New to Claude Skills? Learn how to install them →

yuan1z0825 on GitHub

Research Writing Pipeline

Free

Streamline your scientific writing with structured proposal-first methodologies.

Get this skill

Free · Opens the source repo

What Research Writing Pipeline does

The Research Writing Pipeline is a specialized tool designed to enhance the process of composing, revising, and auditing scientific documents such as research proposals and reports. It operates under a proposal-first framework, ensuring that all writing is grounded in evidence and structured logically. This skill is particularly useful for researchers and academics who need to produce high-quality, evidence-based proposals that adhere to rigorous standards.

The pipeline emphasizes a systematic approach to writing, requiring users to establish a research canon and evidence table before drafting any prose. This ensures that all claims made in the document are supported by solid evidence. The tool provides three distinct modes of operation: compose, revise, and hybrid. Each mode is tailored to different stages of the writing process, allowing users to work with either initial ideas or existing drafts effectively.

In addition to its structured writing approach, the Research Writing Pipeline includes a quality assurance (QA) mode that evaluates the content based on multiple criteria. This four-layer QA pipeline assesses the validity of claims, the presence of citations, and the overall coherence of the document. The skill also allows for dynamic expert consultation, enabling users to summon relevant expertise based on the specific challenges they face in their writing.

Overall, this skill is ideal for researchers who are looking to streamline their writing process while ensuring that their work meets high academic standards. By following the prescribed methodologies, users can enhance the clarity and impact of their scientific communications.

When to use it

Use this skill when you need to draft, revise, or audit research proposals and related documents with a focus on evidence and logical argumentation.

When not to use it

This skill may not be suitable for general writing tasks outside of scientific or academic contexts, as it is specifically tailored for structured proposal writing.

What you can build with it

Drafting a New Research Proposal

Use the compose mode to outline your ideas and establish a research canon before writing.

Revising an Existing Proposal

Switch to the revise mode to improve an already drafted proposal by focusing on evidence and structure.

Conducting a Quality Review

Enter QA mode to assess your document against quality metrics and receive targeted feedback.

How to install Research Writing Pipeline

View source

1. Install with the skills CLI

npx skills add yuan1z0825/nature-skills/nature-proposal-writer --agent claude-code

2. Or install it manually

Download the skill folder and drop it into ~/.claude/skills/ for all projects, or .claude/skills/ to scope it to one repo. Restart Claude Code so it picks up the new skill.

Anthropic's agentic coding CLI, and the reference implementation of Agent Skills. Drop a skill folder into ~/.claude/skills and Claude Code loads it automatically whenever a task matches the skill's description. Claude Code docs

Inside SKILL.md

Written by yuan1z0825

researchwrite — proposal-first 科研写作 pipeline

受 autonovel(状态机+打分)、professor(动态专家)、brainstorming(入口追问)、anti-AI-writing(语言清理)启发的科研写作状态机。不是通用"帮我写论文"prompt。

researchwrite 是为兼容既有安装保留的触发名称,仓库目录名仍为 nature-proposal-writer。不要在没有迁移方案时修改 frontmatter 名称。需要初始化模板、 按条件加载参考资料或执行导出脚本时,先读取 manifest.yaml,只加载与当前条件匹配的路径。

核心原则

  1. 证据先于文字 — 起草前必须建立或读取 research_canonevidence_table
  2. 论证先于章节 — 写正文前必须完成 argument_map
  3. 契约先于段落 — 每节需要 purpose / allowed claims / forbidden claims / inputs / validation
  4. 范围先于完备 — 如果是分阶段写作,先锁定阶段边界
  5. 动态专家,不设固定池 — 用 professor 按失败模式召唤对应专家
  6. 内容先于语言 — 诊断科学逻辑后再做 anti-slop / 语言打磨
  7. 不自动升级事实 — 永远不把 "may indicate" 改成 "proves",除非有证据支撑
  8. 删除胜于解释 — 当某主张不可行,直接删除。正文干净,解释留给答辩
  9. 该停就停 — 平台期、专家冲突、证据缺失是停止理由,不是润色理由

模式分派

输入模式
题目、方向、模糊想法compose — 加载 references/compose-mode.md
已有段落/章节/完整 proposalrevise — 加载 references/revise-mode.md
已有草稿 + 扩写/补写/重构hybrid — 加载 references/hybrid-mode.md

模糊时推断默认模式。只在选择会改变工作流时才问用户。

项目结构

工作目录<outputs>/researchwrite/<project-slug>/

标准文件

00_scope.md              写作任务边界
01_research_canon.md     硬事实和约束
02_evidence_table.md      claim → evidence 映射表
03_argument_map.md        论证架构
04_section_contracts.md   每节的 purpose / inputs / allowed & forbidden claims
05_style_guide.md         风格、术语、禁用表达
state.json                项目状态(mode、round、scores、technical_debts)
sources/                  用户材料、文献、数据
drafts/                   草稿和分节文件
revision_briefs/          修订简报
qa_logs/                  诊断、专家审查、anti-slop、打分记录
exports/                  最终输出(.md + .docx)

新建项目时按 manifest.yamltemplates.on_demand 路由从 templates/ 取空模板。 references/worked-example-quaternary-proposal.md 提供了基于通用材料科学领域的完整填写示例。

Reference 文件索引

完整 reference 按任务需要加载,不要一次性注入全部文件:

Reference加载时机
references/compose-mode.md模式 = compose(9 步流程)
references/revise-mode.md模式 = revise(9 步流程)
references/hybrid-mode.md模式 = hybrid
references/evaluation-rubric.md使用 8 维 × 4 锚点评分体系时
references/research-anti-slop.md清理中文 proposal 的模板化和空泛表达时
references/chinese-review-writing-style.md撰写或修订中文综述时
references/stopping-rules.md判断继续迭代、拆分范围或停止时
references/professor-dispatch.md按失败模式分派动态专家时
references/foundation-files.md建立或修复 foundation 五文件时
references/project-structure.md初始化项目目录或维护 state.json
references/export-archive.md导出和归档 .md / .docx
references/partial-proposal-scope.md分阶段写作并防止范围蔓延时
references/ref-renumbering-cascade.md处理参考文献增删和编号级联时
references/review-paper-framework.md设计综述论文框架时
references/review-critique-methodology.md建立综述的批判性分析时
references/validation-checklist.md自动检查主张、引用、编号和可复现性时
references/gpt-handoff-revision-brief.md生成跨 agent 修订交接简报时
references/within-approved-proposal.md在已批准的 proposal 框架内扩写时
references/worked-example-quaternary-proposal.md需要查看完整 foundation 文件填写示例时
references/降承诺提案模式.md证据不足,需要降低承诺强度时

运行交付

每次运行结束时输出:

  1. 当前文件路径或修订后文本
  2. 当前分数/状态
  3. 剩余风险
  4. 一条建议的下一步

QA Mode — 四层质量保障 Pipeline

当用户说"审查这段/跑 QA/检查方案/过 pipeline"时,进入 QA 模式。

情境挡位

挡位适用场景阈值
paper投稿论文7.0
proposal研究方案/开题7.0
internal内部汇报/周报5.0
quick快速扫读

Pipeline 顺序(先内容后语言)

Gate 2: professor Convener(内容层)
  ├── 论文 → 方法论专家 + 领域专家
  ├── proposal → 可行性专家 + 创新性专家
  └── 文献综述 → 覆盖面专家 + 批判深度专家
  │
  ▼
Gate 1: avoid-ai-writing 模式 detect-only
  └── 只对英文有效,中文跳过或降级为手工扫读
  │
  ▼
Gate 3: auto-validation(格式/完整性层)
  └── 每个 claim 有 citation?方法可复现?编号连续?
  │
  ▼
Gate 4: 评分阈值(分维度打分)
  ├── 总分≥阈值 → 通过 ✅
  └── 总分<阈值 → 按低分维度定向回退(不超过 3 轮)

Gate 2 在 Gate 1 之前 — 避免改了句子后被专家打回重写。

使用方法

用户: "用 researchwrite 审查这段 discussion,paper 挡位"
  → 自动走 Gate 2 → Gate 1 → Gate 3 → Gate 4
  → 返回审查报告 + 定向修改建议

用户: "快速扫一下这个邮件"
  → 走 quick 挡位,只标记 P0 问题

配置你的研究域

首次使用告诉 agent 你的研究背景,agent 会调用 professor 建立领域专家知识。后续写作中专家审查会基于你的领域。

Frequently asked questions about Research Writing Pipeline

Similar skills