Fin Paper Writing logo

Fin Paper Writing

Community
csmar432
fin-paper-writing

经济金融论文写作编排器。根据PAPER_OUTLINE.md和REFINED_DESIGN.md,编排调用fin-paper-draft(正文写作)、fin-paper-figure(图表生成)、fin-review-loop(review循环),管理版本并确保章节间的一致性。

Overview

Publishercsmar432
Repositoryfinai-research
Skill namefin-paper-writing
Stars
108
Forks
4
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by csmar432 on GitHub. Read the source before you install it.

Installation

Install the Fin Paper Writing AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/csmar432/finai-research/tree/main/.agents/skills/fin-paper-writing
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/csmar432/finai-research.git /tmp/finai-research
mkdir -p .claude/skills
cp -r /tmp/finai-research/.agents/skills/fin-paper-writing .claude/skills/fin-paper-writing
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Fin Paper Writing in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Fin Paper Writing on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Fin Paper Writing is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

经济金融论文写作编排器

编排从大纲到可投稿论文的全流程,管理章节间的引用一致性和版本迭代。

触发条件

触发关键词:写论文、paper writing、论文写作、开始写作

与 fin-full-pipeline 的关系

fin-full-pipeline 负责"研究"阶段(lit→idea→novelty→design), 本技能负责"写作"阶段(outline→draft→figure→review→convert)。

fin-full-pipeline 终点
       ↓ PAPER_OUTLINE.md + REFINED_DESIGN.md
fin-paper-writing(写作编排)
       ↓
┌──────────────────────────────────────────┐
│ 阶段A: 正文写作 (fin-paper-draft)         │
│   → 生成 draft_v{N}/introduction.tex     │
│       ↓ checkpoint                       │
│ 阶段B: 图表生成 (fin-paper-figure)       │
│   → 生成 draft_v{N}/figures/*.pdf       │
│       ↓ checkpoint                       │
│ 阶段C: 整合检查(章节一致性)              │
│   → 验证引用、表编号、图编号              │
│       ↓ checkpoint                       │
│ 阶段D: 对抗性Review (fin-review-loop)    │
│   → 多轮review直到达标                   │
│       ↓ checkpoint                       │
│ 阶段E: 编译 (fin-paper-convert)          │
│   → 生成 main.pdf                        │
│       ↓                                  │
│ 阶段F: 投稿前检查 (fin-submit-check)      │
│   → SUBMIT_CHECK_REPORT.md              │
└──────────────────────────────────────────┘
       ↓
输出: draft_v{N}/main.pdf(可投稿)

前置条件检查

Phase A: Pre-writing Checklist

在开始写作前,验证以下文件存在:

检查项文件路径说明
[ ] 论文大纲output/fin-manuscript/PAPER_OUTLINE.md章节结构和内容要点
[ ] 研究设计output/fin-refinement/REFINED_DESIGN.md变量定义、识别策略
[ ] 实证结果output/fin-experiments/results/表格/图表文件
[ ] 参考文献draft_v{N}/references.bibBibTeX格式
[ ] 研究简报FIN_BRIEF.md目标期刊、语言

如果任何文件缺失,停止并提示用户先完成前置步骤。

章节写作顺序(关键原则)

Phase B: Chapter Writing Order

始终使用此顺序,不要按论文的自然顺序写:

推荐顺序(而非论文顺序):
1. Data & Methodology → 先写方法,数据最清晰
2. Descriptive Statistics → 紧跟方法
3. Main Results → 结果最清晰,容易写
4. Mechanism Analysis → 如果有
5. Heterogeneity Analysis → 如果有
6. Robustness Checks → 紧凑汇总
7. Conclusion → 有全篇在手,总结更全面(最难)
8. Introduction → 最后写intro,知道全貌,贡献更清晰
9. Literature Review & Hypotheses → H1/H2/H3框架
10. Abstract → 写完所有章节后,自然形成摘要

为什么这样排序?

  • 方法和数据是基础,写完就清楚论文的"做什么"
  • 结果是论文核心,先写建立信心
  • 结论需要全篇在手,最后写才能全面总结
  • 引言需要知道结论,才能准确描述贡献
  • 摘要需要知道全文,才能精炼概括

编排协调

Phase C: Orchestration

编排调用以下子技能:

python
# 协调关系图:
fin-paper-writing (编排器)
    │
    ├──→ fin-paper-draft (正文写作)
    │       每个章节独立生成
    │
    ├──→ fin-paper-figure (图表生成)
    │       图表生成后与正文交叉引用
    │
    ├──→ fin-ref-paper (参考文献管理)
    │       确保引用格式一致
    │
    └──→ fin-review-loop (Review循环)
            每章节完成后Review

调用方式(Cursor Skill语法):

markdown
<!-- 写方法论章节 -->
Skill: fin-paper-draft
"methodology"

<!-- 生成图表 -->
Skill: fin-paper-figure
"main_results"

<!-- 管理参考文献 -->
Skill: fin-ref-paper
"update"

<!-- Review章节 -->
Skill: fin-review-loop
"methodology_v1"

版本管理

Phase D: Version Management

VERSION_MANIFEST.md

维护版本变更记录:

markdown
# 版本清单

**论文标题**: [标题]
**目标期刊**: [期刊]
**当前版本**: draft_v{N}

## 版本历史

| 版本 | 日期 | 主要变更 | 作者 |
|------|------|---------|------|
| draft_v1 | 2026-06-13 | 初始草稿 | Agent |
| draft_v2 | 2026-06-14 | 补充机制检验 | Agent |
| draft_v3 | 2026-06-15 | 根据Review修改引言 | Human |

## 当前状态

| 章节 | 状态 | 字数 | 最后修改 |
|------|------|------|---------|
| introduction.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 1,234 | 2026-06-13 |
| literature.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 2,345 | 2026-06-13 |
| methodology.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 1,567 | 2026-06-12 |
| results.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 3,210 | 2026-06-12 |
| robustness.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 890 | 2026-06-13 |
| conclusion.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 678 | 2026-06-13 |
| abstract.tex | [DRAFT/REVIEWING/REVISED/FINAL] | 250 | 2026-06-13 |
| references.bib | [DRAFT/REVIEWING/REVISED/FINAL] | — | 2026-06-13 |
CHAPTER_STATUS.md

追踪每个章节的详细状态:

markdown
# 章节状态追踪

## 章节详情

### introduction.tex
- **状态**: DRAFT
- **字数**: 1,234
- **创建时间**: 2026-06-13 10:00
- **最后修改**: 2026-06-13 14:30
- **Review轮次**: 0
- **检查清单**: 
  - [x] 动机充分
  - [x] 3个贡献点清晰
  - [ ] 字数符合期刊要求
  - [x] Roadmap包含
- **备注**: 需补充第3个贡献点

### methodology.tex
- **状态**: REVIEWING
- **字数**: 1,567
- **Review轮次**: 1
- **Review反馈**: 通过
版本迭代流程

当需要重大修改时:

markdown
## 创建新版本流程

1. 将 draft_v{N}/ 复制为 draft_v{N+1}/
2. 在 VERSION_MANIFEST.md 中记录:
   | draft_v{N+1} | [日期] | [修改原因] | Agent/User |
3. 在 CHAPTER_STATUS.md 中标记需修改的章节为"重写"
4. 执行修改
5. 重新执行 阶段C(一致性检查)→ 阶段D(Review)→ 阶段E(编译)→ 阶段F(检查)

一致性检查

Phase E: Consistency Checks

每章节完成后,自动执行以下检查:

检查1:图表编号一致性
图编号FIGURE_PLAN中的位置正文引用状态
fig1_sample3.1(样本构建)\ref{fig1_sample} in §3.1✅/❌
fig2_trend3.2(描述统计)\ref{fig2_trend} in §3.2✅/❌
fig3_main4.1(主结果)\ref{fig3_main} in §4.1✅/❌
fig4_hetero4.3(异质性)\ref{fig4_hetero} in §4.3✅/❌
fig5_mechanism4.4(机制)\ref{fig5_mechanism} in §4.4✅/❌
fig6_placebo4.5(稳健性)\ref{fig6_placebo} in §4.5✅/❌
检查2:表格编号一致性
表编号TABLE_PLAN中的位置正文引用状态
tab:var_def3.2(变量定义)\ref{tab:var_def}✅/❌
tab:summary3.3(描述性统计)\ref{tab:summary}✅/❌
tab:main4.1(主回归)\ref{tab:main}✅/❌
tab:heterogeneity4.3(异质性)\ref{tab:heterogeneity}✅/❌
tab:robustness4.5(稳健性)\ref{tab:robustness}✅/❌
检查3:假设与结果一致性
假设章节位置结果位置对应结论状态
H1§2.3§4.1[结论]✅/❌
H2§2.3§4.3[结论]✅/❌
H3§2.3§4.4[结论]✅/❌
检查4:引用一致性
  • 所有 \cite{} 都在 references.bib 中存在
  • 引用格式统一(JF风格/GB-T-7714)
  • 无"幽灵引用"(引用但无对应条目)
检查5:文字质量检查(新增)

每章节完成后,必须执行以下 AI 味和底气检测:

AI 味检测清单(扫描到任一关键词即为不合格):

扫描关键词(中文):值得注意的是 | 综上所述 | 从某种意义上 | 客观来说 | 在一定程度上 | 某种程度上 | 相对而言 | 近年来 | 随着时代发展 | 当今社会 | 可能存在 | 有待进一步研究 | 首先我们需要 | 非常重要的 | 具有重大意义的 | 已经被 | 可以被认为 | XX研究表明

底气指数评估(结论段必须满足≥1项):

底气要素检测标准
具体数字β = X.XXX,95% CI [...],p = X
经济规模占均值/标准差的 X%
机制描述通过 X 渠道,而非 Y 渠道
对比发现A 效应是 B 效应的 X 倍
理论对应与 [理论名] 的预测一致/相反

表格引导句检查:每张表格前是否有≥2句引导(主题句 + 来源句 + 核心发现句),每张表格后是否有≥1句解读。

图表注释解读检查:每个图表的 caption/注释是否包含"说明什么"(而不仅是"展示什么"),是否包含具体数字或对比描述。

语言质量检查不通过的章节,不得进入 Review 环节,必须先重写。


语言质量标准(新增)

T07:语言质量 Review 环节

在 Phase C(Orchestration)中,每章节写作完成后,必须触发语言质量 Review:

markdown
<!-- 每章节完成后执行 -->
Skill: fin-paper-draft
"language_check"

语言质量 Review 标准:

检查项标准不通过处置
AI 味关键词全文出现 ≤ 0 次重写对应段落
底气指数结论段满足 ≥ 1 项重写结论段
表格引导句每表前后 ≥ 3 句补充引导句
图表注释每图 ≥ 1 句解读补充注释
段落数符合 PAPER_OUTLINE.md 下限扩充或合并

T08:CHAPTER_STATUS.md 语言质量字段

更新 CHAPTER_STATUS.md 模板,增加语言质量追踪字段:

markdown
### introduction.tex
- **状态**: DRAFT
- **字数**: 1,234
- **语言质量检查**:
  - [ ] AI 味关键词: 0次(要求 ≤ 0)
  - [ ] 底气指数: 结论段含具体数字(是/否)
  - [ ] 表格引导句: 每表前后 ≥ 3句(是/否)
  - [ ] 图表注释: 每图 ≥ 1句解读(是/否)

VERSION_MANIFEST.md 新增语言质量维度

版本记录中新增语言质量维度:

markdown
| 版本 | 日期 | 主要变更 | 语言质量 | 作者 |
|------|------|---------|---------|------|
| draft_v1 | 2026-06-13 | 初始草稿 | AI味0/底气足 | Agent |

输出文件结构

output/fin-manuscript/draft_v{N}/
├── introduction.tex      # 引言
├── literature.tex         # 文献综述与假说
├── methodology.tex        # 数据与实证方法
├── results.tex            # 实证结果
├── robustness.tex         # 稳健性检验
├── conclusion.tex         # 结论
├── abstract.tex           # 摘要
├── references.bib         # 参考文献
├── tables/                # 表格文件
│   ├── tab1_var_def.tex
│   ├── tab2_summary.tex
│   └── ...
├── figures/               # 图表文件
│   ├── fig1_sample.pdf
│   ├── fig2_trend.pdf
│   └── ...
├── VERSION_MANIFEST.md    # 版本清单
├── CHAPTER_STATUS.md      # 章节状态
├── CONSISTENCY_CHECK.md   # 一致性检查报告
└── main.pdf              # 编译后的PDF

Checkpoint(检查点)

每个Phase完成后必须暂停,等待用户确认:

检查点内容用户操作
Phase A完成所有前置文件验证确认开始写作
每章节完成章节草稿生成Review通过后继续
Phase C完成一致性检查报告确认无问题
Phase D完成Review通过报告确认进入编译
Phase E完成PDF编译成功确认格式
Phase F完成投稿检查报告确认提交

Review重点(按章节)

章节Review重点
introduction动机是否充分?贡献是否夸大?字数是否合规?
literature文献覆盖是否完整?假设推导是否有理论支撑?
methodology识别策略是否合理?数据处理是否透明?
results表格解读是否准确?经济意义讨论是否充分?
conclusion是否过度推断?局限性是否诚实?

控制标志

标志默认值说明
STRICT_MODEtrue每章写完后必须通过检查清单才能继续
REVIEW_ROUNDS4最大review轮次
AUTO_COMPILEfalse每次章节修改后是否自动编译
VERSION_CONTROLtrue重大修改是否自动创建新版本
CHECKPOINT_ENABLEDtrue每个Phase后是否暂停等待确认

关键原则

  1. 方法论先写。先写methodology和results,有助于理清思路,避免intro空洞。
  2. 章节独立追踪。每个章节有自己的状态、字数、最后修改时间。
  3. 一致性是底线。图表编号、表编号、公式编号必须在正文和计划文件中完全一致。
  4. Review在整合后。等图表和正文都写完再做Review,避免重复Review。
  5. 版本是审计记录。每次重大修改都要记录变更原因和审查意见采纳情况。
  6. 中文论文写作顺序略有不同。中文期刊常按顺序写(引言→文献→方法→结果→结论)。
  7. 写作过程要记录。在VERSION_MANIFEST.md中实时更新状态。
  8. 终稿前必须通过submit-check。格式和匿名性是编辑初审的门槛。
  9. 占位符规则。首次生成用[coef]、[se]、[N]占位,由数据填充。
  10. 每次保存后自动更新状态。章节完成、修改、保存后立即更新CHAPTER_STATUS.md。

Frequently asked questions

What does the Fin Paper Writing AI skill do?

经济金融论文写作编排器。根据PAPER_OUTLINE.md和REFINED_DESIGN.md,编排调用fin-paper-draft(正文写作)、fin-paper-figure(图表生成)、fin-review-loop(review循环),管理版本并确保章节间的一致性。

Why use Fin Paper Writing on TypingMind?

Because you install it once and use it with any model. Fin Paper Writing is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Fin Paper Writing in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/csmar432/finai-research/tree/main/.agents/skills/fin-paper-writing. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Fin Paper Writing?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Fin Paper Writing?

As many as you like. As long as a model supports skills, you can use Fin Paper Writing with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Fin Paper Writing AI skill free?

Yes. It is published on GitHub by csmar432 under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇