/stage2-to-4-plan-write-cite.md).
  • Stage 3 — Write: Draft sections in sequence (Draft 0 Intro, Literature Review, Methodology, Results and Discussion from the synthesis engine, Conclusion, rewritten Final Intro, Abstract last) with claim_evidence_map.md (references/stage2-to-4-plan-write-cite.md).
  • Stage 4 — Cite: Format in-text citations and bibliography (APA 7, MLA 9, Turabian, Vancouver, Harvard, Chicago, or IEEE) and verify 1:1 parity and DOIs (templates/citation-style-templates.md).
  • Stage 5 — Revise: Run the 26 prose style refinement patterns, English-US grammar check, mechanical and semantic gates, and independent peer-review audit (references/stage5-to-7-revise-output-post.md).
  • Stage 6 — Validate: Run output guardrails, quality rubric evaluation, and claim/citation regression checks (protocols/guardrails-and-evaluation.md).
  • Stage 7 — Output: Convert to .md, .tex, or .docx, complete the mode reporting checklist in reporting_checklist_result.md, verify citation integrity, and prepare cover_letter.md (protocols/validation-and-reporting-checklists.md).

Stage 1 — Explore and extract

  1. Confirm input mode (Search, Folder, or Hybrid), synthesis mode, topic, Scopus Quartile target (Q1–Q4 or unfiltered), and year range (default: last 5 years).
  2. Derive search queries from the mode RQ framework elements and calibrate candidate selection to the mode.
  3. Search OpenAlex first, followed by Semantic Scholar, Crossref, DOAJ, PubMed, or arXiv; for Folder or Hybrid mode, read local .pdf and .tex files and prefer full text over abstract metadata.
  4. Collect ~1.5x the target count (typical target: 15–40 papers scoring >= 7/10 relevance), deduplicate by DOI and normalized title, and verify bibliographic metadata. Mark unresolved records UNVERIFIED — check manually.
  5. Build papers.json as the single source of truth. Tag every extracted factual statement with [AB] (abstract-only) or full-text section markers ([I-1], [M-2], [R-1], [D-2], [L-1], [C-1], [F-1]).
  6. Populate the shared columns (Title & Authors, Journal, Year, Purpose, Method, Key Findings, Limitations, Gaps, Theory Used, Novelty/Contribution, Future Studies, DOI & Publisher) plus the mode-specific columns in references/stage1-explore-and-extraction.md, then render literature_matrix.md.
  7. Synthesize gap_matrix.md (>= 3 citations per gap; 0 pseudo-gaps) and research_map.md.

Stage 2 to Stage 4 — Plan, write, and cite

  1. Formulate research questions with FINER criteria (research_question.md), compare target journals (journal_comparison.md), and build paper_outline.md (Intro 15–20%, Body 60–70%, Conclusion 10–15%) with topic sentences first.
  2. Score the outline with the 7-dimension reviewer rubric (require >= 28/35), map reporting checklist items (framework_compliance.md), and lock contribution_statement.md.
  3. Draft in sequence: Draft 0 Intro -> Literature Review (thematic/construct synthesis) -> Methodology -> Results & Discussion (synthesis engine output) -> Conclusion -> rewritten Final Intro -> Abstract (150–250 words, 3–6 keywords).
  4. Maintain claim_evidence_map.md and format all citations and bibliography.md per templates/citation-style-templates.md with 1:1 parity (citation_validation.md).

Stage 5 to Stage 7 — Revise, validate, and export

  1. Refine prose using the 26 stylistic patterns (style_refinement_report.md) and verify English-US grammar and spelling (grammar_report.md).
  2. Check the mechanical gate (~21-word mean sentence length, max 40 words, active voice, evidence-calibrated modality, no em-dashes, claims-based headings), semantic gate, and independent peer-review audit (peer_review_audit_findings.md).
  3. Run protocols/guardrails-and-evaluation.md before final output:
claims_checked: N
citation_mismatches: N
unverified_sources: N
guardrail_violations: N
rubric_mean_score: X.X / 4.0
status: PASS | PASS_WITH_FLAGS | FAIL
  1. Complete the mode reporting checklist from protocols/validation-and-reporting-checklists.md in reporting_checklist_result.md, verify citation integrity (overlap_and_citation_report.md), convert to .md, .tex, or .docx, and prepare cover_letter.md or response_to_reviewer.md.

Handling missing or ambiguous information

  • Unspecified synthesis mode or RQ framework: Pause before searching (Gate 0) and ask the user to select or confirm one of the 6 modes.
  • Missing empirical statistics or study data: Never invent numbers, p-values, or outcomes. Insert [DATA] and ask the user to supply the missing values.
  • Unavailable field in a source paper: Mark the extraction cell with —.
  • Abstract-only access: Record only claims stated in the abstract with [AB] and mark full-text-only columns as —.
  • Unresolved DOI or metadata: Preserve available source text and mark UNVERIFIED — check manually.

Core guidelines

  1. Lock the synthesis mode, reporting standard, and RQ framework (Gate 0) before literature search or extraction.
  2. Never invent references, DOIs, quartiles, data, quotations, or study findings.
  3. Keep papers.json and literature_matrix.md as the verified evidence base with explicit provenance markers.
  4. Write all manuscript sections in English-US academic style following the required drafting sequence.
  5. Pass human-in-the-loop quality gates, output guardrails, and the mode reporting checklist before final delivery.

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.