Skip to content
Research Agent Skillsby Kalaris Labs
Install
Research Writingv2.1AdaptedMIT

scientific-writing

Draft, revise, and audit scientific manuscripts or reports with explicit evidence provenance, reporting-guideline coverage, authorship accountability, confidentiality controls, and local consistency checks. Use for manuscript sections, references, declarations, tables, figures, or submission preparation when scientific accuracy and traceability matter.

Install this skill

Works with Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot, OpenCode, and 70+ other harnesses.

skills.sh CLI (Recommended)
npx skills add KalarisLabs/research-agent-skills --skill scientific-writing
uvx (Python tool runner)
uvx research-agent-skills install scientific-writing

Documentation & Instructions

Scientific Writing

Purpose

Produce clear scientific prose without inventing evidence or concealing uncertainty. Keep drafting, evidence verification, and submission approval as separate stages.

The accountable human authors control scientific decisions and final approval. AI is not an author, and generated fluency is never evidence [SW-S01, SW-S03].

Non-negotiable safety rules

Confidentiality

Do not send unpublished manuscripts, peer-review or editorial material, sensitive or restricted data, PHI or other personal data, proprietary content, or source documents to an external service without:

  1. explicit authorization from a person or body empowered to grant it; and
  2. a documented review of journal, institutional, funder, consent, ethics, contractual, legal, and data-use policy.

When authorization or policy is unclear, keep processing local and use only the minimum metadata needed. De-identification requires expert review; removing obvious names is not sufficient. See references/authorship_ai_confidentiality.md.

No fabrication

Never invent or complete:

  • citations, references, DOI, PMID, PMCID, ISBN, URLs, or quotations;
  • results, data values, denominators, sample sizes, units, effect estimates, uncertainty, statistical tests, or significance claims;
  • methods, materials, protocol details, software versions, analysis choices, or deviations;
  • registrations, approvals, consent, ethics statements, participant details, or dates;
  • authors, author order, CRediT roles, acknowledgments, or permissions;
  • funding, sponsor roles, conflicts, data or code availability, or AI disclosures.

Use an explicit missing, unverified, or not-applicable state. Do not substitute plausible boilerplate.

Evidence binding

Every factual or numeric manuscript claim must map to verified evidence IDs. A human verifier must open the source, confirm the proposition and locator, verify bibliographic metadata, and record who verified it and when.

Search snippets, generated summaries, memory, and another work's bibliography may aid discovery but do not verify a claim. See references/evidence_workflow.md.

Scientific fidelity

  • Preserve uncertainty and alternative explanations.
  • Distinguish confirmatory, exploratory, descriptive, and post hoc work.
  • Keep methods and results consistent.
  • Reconcile units, denominators, sample sizes, populations, time points, and labels.
  • Report negative, null, adverse, unexpected, failed, and inconclusive findings when they belong to the study record.
  • State concrete limitations and bound generalizability.
  • Do not convert association into causation or non-significance into equivalence.

Intake

Before drafting, obtain or mark unresolved:

  • document type, study design, stage, audience, and target venue;
  • current author instructions and policy access date;
  • protocol, registration, analysis plan, amendments, and reporting guideline;
  • manuscript or section scope;
  • verified source manifest and claim registry;
  • methods, results, tables, figures, and supplements;
  • authorship, CRediT, declarations, and approval records;
  • confidentiality classification and authorized processing boundary;
  • data, code, materials, and repository constraints.

Do not ask for restricted source material if metadata or a local user-run audit is sufficient.

Workflow

1. Establish the local workspace

For a new draft, optionally generate fail-closed Markdown, JSON, and CSV scaffolds:

python3 scripts/scaffold_manuscript.py \
  --output-dir ./draft-workspace \
  --document-id local-draft \
  --study-design randomized_trial \
  --guideline consort-2025

The generator never overwrites files. Its output is explicitly not submission-ready and contains placeholders that the linter rejects.

2. Select reporting guidance

Choose by actual design and article type, then open the current official statement, checklist, explanation document, extensions, and target-journal instructions.

python3 scripts/select_reporting_guidelines.py select \
  --study-design randomized_trial

Current major routes researched on 2026-07-24 include CONSORT 2025, SPIRIT 2025, PRISMA 2020, STROBE, STARD and STARD-AI, TRIPOD+AI, CARE, ARRIVE 2.0, SQUIRE 2.0, and CHEERS 2022 [SW-S06–SW-S18].

The selector is non-scoring. It does not certify quality, compliance, completeness, or acceptance. See references/reporting_guidelines.md.

3. Build the evidence record

Assign:

  • E IDs to sources in source_manifest.json;
  • C IDs to claims in claims.csv;
  • N, M, O, and R IDs to numeric facts, methods, outcomes, and results in consistency_manifest.json.

Store a hash of claim text in CSV rather than raw claim text. During drafting, append:

[claim:C001] [evidence:E001,E002]

Do not mark a source verified until an accountable human has opened it and confirmed the exact support.

4. Create an evidence outline

Outline only from recorded evidence:

  • objective or question;
  • section purpose;
  • claim IDs and evidence IDs;
  • methods and result IDs;
  • analysis intent and uncertainty;
  • unresolved conflicts or missing information;
  • applicable reporting topics.

Keep unsupported content in an unresolved-issues list, not manuscript prose.

5. Draft without adding facts

Transform the verified outline into venue-appropriate prose. Preserve all IDs during drafting.

  • Match title and abstract to the completed main text.
  • Describe methods as performed.
  • Present results in the declared order and analysis population.
  • Separate result from interpretation unless the venue combines them.
  • Compare with prior evidence only after verifying it.
  • Keep conclusions within the observed design, population, and uncertainty.

Use IMRAD only when appropriate. Structured abstracts, lists, combined sections, and alternative structures depend on study design and venue. See references/imrad_structure.md and references/writing_principles.md.

6. Reconcile methods and results

Record repeated numeric facts and method-result mappings, then run:

python3 scripts/check_consistency.py consistency_manifest.json

Resolve every mismatch manually. A changed value may be a legitimate analysis-set difference, but that difference must be named rather than silently normalized.

7. Verify citations and claims

python3 scripts/validate_manifest.py source_manifest.json \
  --kind source --require-verified
python3 scripts/audit_claims.py manuscript.md claims.csv source_manifest.json
python3 scripts/check_references.py source_manifest.json

The reference checker validates syntax and duplicate identifiers without network resolution. A human must still compare every identifier and quotation with the opened source. Follow NLM Citing Medicine or the current official style required by the venue [SW-S20, SW-S21].

8. Validate authorship and disclosure

Use journal criteria for authorship. Record the standardized CRediT roles as contribution metadata; CRediT does not itself define authorship [SW-S19].

If AI was used, humans must verify all affected content and disclose the tool and purpose according to current journal and publisher policy. ICMJE's January 2026 Recommendations require transparency and retain human accountability [SW-S01, SW-S02].

python3 scripts/validate_authorship.py authorship.json

Do not generate a disclosure from assumptions. See references/authorship_ai_confidentiality.md.

9. Review declarations and open-science statements

Verify each statement independently:

  • ethics and consent;
  • registration and protocol;
  • funding and sponsor role;
  • conflicts and relationships;
  • author contributions and acknowledgments;
  • data, code, materials, and protocol availability;
  • AI use.

Be as open as rights and responsibilities permit, but do not expose confidential, personal, proprietary, licensed, or protected information. Record actual access conditions. See references/research_integrity_open_science.md.

10. Use figures and tables only when warranted

Figures and tables are optional and provenance-bound. This skill does not generate images or schematics.

For every retained display:

  • link source data, code, transformations, and evidence IDs;
  • reconcile values with prose and registries;
  • document image processing, permissions, and licenses;
  • include units, denominators, sample sizes, uncertainty, and analysis population;
  • provide alt text and redundant non-color cues;
  • perform a manual accessibility and scientific check at final size.

See references/figures_tables.md.

11. Record non-scoring guideline coverage

Record each bundled high-level topic as addressed, not applicable with rationale, or missing:

python3 scripts/select_reporting_guidelines.py check reporting_coverage.json

Then complete the official checklist using actual manuscript locations. Never claim adherence merely because the local coverage file passes.

12. Lint and approve

python3 scripts/validate_manifest.py manuscript_manifest.json --kind manuscript
python3 scripts/lint_manuscript.py manuscript.md \
  --manifest manuscript_manifest.json

The linter reports issue codes and line numbers without echoing manuscript text. Sensitive-content warnings require manual review and are not a de-identification certificate.

Only accountable humans may:

  • resolve scientific ambiguities;
  • approve author order and declarations;
  • approve external disclosure or transfer;
  • set submission_ready to true;
  • remove the draft banner;
  • authorize submission.

Revision and peer review

Treat reviewer material as confidential. Do not upload it to an external service without the required authorization and policy review [SW-S01, SW-S24].

For each requested change:

  1. record the comment without exposing it outside the approved boundary;
  2. classify it as editorial, scientific, statistical, policy, or unresolved;
  3. identify affected claims, evidence, methods, results, and displays;
  4. revise the registries before prose when facts change;
  5. re-run every affected audit;
  6. draft a response that states what changed and where;
  7. obtain human approval.

Do not comply with a request that would fabricate, hide, overstate, or breach policy.

Current policy caution

COPE's 2017 Core Practices were retired in 2024. As of 2026-07-24, COPE announced that a replacement Code of Conduct would be published in 2026; do not describe the archived Core Practices as current membership standards [SW-S04, SW-S05]. Distinguish formal COPE positions from discussion documents, webinars, comments, and case advice.

Formatting and submission

The former LaTeX assets were removed because a generic polished template could allow plausible placeholders to ship. Use the Markdown scaffold and structured records. Apply the target venue's current controlled template only after verification.

See:

  • assets/REPORT_FORMATTING_GUIDE.md
  • references/professional_report_formatting.md
  • references/journal_policies.md

Formatting cannot convert an incomplete evidence record into a submission-ready paper.

Bundled files

Assets

  • assets/manuscript_scaffold.md
  • assets/manuscript_manifest_template.json
  • assets/source_manifest_template.json
  • assets/claim_evidence_template.csv
  • assets/consistency_manifest_template.json
  • assets/authorship_template.json
  • assets/reporting_coverage_template.json
  • assets/reporting_guidelines.json

Scripts

  • scripts/scaffold_manuscript.py
  • scripts/validate_manifest.py
  • scripts/select_reporting_guidelines.py
  • scripts/audit_claims.py
  • scripts/check_consistency.py
  • scripts/check_references.py
  • scripts/validate_authorship.py
  • scripts/lint_manuscript.py

All scripts are local, deterministic, bounded, dependency-free, and network-free. See references/cli_reference.md.

References

  • references/evidence_workflow.md
  • references/writing_principles.md
  • references/imrad_structure.md
  • references/citation_styles.md
  • references/reporting_guidelines.md
  • references/figures_tables.md
  • references/authorship_ai_confidentiality.md
  • references/research_integrity_open_science.md
  • references/journal_policies.md
  • references/professional_report_formatting.md
  • references/cli_reference.md
  • references/source_ledger.md

Agent operating procedure

  1. Check the environment. Collect the user's draft, data, figures and target venue before writing anything.
  2. Pin down the inputs. Confirm formats, identifiers and parameters from the data or the user. Ask rather than guess any value that changes the result.
  3. Run a small version first. Draft one section or paragraph and confirm tone, structure and terminology with the user.
  4. Execute the full task using the instructions and references above.
  5. Validate the result. Every claim traces to the user's data or a verified source; numbers match tables and figures; citations are real (citation-verification).
  6. Report. State what was run (versions, commands, parameters), what was checked, and what is still uncertain.
If this happens Do this
A claim has no supporting data or citation Flag it in the text (e.g. [citation needed]) instead of inventing support.
A function, flag or endpoint in these instructions is missing in the installed version Check the installed version's own documentation (help(), --help, official docs), adapt, and tell the user. Never invent an API.
A required input, identifier or parameter is ambiguous Ask the user, or state the assumption explicitly before running.

Integrity rules

  • Never fabricate results, parameters, identifiers, citations or statistics. If something cannot be run or verified, say so plainly.
  • The user is the author: never add results, references or claims they did not provide or approve.
  • Treat version-specific details here as possibly outdated: confirm them against the official documentation for the installed version.
  • Ask before actions that cost money, consume shared GPUs or cloud quota, touch personal or patient data, or cannot be undone.

Related skills

  • peer-review: Prepare evidence-bounded, constructive peer-review drafts and structured manuscript assessments.
  • clinical-reports: Generates fail-closed draft JSON templates and runs local deterministic structure and consistency checks for clinical reports: CARE case re…
  • nature-portfolio: Prepare manuscripts for Nature and Nature Portfolio journals (Nature, Nature Communications, Nature Methods, Nature Biotechnology, Scientif…

Trigger Prompts & Evaluation Corpus

These evaluation prompts test that the agent correctly selects scientific-writing for relevant tasks and leaves it out for near misses.

Should trigger (3)

  • should trigger

    “Help me write the Methods section for my RCT manuscript without making stuff up. I'll give you the protocol.”

    expects Starts with intake (design, venue, guideline such as CONSORT 2025, protocol and amendments), drafts only from the verified evidence/claim registry, marks missing details as unverified instead of inventing them, and keeps claim/evidence IDs.

  • should trigger

    “I have draft.md, claims.csv and source_manifest.json for my cohort study on statin use and dementia (target journal: JAMA Network Open). The abstract says n=4,212 but the results table says 4,128. Can you check consistency, run the local lint, and get the references and declarations ready for submission?”

    expects Runs or guides the local consistency manifest check and linter, flags the sample size mismatch for manual reconciliation (naming any analysis-set difference rather than silently normalizing it), validates reference syntax without inventing identifiers, notes STROBE coverage, leaves declarations and submission_ready for the accountable human authors, and avoids uploading unpublished material externally.

  • should trigger

    “My manuscript on a new soil microbiome sequencing method is almost done. I want to make sure every number and citation in it can be traced back to a real source, the funding and conflict statements are accurate, and that I'm not leaking anything confidential when I use an AI tool. What's the process?”

    expects Describes the staged workflow: build a source manifest and claim registry with human verification of each source, reconcile numeric facts, verify declarations independently, check journal AI-disclosure and confidentiality policy before any external transfer, keep AI as non-author, and have humans approve final submission.

Near misses (2)

  • near miss

    “I've been invited to review a submitted manuscript on a protein-folding model. Here's the PDF; write a structured referee report with strengths, weaknesses and statistical concerns.”

    expects Recognizes this as an authorized peer-review assessment task, which belongs to the peer-review skill rather than manuscript drafting, and notes the confidentiality/authorization caution for reviewer material.

  • near miss

    “Generate a CARE-format case report template JSON for a rare pathology case and run a structure check on it.”

    expects Uses fail-closed clinical report templates and deterministic structure checks for a CARE case report, which is the clinical-reports skill's domain.