paper-skill
End-to-end, Chinese-first academic research and publication workflow for reading papers, literature reviews, research framing, manuscript planning and writing, the eight-step paper-system workflow, Chinese-English translation, LaTeX, figures, citation verification, reporting-guideline checks, journal selection, peer review, rebuttals, revisions, and submission packages. Use for SCI, IEEE, Nature-family, IJCAI, TRO, theses, systematic reviews, clinical/observational/animal studies, and any task where claims, evidence, references, manuscript files, or journal requirements must be checked rather than invented.
Safety scan
A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.
Badge for your README
Markdown
[](/listings/clin-c--paper-skill)HTML
<a href="/listings/clin-c--paper-skill"><img src="/api/listings/clin-c--paper-skill/badge.svg" alt="findsafeskills safety rating" /></a>Install
What was scanned
The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.
SKILL.md · Manifest · 163 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- ---
- name: paper-skill
- description: End-to-end, Chinese-first academic research and publication workflow for reading papers, literature reviews, research framing, manuscript planning and writing, the eight-step paper-system workflow, Chinese-English translation, LaTeX, figures, citation verification, reporting-guideline checks, journal selection, peer review, rebuttals, revisions, and submission packages. Use for SCI, IEEE, Nature-family, IJCAI, TRO, theses, systematic reviews, clinical/observational/animal studies, and any task where claims, evidence, references, manuscript files, or journal requirements must be checked rather than invented.
- ---
- # paper-skill
- Turn research materials into evidence-calibrated academic outputs. Treat the manuscript as a traceable system of claims, evidence, citations, figures, and submission requirements—not as a free-form writing task.
- ## Non-negotiable rules
- 1. Never invent data, experiments, statistics, quotations, references, DOI values, journal policies, author contributions, ethics approvals, funding, or revision locations.
- 2. Separate three states explicitly: `VERIFIED`, `UNVERIFIED`, and `AUTHOR_INPUT_NEEDED`.
- 3. Preserve scientific meaning before improving style. Flag contradictions instead of silently repairing them.
- 4. Map every material claim to evidence. Downgrade or remove unsupported claims.
- 5. Verify time-sensitive journal policies and bibliographic metadata against primary sources when tools permit. Record source and access date.
- 6. Never promise acceptance, impact factor outcomes, or detector evasion. Quality scores are internal triage, not publication predictions.
- 7. Keep a change ledger for manuscript edits: location, original, revision, reason, evidence affected.
- Read [references/verification-policy.md](references/verification-policy.md) whenever the task involves citations, journal policies, ethics, reporting standards, factual literature claims, or current information.
- ## Route the task
- First classify the request, then load only the indicated resources. For “论文体系8步”, “八步论文流程”, or continuing a numbered stage of that workflow, select `paper-system`; ordinary paragraph edits remain scoped to `write`.
- | Route | Typical requests | Required resources |
- |---|---|---|
- | `paper-system` (`eight-step`) | 论文体系八步:故事线、核心实验、完整实验、中文稿、文献核查、翻译排版、绘图、精简定稿;start or resume any stage | [workflows/paper-system-eight-steps.md](workflows/paper-system-eight-steps.md), then only the selected stage's prompts |
- | `read` | Read, summarize, critique, bilingual notes | [references/paper-writing-prompts.md](references/paper-writing-prompts.md) |
- | `literature` | Search strategy, review, gap map, related work | [workflows/literature-review.md](workflows/literature-review.md), [citation-integrity.md](citation-integrity.md) |
- | `write` (`plan-write`) | Storyline, outline, draft, rewrite, polish, translate; personal writing style and full-manuscript structure | Apply the writing profile below, the matching section guide, and [references/output-contracts.md](references/output-contracts.md) |
- | `figures` | Method diagram, result figure, caption, figure audit | [references/figure-prompts.md](references/figure-prompts.md), [workflows/figure-audit.md](workflows/figure-audit.md) |
- | `review` | Pre-submission review, rejection risks, adversarial audit | [quality-gates.md](quality-gates.md), [references/output-contracts.md](references/output-contracts.md) |
- | `rebuttal` | Reviewer response, R&R, rebuttal, revision verification | [references/output-contracts.md](references/output-contracts.md), [quality-gates.md](quality-gates.md) |
- | `citations` | DOI audit, hallucination scan, claim-source match | [citation-integrity.md](citation-integrity.md), [workflows/reference-tools.md](workflows/reference-tools.md) |
- | `journal` | Venue selection, transfer, policy, formatting | [journal-strategy.md](journal-strategy.md), [references/verification-policy.md](references/verification-policy.md) |
- | `submit` (`submission`) | Cover letter, highlights, declarations, final package | [workflows/submission-package.md](workflows/submission-package.md), [references/output-contracts.md](references/output-contracts.md) |
- | `reporting` | CONSORT, PRISMA, STROBE, ARRIVE, CARE, SPIRIT | [references/reporting-guidelines.md](references/reporting-guidelines.md), [references/verification-policy.md](references/verification-policy.md) |
- | `full-verify` | DOI lookup, journal policy evidence, manuscript/rebuttal consistency | [workflows/full-verification.md](workflows/full-verification.md) |
- For IEEE Transactions on Robotics, also load [references/venues/ieee-tro.md](references/venues/ieee-tro.md). Treat every venue profile as a dated starting point and re-check its official links before submission.
- For a multi-route request, load the smallest combination that covers it. Do not load the general prompt bank unless the routed resources are insufficient; then use [references/prompt-bank.md](references/prompt-bank.md).
- ### Personal writing style and manuscript structure
- For `write`, use [references/personal-writing-style.md](references/personal-writing-style.md) as the default writing profile. Requests such as “按我的论文风格”, “个人写作风格”, or “全文结构规范” select this profile explicitly; a structural review can use it without rewriting the manuscript. Read its application rules and the sections relevant to the requested scope.
- Build the argument from task/scientific question → capability needed → concrete obstacle and consequences → design principle → mechanism → evidence → scoped conclusion. Explain requirements backward from the goal, then describe the method forward along its actual dependencies. Keep claims, contributions, methods, figures, and evidence aligned.
- Explicit author preferences and verified venue requirements take precedence over these stylistic defaults. Preserve scientific meaning and evidence limits. Adapt to method, discovery, or theory papers; do not impose fixed counts of sections, contributions, modules, experiments, or sentences. A paragraph edit stays a paragraph edit. Deliver the requested prose first, with proportional verification notes and author questions outside the manuscript.
- ### Section-writing selector
- | Target | Load |
- |---|---|
- | Paragraph flow or full-section structure | [references/section-guides/paragraph-architecture.md](references/section-guides/paragraph-architecture.md) |
- | Abstract | [references/section-guides/abstract.md](references/section-guides/abstract.md) |
- | Introduction | [references/section-guides/introduction.md](references/section-guides/introduction.md) |
- | Related Work | [references/section-guides/related-work.md](references/section-guides/related-work.md) |
- | Method | [references/section-guides/method.md](references/section-guides/method.md) |
- | Experiments/Results | [references/section-guides/experiments.md](references/section-guides/experiments.md) |
- | Discussion/Conclusion | [references/section-guides/discussion-conclusion.md](references/section-guides/discussion-conclusion.md) |
- For worked patterns, load [examples/section-rewrites.md](examples/section-rewrites.md). The examples are synthetic; reuse reasoning structure, never their facts.
- ## Core workflow
- ### 1. Intake and provenance
- Create a compact intake record:
- - task and intended artifact;
- - paper type and discipline;
- - target venue and deadline, if supplied;
- - authoritative source files and version order;
- - available evidence: datasets, experiments, figures, tables, code, references;
- - required output format;
- - unresolved author decisions.
- If several manuscript versions exist, ask which is authoritative only when it cannot be inferred safely. Never merge conflicting values silently.
- ### 2. Build the argument model
- Before drafting or reviewing, state:
- - research question;
- - one-sentence central claim;
- - supporting claims warranted by the research and requested scope;
- - evidence supporting each claim;
- - scope conditions and limitations;
- - novelty relative to named prior work.
- Use IDs (`C1`, `E1`, `R1`, `F1`) so claims, evidence, references, and figures can be traced.
- ### 3. Choose the operation
- - **Draft:** build from the argument model, not from generic section templates.
- - **Rewrite:** diagnose structure first; preserve facts, equations, citations, and numbers.
- - **Translate:** translate meaning and rhetorical function; keep a terminology table and ambiguity log.
- - **Review:** separate fatal validity issues, major evidence gaps, minor clarity issues, and optional improvements.
- - **Rebuttal:** map every reviewer point to response, manuscript change, exact location, and verification status.
- - **Submission:** reconcile title, abstract, highlights, figures, declarations, and cover letter against the manuscript.
- ### 4. Verify
- Run checks proportional to the task:
- - claim ↔ evidence;
- - claim ↔ citation;
- - text ↔ figure/table;
- - response letter ↔ revised manuscript;
- - journal policy ↔ official current source;
- - reporting checklist ↔ study design;
- - terminology, symbols, units, sample sizes, and metric consistency.
- When a submission-package JSON is available, run:
- ```bash
- python tools/paper_skill_gate.py path/to/package.json
- ```
- Use `--format json` for machine-readable output and `--strict` to treat warnings as a nonzero exit.
- For a complete online and cross-file verification run, use:
- ```bash
- python tools/run_full_workflow.py path/to/workflow-config.json --output full-report.json
- ```
- ### 5. Deliver with uncertainty visible
- Every substantial output must include:
- 1. the requested artifact;
- 2. evidence/traceability notes appropriate to the task;
- 3. unresolved items labeled `AUTHOR_INPUT_NEEDED`;
- 4. verification summary—what was checked, what was not, and why;
- 5. next gate, if the artifact is part of a longer workflow.
- Follow the schemas in [references/output-contracts.md](references/output-contracts.md).
- ## File and artifact handling
- - Preserve the original; write revisions to a new file unless replacement is explicitly requested.
- - For DOCX, PDF, spreadsheets, or slides, use the relevant artifact skill and render/inspect the final output.
- - For LaTeX, compile when a TeX toolchain is available; inspect warnings, references, floats, and rendered pages.
- - For figures, inspect axes, units, legends, panel labels, sample sizes, uncertainty, and caption agreement.
- - Never claim a file was modified, compiled, or visually checked unless it actually was.
- ## Completion gates
- Do not call work “submission-ready” while any critical item remains:
- - unsupported central claim;
- - unresolved contradictory data;
- - fabricated or unverified high-risk citation;
- - missing ethics/registration information when applicable;
- - reviewer response without a matching manuscript change;
- - figure/table inconsistent with the text;
- - journal requirement not checked against a current official source.
- Use `READY`, `READY_WITH_WARNINGS`, or `NOT_READY`; always state the blocking evidence.
README.md · README · 283 lines
View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)
This file has changed since it was scanned, so this is not exactly the text that was scanned, and no line is marked. The findings are listed with the safety scan above.
- # paper-skill
- [项目介绍](#paper-skill) · [AI 提示词](https://clin-c.github.io/paper-skill/ai-prompts/) · [安装指南](docs/installation.md)
- > 面向中文科研作者的 evidence-first 英文投稿工作流。
- [](https://github.com/cLin-c/paper-skill/stargazers)
- [](https://github.com/cLin-c/paper-skill/actions/workflows/toolchain.yml)
- [](VERSION)
- [](LICENSE)
- [](https://github.com/openai/codex)
- [](https://claude.ai/code)
- <p align="center">
- <img src="assets/paper-skill-hero.png" alt="paper-skill evidence-first publication workflow" width="100%">
- </p>
- <p align="center">
- <img src="https://img.shields.io/badge/Google%20Scholar-literature%20map-0B57D0?style=for-the-badge&logo=googlescholar&logoColor=white" alt="Google Scholar literature map">
- <img src="https://img.shields.io/badge/arXiv-preprint%20workflow-B31B1B?style=for-the-badge&logo=arxiv&logoColor=white" alt="arXiv preprint workflow">
- <img src="https://img.shields.io/badge/LaTeX-manuscript%20ready-008080?style=for-the-badge&logo=latex&logoColor=white" alt="LaTeX manuscript ready">
- <img src="https://img.shields.io/badge/IEEE-submission%20style-00629B?style=for-the-badge&logo=ieee&logoColor=white" alt="IEEE submission style">
- <img src="https://img.shields.io/badge/Overleaf-writing%20pipeline-47A141?style=for-the-badge&logo=overleaf&logoColor=white" alt="Overleaf writing pipeline">
- </p>
- <p align="center">
- <a href="#三个旗舰入口"><img src="https://img.shields.io/badge/Storyline-paper%20arc-111827?style=for-the-badge&logo=readme&logoColor=white" alt="storyline module"></a>
- <a href="#三个旗舰入口"><img src="https://img.shields.io/badge/Translation-Chinese%20to%20English-D64545?style=for-the-badge&logo=deepl&logoColor=white" alt="translation module"></a>
- <a href="#三个旗舰入口"><img src="https://img.shields.io/badge/Reviewer%20Response-R%26R%20matrix-8B5CF6?style=for-the-badge&logo=gitbook&logoColor=white" alt="review response module"></a>
- <a href="citation-integrity.md"><img src="https://img.shields.io/badge/Citation%20Integrity-DOI%20audit-2563EB?style=for-the-badge&logo=zotero&logoColor=white" alt="citation integrity module"></a>
- <a href="quality-gates.md"><img src="https://img.shields.io/badge/Quality%20Gates-7D%20check-0F766E?style=for-the-badge&logo=checkmarx&logoColor=white" alt="quality gates module"></a>
- <a href="references/figure-prompts.md"><img src="https://img.shields.io/badge/Figures-visual%20contract-F59E0B?style=for-the-badge&logo=figma&logoColor=white" alt="figures module"></a>
- </p>
- 把中文初稿、实验结果、图表和参考文献转换为可核查的英文投稿材料:
- ```text
- 中文研究材料
- → 论文故事线与 claim–evidence map
- → 英文摘要 / 正文 / Cover Letter
- → 引用、图表、审稿回复一致性检查
- → VERIFIED / UNVERIFIED / AUTHOR_INPUT_NEEDED
- → READY / READY_WITH_WARNINGS / NOT_READY
- ```
- 它不是“帮我写得更像论文”的提示词合集。它把论文视为一套可追踪的主张、证据、引用、图表和投稿要求,并明确拒绝编造数据、实验、DOI、政策与修改位置。
- [English README](README_EN.md) · [安装说明](docs/installation.md) · [论文体系八步](#论文体系八步能力88条提示词) · [论文写作顺序](#论文写作顺序) · [公开 benchmark](benchmarks/README.md) · [完整示例](examples/README.md)
- ## 30 秒开始
- macOS / Linux:
- ```bash
- curl -fsSL https://raw.githubusercontent.com/cLin-c/paper-skill/main/install.sh | bash
- ```
- Windows PowerShell:
- ```powershell
- irm https://raw.githubusercontent.com/cLin-c/paper-skill/main/install.ps1 | iex
- ```
- 然后直接说:
- ```text
- 使用 $paper-skill。我的论文是中文初稿,目标期刊是 IEEE TRO。
- 先建立 claim–evidence map,再改写英文摘要;不要补写不存在的数据和引用,
- 最后列出 VERIFIED、UNVERIFIED、AUTHOR_INPUT_NEEDED 和投稿阻塞项。
- ```
- Claude Code 等 slash-command 平台可使用 `/paper-skill`。安装器只写入本机已存在的平台目录;若没有检测到平台,则默认安装到 Codex。可选的命令行核验工具需要 Python 3.10+。
- ## 个人论文写作风格与全文结构
- `write` 已接入作者的[个人论文写作风格与全文结构规范](references/personal-writing-style.md),覆盖标题与摘要、引言、相关工作、方法、实验、讨论、图表附录、段落衔接和主张—证据核对,保留原规范全部 14 部分及可复用指令。
- 默认按“研究问题 → 具体障碍 → 设计原则 → 方法机制 → 研究证据 → 有范围的结论”组织写作。章节、贡献和实验数量随研究内容调整;你明确指定的其他风格、写作范围和目标刊会要求优先。无需再次上传原始 Word 文档。
- ```text
- 使用 $paper-skill,按我的个人论文写作风格与全文结构规范,
- 根据附件重写引言和方法。中文输出,保留现有公式、术语和已确认数据;
- 把研究需求、具体困难、设计原则、模块依赖与证据逐项对应。
- 先给改写正文,再列出关键修改和待作者确认的问题,不编造缺失信息。
- ```
- 只润色摘要或一段文字时,说明范围即可;全文规范不会要求额外生成其他章节。英文写作使用相同论证逻辑,按英文习惯组织句子并保持证据强度。
- ## 论文体系八步能力(88条提示词)
- `paper-system`(别名 `eight-step`)已接入作者《AI提示词_论文体系8步_本地预览.html》中的全部 **88 条学术论文提示词**,支持从头推进、指定单步和继续已有进度。各步骤包含任务、产出与完成标准;技能按当前任务选用提示词并执行。
- | 步骤 | 任务与提示词入口 | 数量 |
- |---|---|---|
- | 1 | [设计故事线](references/paper-system/step-1.md) | 13 |
- | 2 | [做实验,解决核心难题](references/paper-system/step-2.md) | 3 |
- | 3 | [根据故事线完成所有实验](references/paper-system/step-3.md) | 10 |
- | 4 | [论文写作中文初稿](references/paper-system/step-4.md) | 18 |
- | 5 | [参考文献核查](references/paper-system/step-5.md) | 4 |
- | 6 | [翻译排版](references/paper-system/step-6.md) | 10 |
- | 7 | [绘图修改](references/paper-system/step-7.md) | 18 |
- | 8 | [按照要求进行排版修改,包括精简缩短篇幅](references/paper-system/step-8.md) | 12 |
- 第2步验证核心方法,第3步补齐全文证据;第6步完成翻译和初排版,第8步根据最终图文与篇幅要求定稿。文献阅读、草图与实验记录可同步开展,故事线随真实证据调整。研究目标可从**效果领先、能力突破、机理分析**三个角度确定。
- ```text
- 使用 $paper-skill 的 paper-system 论文体系八步能力。
- 先根据附件判断当前进度,从尚未完成且材料具备的步骤继续。
- 选择适用提示词并直接完成任务,按真实证据调整故事线,不编造数据或引用。
- 交付本步成果、主张—证据对应、未解决事项和下一步。
- ```
- 也可直接说:“使用 $paper-skill,按论文体系八步的第4步,根据附件完成中文初稿,按我的个人论文风格写作。”
- 完整说明见[八步工作流](workflows/paper-system-eight-steps.md)。提示词保留来源编号;个人路径已改为填写项,历史案例中的项目名称、数量与刊会要求按当前任务调整,不承诺录用概率。原HTML中的日常工作和合同提示词不纳入论文能力。
- ## 论文写作顺序
- 以下是八步流程中**第4步中文初稿内部的起草顺序**,也适用于独立写作任务。
- 从研究材料开始写作时,按以下顺序推进;这是起草与修改的工作顺序,成稿的章节排列按论文类型和目标刊会要求确定。
- 1. **明确核心问题、设计原则与证据对应关系**:先确定要回答什么问题、方法为什么这样设计,以及已有证据支持哪些主张。
- 2. **组织方法与实验**:说明方法机制、输入输出和实际依赖,再用实验或理论证据逐项回应主张;根据真实证据调整主张范围。
- 3. **完成引言、相关工作与贡献**:解释研究必要性、已有工作的边界和本文增量,使贡献与方法、证据对应。
- 4. **最后压缩摘要并确定标题**:从已经确认的正文与结果提炼完整论证,保持数值、术语和结论范围一致。
- 修改依次检查:**逻辑 → 一致性 → 表达 → 目标刊会要求**。分别核对需求与证据是否成立,术语、变量、数据、图表和实现是否一致,段落与句子是否清楚,以及篇幅、匿名、引用和文件格式是否符合要求。
- 详细规则见[第 13 节:写作顺序与修改重点](references/personal-writing-style.md#13-写作顺序与修改重点);完整 AI 提示词见[第 14 节:可直接复用的全文写作指令](references/personal-writing-style.md#14-可直接复用的全文写作指令)。也可以直接使用:
- ```text
- 使用 $paper-skill,按我的个人论文写作风格与全文结构规范,
- 根据附件按以下顺序完成指定范围的写作:
- 核心问题、设计原则与证据对应关系 → 方法与实验 → 引言、相关工作与贡献 → 摘要与标题。
- 根据真实证据调整主张范围,不编造数据、实验或引用。
- 完成后依次检查逻辑、一致性、表达和目标刊会要求,列出待作者确认的问题。
- 本次写作范围:【填写章节或全文】;目标刊会:【填写】;语言:【中文或英文】。
- ```
- ## 三个旗舰入口
- | 入口 | 适用场景 | 核心产物 |
- |---|---|---|
- | `write` | 中文材料转英文稿、重构摘要或章节 | 故事线、claim–evidence map、英文稿、变更记录 |
- | `review` | 投稿前审查、引用核验、图表核对 | 致命问题、证据缺口、引用风险、就绪状态 |
- | `submit` | Rebuttal、R&R、Cover Letter、投稿包 | 逐点回复矩阵、精确修改位置、声明、最终门控 |
- 窄任务也可以直接调用 `literature`、`figures`、`citations`、`journal`、`reporting` 或 `full-verify`。主 [SKILL.md](SKILL.md) 仅负责路由,详细规则按需加载,避免一次塞入全部上下文。
- ### 中文粗稿到 T-RO 的首次运行顺序
- 1. 上传权威版本的稿件、图表、参考文献/BibTeX、结果表和补充材料;说明版本优先级。
- 2. `write`:建立研究问题、中央主张、claim–evidence map 和章节修改计划。
- 3. `review`:核对数字、图表、引用、过度主张和匿名化风险。
- 4. `journal`:按 [T-RO 官方来源快照](references/venues/ieee-tro.md) 重新访问官方页面并生成带访问日期的要求表。
- 5. `submit`:生成 Cover Letter、声明、投稿清单;只有关键证据全部解决后才能输出 `READY`。
- 理想输入还包括作者顺序、单位、ORCID、基金、伦理审批、COI、数据/代码可用性、相关预印本和截止日期。分享材料前请移除不必要的个人、患者、审稿或机构敏感信息。
- ## 为什么它不一样
- | 常见问题 | paper-skill 的处理方式 |
- |---|---|
- | AI 把中文逐字翻成生硬英文 | 保留科学含义和修辞功能,维护术语表与歧义日志 |
- | 为了“润色”强化了没有证据的结论 | 每项 material claim 映射到证据,超出证据时降级或删除 |
- | 生成看起来真实的 DOI 或参考文献 | Crossref/OpenAlex 核验;未核验内容明确标记 `UNVERIFIED` |
- | 回复称“已修改”,但稿件中找不到 | reviewer point → response → manuscript change → exact location |
- | 图表数字与正文不一致 | figure/table ↔ claim ↔ manuscript consistency audit |
- | 不知道稿件是否真的可以投稿 | 输出 `READY`、`READY_WITH_WARNINGS` 或 `NOT_READY` 及阻塞证据 |
- ## 可执行工具,而不只是提示词
- ```bash
- python tools/paper_skill_gate.py examples/submission-package-check.json
- python tools/run_full_workflow.py examples/full-workflow-config.json --output full-report.json
- python tools/run_benchmark.py
- python -m unittest discover -s tests
- ```
- | 工具 | 用途 |
- |---|---|
- | [paper_skill_gate.py](tools/paper_skill_gate.py) | 身份、来源、主张证据、引用、报告规范、投稿包、未决输入七项门控 |
- | [scholarly_verify.py](tools/scholarly_verify.py) | 使用 Crossref/OpenAlex 核验 DOI、标题和年份 |
- | [journal_policy_verify.py](tools/journal_policy_verify.py) | 保存官方政策 URL、时间、证据片段和内容哈希 |
- | [revision_trace.py](tools/revision_trace.py) | 核对审稿回复中的修改位置和修订文本 |
- | [run_full_workflow.py](tools/run_full_workflow.py) | 串联投稿门控、引用、政策和修订一致性检查 |
- ### 图 1:引用侦探——从主张追到真实来源
- 这张图说明 `citation-integrity` 的核心流程:把论文主张连接到参考文献,核对 DOI/标题,并把疑似幻觉或错引标成待审问题。
- <p align="center"><img src="assets/comic-citation-detective.png" alt="引用核验漫画:从主张追踪到真实来源" width="100%"></p>
- ### 图 2:论文全流程——从散乱材料到可核验投稿包
- 这张流程图展示 paper-skill 的主线:识别论文类型、建立故事线、写作与翻译、检查证据和图表,最后生成审稿回复与投稿材料。
- <p align="center"><img src="assets/paper-workflow.png" alt="论文写作与投稿全流程图" width="100%"></p>
- ### 图 3:投稿工作流——把每个阶段连接起来
- 这张漫画把完整投稿过程串成连续步骤:整理研究问题、形成英文稿、核对引用和图表、处理审稿意见,并在最终提交前保留未解决事项。
- <p align="center"><img src="assets/comic-submission-workflow.png" alt="从混乱初稿到核验投稿包的投稿流程漫画" width="100%"></p>
- ### 图 4:学术绘图——先定义证据,再设计视觉表达
- 这张图强调 figures 工作流不是先追求好看,而是先回答“图要证明什么”,再选择方法图、结果图或图形摘要,并检查 caption 与正文的一致性。
- <p align="center"><img src="assets/figure-generation.png" alt="学术图表生成与核验流程图" width="100%"></p>
- ### 图 5:小猪的投稿故事——一次完整的 paper-skill 使用示例
- 这是项目的叙事入口:小猪研究者经历“初稿混乱 → 故事线重构 → 主张、实验、引用和图表核验 → 审稿回复与投稿包整理”。它是帮助新用户快速理解产品价值的虚构示例,不代表录用保证。
- <p align="center"><img src="assets/comic-pig-paper-skill-story.png" alt="小猪研究者使用 paper-skill 改进论文并准备投稿的四格漫画" width="100%"></p>
- ## 公开、可复现的 benchmark
- ```bash
- python tools/run_benchmark.py
- ```
- 当前确定性安全门控 benchmark:**8/8 分类正确**。
- 覆盖:完整投稿包、身份信息缺失、无证据主张、断裂证据链接、可疑 DOI、伪 `VERIFIED` 引用、未解决作者输入、报告规范缺口。所有用例位于 [benchmarks/gate-cases.json](benchmarks/gate-cases.json),CI 会在 Windows 和 Linux 上重复运行。
- 这个结果只衡量结构化投稿包的风险分类,不代表语言质量、科学创新性或录用概率。项目不会用内部评分预测论文录用。
- ## 工作流结构
- ```text
- SKILL.md concise router
- ├── references/section-guides/ Abstract → Discussion 的分节写作规则
- ├── workflows/ 文献、图表、引用、投稿包、完整核验
- ├── citation-integrity.md 引用真实性与 claim-source 匹配
- ├── journal-strategy.md 选刊、转投和政策核验
- ├── quality-gates.md 审稿、R&R 和投稿前门控
- ├── tools/ 确定性检查与在线核验脚本
- ├── benchmarks/ 公开评测用例
- └── examples/ 合成、可复制的 before/after 示例
- ```
- 详细写作能力覆盖 Abstract、Introduction、Related Work、Method、Experiments/Results、Discussion/Conclusion、翻译、学术图表、CONSORT、PRISMA、STROBE、SPIRIT、ARRIVE、CARE、CRediT、伦理与利益冲突声明。
- ## 可靠安装与升级
- 从 v0.8.0 开始,安装器会:
- - 对 Git 安装执行 fast-forward 更新;
- - 识别早期复制式/非 Git 安装;
- - 在迁移前创建带时间戳的备份;
- - 克隆或验证失败时恢复原安装;
- - 显示迁移前后版本;
- - 在 CI 中验证 Windows 与 Linux 迁移路径。
- 如果你曾安装旧版,重新运行上面的单行命令即可。旧目录不会被静默覆盖。
- ## 项目质量门槛
- 每次 push 和 pull request 都必须通过:
- - Windows + Linux 单元测试;
- - 公开 benchmark;
- - 示例投稿包门控;
- - `SKILL.md` 长度与路由检查;
- - UTF-8 与常见中文乱码检查;
- - `VERSION` 和 `agents/openai.yaml` 一致性检查;
- - 旧安装安全迁移测试。
- ## 社区
- - [提交真实工作流需求](https://github.com/cLin-c/paper-skill/issues/new?template=example-request.yml)
- - [报告问题](https://github.com/cLin-c/paper-skill/issues/new?template=bug-report.yml)
- - [提出功能建议](https://github.com/cLin-c/paper-skill/issues/new?template=feature-request.yml)
- - [路线图](ROADMAP.md)
- 项目优先接受可复现的真实科研场景、失败案例、匿名化 before/after 和 benchmark 用例,而不是继续堆叠泛化提示词。
- ## License
- MIT。写作工作流的外部灵感与授权信息见 [ATTRIBUTIONS.md](ATTRIBUTIONS.md)。
Ratings and comments
Sign in to rate this skill.
Sign inFree to use. If it’s helped, pass it on.