agentsclimarketplace

Academic writing

Skill Tanjim-Noor/academic-writing-skill/skills/academic-writing

Portable evidence-grounded academic writing Agent Skill for Codex and compatible agents.

Install
npx -y skills add Tanjim-Noor/academic-writing-skill --skill academic-writing

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

2 things to look at

  • no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Comprehensive, portable academic writing workflow for evidence-grounded planning, drafting, revising, and preparing scholarly outputs. Use for journal or conference papers, thesis chapters, literature reviews, theoretical papers, grants, abstracts, citation and integrity audits, reviewer responses, rebuttal audits, academic style, phrase selection, formatting, and submission preparation in the user's requested language.

The file declares its own license as See LICENSE.md and THIRD_PARTY_NOTICES.md. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

9.0 KB, as published. Nobody here has run it

Academic Writing

Produce defensible scholarly work through explicit evidence, argument, integrity, revision, and formatting gates. Follow the user's requested language and venue requirements. Use the English Academic Phrasebank only when the target prose is English.

Non-negotiable rules

  1. Never invent sources, bibliographic metadata, quotations, data, methods, results, policies, or review outcomes.
  2. Distinguish verified evidence, author-provided claims, inference, proposal, and placeholder text.
  3. Do not certify an artifact as verified or submission-ready while critical integrity issues remain.
  4. Verify time-sensitive venue, funder, open-access, disclosure, and submission rules against current primary sources.
  5. Preserve the author's meaning and evidence strength during editing.
  6. Establish content before selecting phrasebank language; never let a stock phrase create a claim.
  7. Ask only for information that materially changes the work and cannot be derived from supplied materials.

Select a mode

ModeUse whenPrimary output
fullBuild an output through all relevant phasesComplete scholarly package
planClarify question, contribution, evidence, and work sequenceWriting brief and phase plan
outlineDesign structure before draftingEvidence-mapped outline
draftDraft one or more sections from adequate evidenceSection draft with evidence markers
literature-reviewSearch, organize, and synthesize prior workReview protocol, matrix, and synthesis
abstractDraft or revise an abstractStructured or unstructured abstract and keywords
grantDevelop a research or funding proposalAims, narrative, impact, and budget rationale
reviseImprove a draft against objectives or feedbackRevision ledger and targeted revision
reviewer-responsePlan responses or audit an existing rebuttalComment ledger and response letter
citation-auditCheck claims, citations, and referencesTraceable integrity report
formatConvert or normalize document presentationVenue-aligned output package
submissionPrepare final administrative materialsSubmission checklist and supporting documents

Read modes-and-intake.md for mode gates and the complete Writing Brief.

Run the core workflow

1. Intake and classify

  • Identify deliverable, audience, discipline, language, venue, length, citation style, output format, deadline, supplied materials, and evidence status.
  • Determine whether the user wants guidance, generation, revision, audit, or formatting.
  • Record assumptions and unresolved decisions.

2. Establish research and evidence readiness

  • Refine the question, contribution, scope, and design.
  • Inventory supplied sources and separate verified from unverified material.
  • Build a search or evidence-completion plan when support is insufficient.
  • Read research-design-and-evidence.md.

3. Design structure and argument

4. Draft section by section

  • Draft from the approved outline and evidence map.
  • Mark unsupported points explicitly instead of filling gaps.
  • Maintain terminology, tense, voice, and contribution boundaries.
  • Keep results separate from interpretation unless the genre combines them.

5. Audit citations, integrity, style, and language

  • Reconcile every in-text citation with the reference list.
  • Trace important claims to adequate sources.
  • Check data, methods, results, quotations, authorship, disclosure, and originality.
  • Calibrate certainty and improve cohesion without changing claim strength.
  • Read citation-integrity-and-ethics.md and style-and-language.md.

6. Review and revise

  • Review contribution, logic, evidence, methods, presentation, and compliance independently.
  • Prioritize critical issues before stylistic preferences.
  • Use a comment-to-change ledger and targeted patches.
  • Stop after at most two substantive loops by default or earlier when no critical issue remains and improvement has converged.
  • Read review-revision-and-response.md.

7. Format and finalize

  • Apply the current venue template and citation style.
  • Validate tables, figures, equations, cross-references, accessibility, and rendered output.
  • Prepare cover letter, declarations, funding/authorship statements, data/code availability, and AI-use disclosure when required.
  • Read publication-and-formatting.md.

Read workflow-core.md for checkpoints, state, recovery, and completion statuses.

Use the phrasebank

  1. Identify the rhetorical function required.
  2. Search the index rather than loading the full bank:
python scripts/search_phrasebank.py --section discussing-findings --query limitation --limit 12
python scripts/search_phrasebank.py --function "being cautious" --entry-type phrase --limit 20
  1. Inspect source page and subheading.
  2. Adapt placeholders, grammar, discipline terms, and certainty.
  3. Audit repetition and confirm the phrase does not introduce unsupported content.

Read phrasebank-routing.md first. Then load only the relevant direct reference:

Use deliverable-specific guidance

  • Grants: read grants-and-proposals.md.
  • Abstracts, theses, reviews, theoretical papers, conferences, and policy briefs: read structures-and-deliverables.md.
  • Multilingual work: draft in the target language; use English phrasebank material only for English passages. Preserve technical terms when translation would reduce precision.
  • Templates: copy the appropriate file from assets/templates/ and adapt it to the current venue.

Completion contract

Return only statuses supported by evidence:

  • planned: brief, scope, and workflow agreed.
  • drafted: requested text exists but has not passed all audits.
  • audited-with-issues: audit complete; unresolved issues listed.
  • verified: required checks passed against available evidence.
  • submission-ready: verified, venue-formatted, rendered, and administratively complete.

Always list unresolved evidence gaps, placeholders, unverified policies, and user decisions separately from the finished prose.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.