research-manuscript-writing
Skill4 starsChanged 32 days agoMITView on GitHub
Develop an evidence-supported story from an initial research idea, then draft or restructure manuscript sections. Use for literature-informed first drafts or generic, list-like writing that misses the research question or section logic, with LaTeX or Word delivery. Not a substitute for missing study results or a default full rewrite for local edits.
In this repository17 other files it gives its agents.
- academic-prose-editingSkill
- academic-writing-auditSkill
- claim-evidence-auditSkill
- experiment-comparison-auditSkill
- figure-caption-auditSkill
- latex-bibliography-auditSkill
- manuscript-release-checkSkill
- method-definition-auditSkill
- nondefensive-writingSkill
- reference-authenticity-auditSkill
- related-work-synthesisSkill
- results-interpretationSkill
- reviewer-comment-triageSkill
- reviewer-response-writingSkill
- revision-integrity-auditSkill
- scientific-story-auditSkill
- word-citation-auditSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
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
- Distinctive: line 19
If a missing fact determines the scientific story, ask only for that fact or identify the dependency and continue supported work.
- Distinctive: line 27
Read section reasoning for any abstract, introduction, literature-review, or method work.
- Distinctive: line 32
Method: Follow dependencies from objects and assumptions through operations to their outputs and use.
- Distinctive: line 34
Read findings and endings for Results, Discussion, Limitations, or Conclusion.
- Distinctive: line 38
Read the prose policy when drafting or revising manuscript prose or audit explanations.
- Distinctive: line 50
Read LaTeX and Word delivery for a formatted draft or source edit.
≈ “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.
Files
Research Manuscript Writing
Understand the study before writing its story. Produce concise, connected prose in which each section advances the research question and each claim has an identifiable basis.
Establish the study before drafting
Read the current manuscript or research packet and identify the requested mode: audit, draft, or authorized file edit. Establish the research question, audience, study design, contribution, strongest supported finding, and its limits. Preserve sound passages and the author's terminology. A method name or fashionable topic cannot supply the study's rationale.
For an idea-led first draft or substantive reframing, read literature investigation and story development before fixing the story. This mode requires at least two separate agents to investigate overlapping decisive questions independently, then reciprocally review the source-supported findings before synthesis. Follow its discovery, prior-attempt, critique, and handoff procedure. Reuse an adequate research record and update only material gaps. If agents or sources are unavailable or outside scope, provide a bounded single-agent or supplied-source account and identify the incomplete checks. Local wording edits do not trigger this workflow.
Before drafting an introduction, establish the venue, track or article type, readership, and relevant submission priorities. Consult current official guidance when a venue-specific expectation needs verification. Distinguish supplied or verified official criteria from inferred audience needs. If the missing target would change the framing, ask while organizing supported material; a requested venue-neutral draft can proceed with that scope stated outside the manuscript. Venue fit changes emphasis, not facts or claim strength.
If a missing fact determines the scientific story, ask only for that fact or identify the dependency and continue supported work. Do not invent motivation, novelty, citations, results, or a missing source capability. A source's silence means unknown, not absent.
Build one connected argument
Form a short internal account of why the problem matters, what existing work enables, what remains unresolved, why the proposed scientific move addresses it, and what this study establishes. Test the strongest supported account against consequential alternatives at the study's actual stage. An unknown outcome may motivate investigation; an unsupported asserted result needs narrowing or withholding. A weakness in one claim does not automatically defeat the study.
Let the evidence revise the initial idea. Keep exploratory debate, search diligence, agent disagreements, and unresolved research choices in preparation notes. The manuscript presents the supported reasoning and consequential qualifications, not an investigation diary or anticipated rebuttal.
Read section reasoning for any abstract, introduction, literature-review, or method work. Use it to develop each requested section's distinct job:
- Abstract: Normally use at most six sentences, connecting task need, unresolved difficulty and contribution, then ending in the principal supported result. Preserve the reasoning while removing secondary detail; author and required venue formats take precedence.
- Introduction: Choose the opening perspective and its path to the research question before drafting. Develop one narrowing argument from relevant progress through gap and motivation to the supported contribution.
- Literature review: Continue that question through a connected account of prior work in its own settings. Do not prescribe a number of directions, list method families, or default to comparisons with the present paper.
- Method: Follow dependencies from objects and assumptions through operations to their outputs and use; do not repeat every representation verbatim.
Read findings and endings for Results, Discussion, Limitations, or Conclusion. Adapt the reasoning to the discipline, study design, and supplied format instead of imposing section headings, universal sentence counts, or a benchmark narrative.
Keep the prose focused
Read the prose policy when drafting or revising manuscript prose or audit explanations. Preserve scientific meaning, valid terminology, necessary conditions, and sound author choices. Neither the manuscript nor the audit may add an unreported property or mechanism.
Ask what each paragraph establishes and why the next follows. Remove generic importance claims, empty transitions, repeated announcements, and assurances that serve no reader purpose. Retain the shortest premise or definition needed for the next inference. Let emphasis and variation follow the reasoning; do not invent author beliefs, failed attempts, surprise, or design history to make prose feel natural.
Check the revision against its purpose
Compare supplied earlier and revised passages for both concern resolution and preservation of valid reasoning, definitions, conditions, evidence links, and interpretation. Match the standard to the design and distinguish scientific validity from expression. Keep supported corrections and useful compression; repair only actual losses. Without the earlier version, do not claim preservation was checked.
Deliver in the requested medium
When adding, returning or validating BibTeX, use each intended publication's actual Google Scholar Cite → BibTeX entry verbatim, with its exported key. Retain its provenance and check identity against primary records through the two independent verifiers. Missing export or unresolved identity leaves the reference pending outside the prose. Preserve existing citation markers during a prose-only edit. Never synthesize BibTeX from source cards, DOI metadata or memory.
Read LaTeX and Word delivery for a formatted draft or source edit. Maintain the existing template, citation system, and cross-references. Prose quality and successful document rendering are separate checks. Never claim a compiled PDF or validated Word layout from text inspection alone.
Return the requested section or document, material unresolved facts, and brief reasons for substantive changes. Keep reviewer-response status and negotiation language outside manuscript exposition. Finish by tracing the question through motivation, contribution, evidence, and conclusion; each section should advance its own job without replaying the whole story. Stop when the requested draft or repair is supported and checked.
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
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.

