科研 Skill

投稿前自我审稿预评估

self-review

Checks a manuscript before submission from a reviewer’s perspective, with leakage checks. Use when self-auditing one’s own manuscript ahead of submission.

发布者 Codar安全审核通过v0.1.0更新于 2026-11-18

SKILL.md

Self-Review Skill

Check the user's own manuscript before submission and produce an actionable list of anticipated reviewer comments, each with a specific fix — not a written review.

Optional Flags

  • --fix: after the report, apply fixes for every issue with fixable_by_ai: true, editing the manuscript in place, then report a diff summary. Never fix fixable_by_ai: false issues (missing data, design flaws). Maximum 2 fix-and-re-review iterations; if the score is still below threshold after the second, stop and report what remains as structural (inside /write-paper this routes to Phase 7.4a Audit Recovery).
  • --json: also emit the structured JSON block (Phase 3c). Default when called from /write-paper Phase 7.
  • --panel: run the multi-agent panel review (Phase 2.6) — domain-expert reviewers in parallel plus an editor synthesis — instead of the single-pass review. Opt-in and off by default, because it spawns N reviewer agents + 1 editor and costs several times more tokens; reserve it for a high-stakes final pass on a top-tier target. Do not combine with --fix: a panel diagnoses and prioritizes; run --fix as a separate pass once the author has triaged the panel's findings.

Severity Framing

  • Fatal: a design flaw existing data cannot fix (data leakage that invalidates all results, no reference standard, label–feature circularity). Submission would likely be rejected. Reserve Fatal for true design-level problems.
  • Fixable: significant but addressable with existing data (missing calibration, unclear exclusion criteria, absent CIs, incomplete reporting). Most issues are Fixable.

Two Objectives: the Floor and the Ceiling

The floor (categories A–K and the gates of Phases 2.5–2.5f) minimizes rejection-for-cause — fabricated citations, numbers that do not reconcile, overclaims, missing checklist items, leakage — and much of it works by adding a hedge, caveat, disclosure or audit trail. Iterated, that over-hardens a manuscript: every finding is correct, yet the whole reads as a defensive audit (over-hedged, audit trail in the body, Abstract buried under caveats, the strongest robustness result hidden in Limitations). The ceiling pass (category L / Phase 2.5g) runs after the floor gates, reads the accurate manuscript as a whole, and recommends only SUBTRACTION — REMOVE, MOVE, or TIGHTEN. It is advisory, never blocks, and cannot relax a floor gate. Report its findings as their own block (Phase 3), not folded into the "add this" comments. Phase 2.5i then reads the floor + ceiling state to declare the loop done, including a zero-edit PASS.

Workflow

Full steps for each part are in references/workflow/; open the one you need:

  • Phase 1: Intake — references/workflow/01-phase-1-intake.md: 1. Get the manuscript — PDF, Word doc, or pasted text. 2. Ask the user: target journal (sets reporting standards and scope); manuscript type (original research / review / perspective / technical note / letter / meta-analysis / case report)…
  • Phase 2: Systematic Check — references/workflow/02-phase-2-systematic-check.md: Work every category the Research-Type Adaptation table (below) marks as applicable, and for each item decide whether a reviewer would raise it as a Major or Minor comment. The per-item check tables are in `references/phases/phase2_systemat…
  • Research-Type Adaptation — references/workflow/03-research-type-adaptation.md: *Meta-analysis: replace C with heterogeneity (I², prediction intervals), publication bias (funnel plot, Egger), and sensitivity/subgroup analyses.
  • Phase 2.5: Numerical Cross-Verification (Internal) — references/workflow/04-phase-2-5-numerical-cross-verification-internal.md: Before the report, verify internal consistency:
  • Phase 2.5a: Numerical Source-Fidelity Audit (External) — references/workflow/05-phase-2-5a-numerical-source-fidelity-audit-exter.md: Numbers can be self-consistent everywhere and still wrong at the source; only a traversal back to the primary source catches that. First run the displayed-arithmetic gate — a stated difference must equal the subtraction of its two disp…
  • Phase 2.5a-2: Design & Power Statistic Provenance — references/workflow/06-phase-2-5a-2-design-power-statistic-provenance.md: Applies only when the manuscript states a sample-size calculation, a power figure, or a detectable-effect claim. These are computed, not copied, so Phase 2.5a cannot check them: re-derive each from the manuscript's own inputs. A value not…
  • Phase 2.5b: Screening-Count Reconciliation from ID Sets (SR/MA + observational tier/stratum) — references/workflow/07-phase-2-5b-screening-count-reconciliation-from-i.md: When to run: any SR/MA manuscript revision, at any stage (before Phase 3); or any observational manuscript presenting an ordinal tier / mutually-exclusive stratum split. Skip otherwise. A wrong prose total survives every other pass bec…
  • Phase 2.5c: Reference Scans (hallucination + adequacy) — references/workflow/08-phase-2-5c-reference-scans-hallucination-adequac.md: 2.5c catches a citation that does not exist or has an invented first author; 2.5c-2 catches a claim with no citation. Both need a bibliography — skip them if there is no refs.bib and no reference list. Run /verify-refs --strict…
  • Phase 2.5d: Cross-Reference QC (Manuscript ↔ rendered DOCX) — references/workflow/09-phase-2-5d-cross-reference-qc-manuscript-rendere.md: In-text Table/Figure citations can resolve to a different caption in the rendered DOCX when the build script carries its own legacy SSOT; no other phase sees this.
  • Phase 2.5e: Confounding Completeness (observational only) — references/workflow/10-phase-2-5e-confounding-completeness-observationa.md: When to run: observational manuscripts (cohort, case-control, cross-sectional, health-screening registry) whose central claim is an adjusted exposure–outcome association. **Skip for RCTs, diagnostic-accuracy, SR/MA, and descriptive stu…
  • Phase 2.5f: Claim-vs-Artifact Cross-Check — references/workflow/11-phase-2-5f-claim-vs-artifact-cross-check.md: This phase checks claims against the artifacts they should trace to — the pre-registration, the protocol, the analysis outputs — where the prose is internally consistent yet disagrees with them. Run the gates (pass the supplement so the co…
  • Phase 2.5g: Editorial-Impression / Defensiveness Scan (the ceiling pass) — references/workflow/12-phase-2-5g-editorial-impression-defensiveness-sc.md: Run this after the floor gates (Phases 2.5–2.5f): it reads the accurate manuscript and recommends what to take back out. It is advisory and non-blocking — it never produces a Major and never gates submission.
  • Phase 2.5h: Baseline Drift (anchor to the last human-approved version) — references/workflow/13-phase-2-5h-baseline-drift-anchor-to-the-last-hum.md: Run after the ceiling pass and before the loop controller (Phase 2.5i). Each refine pass takes the previous AI output as its baseline, so framing bias compounds unseen; this gate compares the manuscript against the **last human-approve…
  • Phase 2.5i: Refinement Terminal-State (the loop controller) — references/workflow/14-phase-2-5i-refinement-terminal-state-the-loop-co.md: Run this last, after the floor gates and the ceiling pass: it reads their qc/*.json artifacts and classifies whether the review → revise → review loop is done, so an accurate manuscript is not over-hardened by a pass it does not need…
  • Phase 2.5j: Refinement Regression (fixed vs broke, across runs) — references/workflow/15-phase-2-5j-refinement-regression-fixed-vs-broke.md: Run each round, after the loop controller. A revision that resolves finding X can introduce finding Y, and the pass-rate hides it; this step reads a run-history ledger (one line per run, the verdict@where fingerprints of its findings) an…
  • Phase 2.6: Multi-Agent Panel Review (--panel, opt-in) — references/workflow/16-phase-2-6-multi-agent-panel-review-panel-opt-in.md: Run this phase only when --panel is passed, after the numerical audits (Phases 2.5–2.5d) so the reviewers see source-verified numbers, and before the Phase 3 report, which it feeds. Two things bind before you spawn anything: the **SS…
  • Phase 3: Report — references/workflow/17-phase-3-report.md: Before writing the comments, skim references/exemplar_findings/ for the finding at hand (cohort-arithmetic mismatch, unadjusted confounder, cross-sectional scope overreach, post-hoc primary / estimand drift). Each models the full shape —…
  • Phase 3b: R0 Numbering (Optional) — references/workflow/18-phase-3b-r0-numbering-optional.md: If the user plans to use /revise once real reviews arrive, offer to append R0-numbered output so they can later tell anticipated (R0) from novel (R1-only) comments:
  • Phase 3c: Structured JSON Output (--json) — references/workflow/19-phase-3c-structured-json-output-json.md: Emit machine-readable JSON only when --json is passed (or another skill consumes this run): append the block to the report and also write it to qc/self_review.json. Read references/phases/phase3c_json_output.md for the schema, fi…
  • Phase 4: Fix Support (on request) — references/workflow/20-phase-4-fix-support-on-request.md: The review ends at Phase 3. Read references/phases/phase4_fix_support.md when --fix was passed or the user asks you to apply or draft fixes for the findings.

示例提示词

帮我以审稿人视角挑一下这篇稿件可能被提的主要问题。

所属插件

临床医学与公共卫生面向临床与公卫研究:研究设计与方案、样本量、队列与 Meta 分析、报告规范核查、医学影像 AI 评估、流行病建模与卫生经济学评价、医学论文写作投稿。

安装该插件会一并获得同领域的 58 个 Skill / MCP。

同插件的其他科研 Skill

常见问题

投稿前自我审稿预评估能做什么?

Checks a manuscript before submission from a reviewer’s perspective, with leakage checks. Use when self-auditing one’s own manuscript ahead of submission。

如何安装投稿前自我审稿预评估?

在科搭 Codar 左侧点击「插件」,在「组件」页签搜索「投稿前自我审稿预评估」并安装;在 Claude Code 或 Codex 中可以发送本页的安装提示词,或下载 Skill 包后解压到 ~/.claude/skills 或 ~/.codex/skills。

投稿前自我审稿预评估需要哪些权限?

该科研 Skill没有声明额外权限;执行具体操作时仍由 Agent 按自身规则请求确认。

投稿前自我审稿预评估经过官方审核吗?

是。SpeedSkills 收录的每个版本都会记录内容摘要与权限声明,当前版本已通过安全审核。