Agent Skills
› brycewang-stanford/Awesome-Journal-Skills
› lancet-submission
lancet-submission
GitHub柳叶刀投稿前最终检查技能，涵盖重要性、注册、报告规范、统计、摘要及伦理声明等完整清单。确保符合期刊要求，防止遗漏或退步，并提供模板辅助。
Trigger Scenarios
稿件完成准备上传时
需确认其他柳叶刀相关技能输出无误时
修改稿重新提交需验证无回退时
Install
npx skills add brycewang-stanford/Awesome-Journal-Skills --skill lancet-submission -g -y
SKILL.md
Frontmatter
{
"name": "lancet-submission",
"description": "Use as the final preflight before submitting to The Lancet — a complete checklist across importance, registration, reporting guideline + flow diagram, statistics, structured abstract, the Research in context panel, ethics\/declarations\/data-sharing, and required files. Bundles a checklist and a cover-letter template."
}
Submission Preflight (lancet-submission)
When to trigger
- The manuscript is "done" and you are about to upload.
- You want a single gate that confirms every other lancet-* skill's output landed.
- A revision is going back and you need to confirm nothing regressed.
Master preflight checklist
Importance & framing
- Clears the importance / global-relevance / practice-or-policy bar (
lancet-fitrung ≥ 3). - Equity/global-health angle articulated where relevant.
- Title declarative and specific; reflects design (e.g., "a randomised controlled trial").
Registration & design
- Trial registered prospectively; registration number in the abstract and Methods.
- Protocol + pre-specified SAP available; primary outcome matches the registered one.
- RCT: randomisation, allocation concealment, blinding, ITT — described.
Reporting guideline & flow diagram
- Correct EQUATOR guideline (CONSORT / STROBE / PRISMA / SPIRIT / ...) with completed checklist, page/line-mapped.
- Flow diagram present (CONSORT for RCT, PRISMA for review); numbers reconcile with Table 1 and analysis populations.
Statistics
- Effect estimates with 95% CI; exact P; absolute + relative effects (NNT where relevant).
- Pre-specified primary analysis; ITT primary + per-protocol sensitivity; missing-data handling.
- Multiplicity strategy for secondary endpoints; subgroups pre-specified + interaction tests.
Structured abstract
- ≤300 words under Background, Methods, Findings, Interpretation, Funding (exact headings).
- Registration number in Methods; primary outcome with effect size + 95% CI; funder named.
Research in context panel
- All three parts present (Evidence before / Added value / Implications).
- "Evidence before this study" documents a systematic search (databases + terms + dates).
Figures & tables
- Table 1 baseline characteristics (no baseline P in an RCT).
- Kaplan–Meier with numbers at risk / forest plot / map — as the design warrants; CIs shown.
Ethics, declarations & data
- Ethics approval + consent (all sites) + Declaration of Helsinki.
- Declaration of interests for all authors (ICMJE).
- Role of the funding source statement (funder role + data access + final responsibility to submit).
- Data sharing statement (ICMJE-compliant for trials).
- ICMJE author contributions (incl. who verified the data); SAGER sex/gender + equity reporting; PPI statement.
References & required files
- References within budget (~30 for an Article), correct style; all in-text citations resolve.
- Main text (IMRaD + panel), abstract, tables, figures + legends, appendix/supplement.
- Reporting checklist; protocol/SAP; cover letter; authors/affiliations/ORCIDs; corresponding author.
Final integrity sweep
- No over-claiming: causal language matches design; no superiority claim from a non-inferiority trial.
- All abstract / flow-diagram / Table 1 / text numbers reconcile.
- Registration number, ethics references, and any data-repository identifiers are correct and live.
- Headings are Findings / Interpretation (not Results / Conclusions).
Templates
templates/checklist.md— copyable preflight checklist.templates/cover_letter_template.md— clinical/global-importance cover-letter scaffold.
Desk-reject patterns to clear before you upload
The Lancet declines most submissions without external review, so the preflight is really a pre-triage simulation. The editor reads the cover letter and abstract for one thing first: does this change clinical practice or public-health policy for a globally relevant problem. Then the submission system enforces the artifacts. Clear these before upload, because each is a fast bounce-back.
| Pre-triage failure | Where it surfaces | Pre-upload fix |
|---|---|---|
| Not practice/policy-changing for a broad, global audience | Cover letter + abstract Interpretation | Re-run lancet-fit; if rung < 3, reroute to a family title |
| Trial not prospectively registered | Methods + abstract | Disclose timing honestly; if retrospective, flag it explicitly |
| Wrong abstract headings | Structured abstract | Use Findings / Interpretation / Funding, not Results / Conclusions |
| No CONSORT/PRISMA flow diagram | Figures | Add the diagram; reconcile every number |
| Missing role-of-funding / data-sharing statement | Declarations | Add both; ICMJE data-sharing is mandatory for trials |
| Research in context panel absent or search-free | Main text | Build the three-part panel on a documented search |
Cover-letter pitch: the one load-bearing paragraph
The Lancet editor wants the practice/policy claim in the first lines, calibrated to the design. State the clinical or public-health problem, the single most important finding with its effect and 95% CI, and what guideline or decision it changes — for whom, and where (the global-relevance hook). Do not pad with adjectives; editors discount "novel" and "important" unless the numbers carry them.
Worked micro-example (illustrative numbers — not real data)
A hypothetical pragmatic, cluster-randomised trial of a community health-worker intervention to improve hypertension control across districts in three low- and middle-income countries.
Preflight verdict (illustrative):
Fit: rung 4 (establishes a scalable public-health intervention) -> GO
Registration: prospective, ISRCTN in abstract + Methods -> OK
Reporting: CONSORT-Cluster + flow diagram, page/line-mapped -> OK
Abstract: 296 words; headings Background/Methods/Findings/Interpretation/Funding -> OK
Primary outcome in Findings: BP control 38.1% vs 27.4%,
adjusted risk difference 10.6 pp (95% CI 7.2-14.0), NNT ~10 (illustrative)
Panel: Evidence-before search (MEDLINE+Embase+Global Index Medicus, dates+terms) -> OK
Declarations: role-of-funding + data-sharing + SAGER -> OK
Cover-letter pitch: "scalable, equity-focused, policy-relevant in LMIC primary care"
VERDICT: GO
Every other lancet-* skill's output has a line on this sheet; the preflight passes only when each lands, the numbers reconcile across abstract/diagram/Table 1, and the cover letter states the policy change in one sentence.
Reviewer / editor-pushback patterns and the venue-specific fix
- "Global-health relevance / equity not addressed." → Add the equity framing (PROGRESS-Plus, LMIC generalisability) in Interpretation and the cover letter; re-check
lancet-fit. - "Trial not prospectively registered." → Disclose registration timing transparently; if retrospective, flag it — the editor may decline, so address it head-on.
- "Primary outcome differs from the registered one." → Reconcile to the registered outcome or explain the dated change explicitly (see
lancet-study-design). - "Reporting checklist incomplete." → Complete and page/line-map the EQUATOR checklist before upload; confirm any current file-format requirement against the journal's author guidelines.
Output format
【Blocking gaps】 [...] (must fix before upload)
【Warnings】 [...] (should fix)
【Files ready】 main / abstract / panel / figures+tables / reporting checklist / protocol+SAP / cover letter / declarations
【Verdict】 GO / NO-GO + the top 3 fixes
【Next】 submit | lancet-rebuttal (after decision)
Anti-patterns
- Do not upload a trial that lacks prospective registration without flagging it.
- Do not submit an RCT without a CONSORT flow diagram or a review without a PRISMA one.
- Do not skip the role-of-the-funding-source or data-sharing statements.
- Do not rely on memory; run the checklist top to bottom.
Version History
- 1839142 Current 2026-07-05 14:02
