Aaai Writing logo

Aaai Writing

Community
HansonLegacy
aaai-writing

AAAI 2027 论文写作全流程编排。从大纲生成到逐节撰写到终稿打磨,针对不同论文类型(理论/模型方法/基准资源/应用驱动)提供特化指导。所有规则蒸馏自 2023-2026 AAAI 获奖论文,格式约束对齐 Author Kit 2027。当用户要写 AAAI 论文、准备 AAAI 投稿、问"AAAI 论文怎么写"时使用。

Overview

PublisherHansonLegacy
RepositoryGreat-AAAI-Writing-Skills
Skill nameaaai-writing
Stars
79
Forks
1
Bundled files
157
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.

  • 157 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

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

Installation

Install the Aaai 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/HansonLegacy/Great-AAAI-Writing-Skills/tree/master
  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/HansonLegacy/Great-AAAI-Writing-Skills.git \
  .claude/skills/aaai-writing
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Aaai 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 Aaai 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 Aaai 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.

AAAI 论文写作系统

核心设计:一个父 Skill 管全局编排 + N 个子模块按需加载。你写到哪一节,就加载哪一节的专属 guidance。 所有规律均标注来源:获奖论文实例(📄)或 Author Kit 规范(📋)。

触发场景

当用户提到以下任意关键词时,加载本 skill:

  • "写 AAAI 论文" / "AAAI 写作" / "准备投 AAAI"
  • "AAAI 投稿" / "AAAI 2027"
  • "帮我写 paper"(上下文是 AAAI 会议时)
  • "生成论文大纲" / "论文结构怎么组织"

工作流总览(5 阶段)

Phase 1: 定位     → 确定论文类型 + 核心贡献形态
Phase 2: 大纲     → 生成章节大纲 + 页数预算 + story arc 验证
Phase 3: 逐节撰写  → 按顺序路由到 sections/ 子 Skill(按需加载)
Phase 4: 整合打磨  → 全篇 flow / claim-evidence 映射 / 术语一致性
Phase 5: 合规+自审 → 格式合规 + AAAI 特化自审 + 双盲检查

Phase 1: 定位(Understand)

Step 1.1: 确定论文类型

加载 modules/paper-taxonomy.md,根据用户的研究内容判定论文类型。

#类型一句话定义核心贡献形态
1理论/算法型定理、证明、复杂度分析、可判定性新定理/新界/新算法 + 严格证明
2模型/方法型新架构、新训练方法、新框架(含 LLM/VLM/小模型/视觉等)新方法/新模块 + benchmark SOTA
3基准/资源型新数据集、新 benchmark、系统性实证研究新资源 + 全面 baseline 评估 + 洞见
4应用驱动型AI 解决真实世界问题(社会影响/医疗/教育等)问题建模 + 领域验证 + 真实影响

判定完成后,将类型编号(1-4)记为全局上下文,后续所有子模块加载时自动注入对应类型的专属 guidance。

Step 1.2: 明确核心贡献

让用户回答(或从讨论中提取)3 个问题:

  1. What's new? 新任务 / 新方法 / 新理论 / 新基准 / 新应用?
  2. Why now? 为什么之前没人做/做不好?技术瓶颈是什么?
  3. So what? 做成了意味着什么?谁会关心?

Phase 2: 大纲(Outline)

Step 2.1: 生成大纲

加载 modules/outline-template.md,根据论文类型选择对应模板,生成包含以下元素的章节大纲:

  • 各章节标题(对齐 AAAI 结构规范)
  • 每节预计页数/栏数
  • 每节核心信息点(3-5 bullet)
  • Abstract 五功能句草稿

Step 2.2: 验证大纲

大纲生成后,逐项验证:

  • 页数预算:先读取具体 event 的页限;Content Appendices 计入该页限,不从通用 Author Kit 推断固定页数
  • Story arc:Abstract → Intro → Method → Experiments 逻辑链完整
  • 贡献-证据映射:每个核心贡献在 Experiments 中有对应验证
  • 图表计划:teaser、pipeline、main result table、ablation 各有位置
  • 格式合规:无禁用包/命令,章节顺序正确(详见 modules/compliance-quick.md)

Phase 3: 逐节撰写(Write)

按 AAAI 论文的标准写作顺序推进。每节只加载对应的子 Skill,不提前加载后续章节。

写作顺序与子 Skill 映射

顺序章节加载文件产出
1Titlesections/title.md1-2 个候选标题
2Abstractsections/abstract.md五步弧线摘要(150-200 词)
3Introductionsections/introduction.md5-7 段引言 + contributions 列表
4Related Worksections/related-work.md组织策略 + 段落草稿
5Methodsections/method.md各模块三元组(设计/动机/优势)
6Experimentssections/experiments.md实验设计 + 表格模板 + 分析框架
7Conclusionsections/conclusion.md总结 + 局限性 + 展望

撰写纪律

  1. 每节写完后运行该节的自查清单(每个 sections/ 文件末尾附带)
  2. 句子级参考:写作时加载 modules/sentence-craft.md——模板库 + Before/After 示例 + 节奏检查
  3. 维护跨节一致性:
    • Abstract 的每个功能句 → Intro 中对应段落
    • Intro 中 contributions bullet → Experiments 中对应结果
    • Method 中每个模块 → Experiments 中对应消融实验
  4. 术语稳定:方法名、组件名在首次定义后全文复用;缩写在 Abstract 和 Intro 各首次出现时均定义一次
  5. 引用纪律:Abstract 不含 \cite;Intro 引用尽早出现(段①/②即开始引用)

Phase 4: 整合打磨(Polish)

所有章节写完后,进行全篇一致性打磨:

4.1 Reverse Outlining

对每节执行反向提纲:

  1. 写出该节的核心论点
  2. 提取每段的 topic sentence
  3. 验证 topic sentence → 核心论点的支撑关系
  4. 删掉或重写无法映射的段落

4.2 Claim-Evidence 映射

对 Abstract 和 Introduction 中的每一个强 claim,标注:

Claim: <具体断言>
Evidence: <实验章节/表号/图号>
Status: ✅ supported / ⚠️ needs evidence / ❌ contradicted

所有 ⚠️ 和 ❌ 项必须解决才能进入 Phase 5。

4.3 术语一致性扫描

  • 方法名/组件名/缩写全文一致
  • 数学符号无重载(同一符号不表示两个不同量)
  • 所有缩写首次出现时已定义

4.4 图表最终检查

加载 modules/figure-design.md,检查:

  • Teaser 图是否在第一页可见
  • 位图分辨率 ≥ 300 dpi;矢量 PDF 不套用 DPI 阈值
  • 图内文字在最终版面中 ≥ 9pt
  • 表使用 booktabs 风格,数字对齐
  • 图注/表注格式正确

Phase 5: 合规+自审(Review & Comply)

5.1 格式合规

使用仓库内置的确定性检查器进行源文件预检;不要依赖未随本仓库安装的外部 skill:

bash
python scripts/aaai27_check.py paper.tex --stage anonymous \
  --technical-appendix unknown --checklist unknown

检查前必须声明 anonymous 或 camera-ready 阶段,并尽可能提供具体 event 的页限、Technical Appendix 政策和 Checklist 提交方式。脚本未拿到 PDF、编译日志或最终打包文件时,会把相应维度标为 NOT_CHECKED,不能据此声称整篇完全合规。写作过程中可用 modules/compliance-quick.md 做快速自查。

深度格式审查用 modules/review/03-aaai-format-compliance.md(8 类别 × AAAI 2027 专项,替代 paper-review 的 IEEE 格式检查)。

5.2 AAAI 特化自审

轻量自查 → 加载 modules/self-review.md(五维度框架)。

深度审查 → 按需加载 modules/review/ 下的 AAAI 特化 prompt:

文件用途使用时机
review/00-aaai-master-workflow.mdAAAI 10 维度完整审查(主入口)投稿前 1-2 周
review/01-aaai-section-review.md章节级专项审查(5 模块)每节写完后
review/02-aaai-red-flags.mdAAAI Reviewer 红旗速查(65 项)随时 Cmd+F
review/03-aaai-format-compliance.mdAAAI 格式合规专项扫描投稿前格式终审
review/04-aaai-double-blind.mdAAAI 双盲合规扫描投稿前 24h
review/05-aaai-reproducibility.mdAAAI Reproducibility Checklist投稿前

5.3 提交前 10 项速查

□ frenchspacing 已开启
□ 无禁用包和命令
□ 图均为 jpg/png/pdf;位图 ≥ 300 dpi,图中文字 ≥ 9pt
□ 无 overfull boxes
□ 无页码
□ Abstract 无 \cite
□ 参考文献格式正确(aaai2027.bst)
□ 若具体 event 要求:可重复性清单已按 embedded/separate 政策填写
□ Camera-ready 打包时:单 .tex 文件及 ZIP 要求已按 event/Author Kit 核对
□ 匿名投稿:PDF metadata 已清除

论文类型 → 专属 Guidance 路由

当论文类型确定后,以下模块会自动注入类型专属内容:

类型paper-types/关键差异
理论/算法型theory.md定理环境、证明草图、叙事钩子开场
模型/方法型model-method.mdpipeline 图、贡献列表、SOTA 表
基准/资源型benchmark-resource.md数据统计、评估协议、baseline 全景
应用驱动型application-driven.md问题建模、领域指标、案例分析

子模块完整索引

横向交叉模块(modules/)

文件用途加载时机
paper-taxonomy.md4 类论文分类 + 贡献形态判定Phase 1
outline-template.md大纲模板(按类型变体)Phase 2
distilled-patterns.md获奖论文蒸馏规律库(句法/段落/图表/记忆点的统计数据)Phase 3-4(全局参考)
sentence-craft.md句子级写作工艺:句法模板库 + Before/After 改写 + 句子节奏(怎么写好每句话)Phase 3-4(写作时逐句参考)
figure-design.mdAAAI 图表设计与获奖案例Phase 3(Method/Experiments 写作时)
caption-writing.mdAAAI 图注/表注写作规范:解剖结构 + 句法模板 + 病灶改写(图注与表注差异化)Phase 3-4(图表定稿时)
self-review.mdAAAI 特化自审清单(轻量版)Phase 5 快速自查
compliance-quick.md分阶段格式速查 + 内置脚本用法Phase 3-5(随时)

AAAI 特化深度审查(modules/review/)

文件用途加载时机
00-aaai-master-workflow.mdAAAI 10 维度完整审查工作流(主入口)投稿前 1-2 周深度审稿
01-aaai-section-review.md章节级专项审查(Abstract/Intro/Method/Experiments/Conclusion 5 模块)每节写完后
02-aaai-red-flags.mdAAAI Reviewer 红旗速查表(65 项,含 19 项 AAAI 新增)随时 Cmd+F 自查
03-aaai-format-compliance.mdAAAI 2027 格式合规专项扫描(8 类别,替代 IEEE 格式检查)投稿前格式终审
04-aaai-double-blind.mdAAAI 2027 双盲合规扫描(含 submission 模式特有检查)投稿前 24h
05-aaai-reproducibility.mdAAAI 2027 可重复性清单审查投稿前

AAAI 审稿模拟器(modules/review-simulator/)

文件用途加载时机
SKILL.md诊断性审稿入口 + Preflight + 4 轮工作流用户要求模拟审稿时
criteria.md7 个科学质量维度及四类论文证据标准Round 2
scoring-rubric.md0–6 Overall、类型权重、coverage、门控和独立 0–5 ConfidenceRound 2-4
review-workflow.md证据账本、交叉核验和确定性聚合步骤全程
scoring-calibration.md50 篇获奖论文的非评分写作参照;不得作为录取阈值Round 3
review-template.md含七维 scorecard、数值分和不确定性的结构化模板Round 4
common-qa.mdReviewer 常问问题库 + Rebuttal 预判Round 4
rules/aaai-review-scoring.json机器可读权重、区间和科学门控Round 4
scripts/aaai_review_score.py确定性评分聚合器Round 4

调用时机:仅当用户说"模拟审稿"、"审这篇 paper"、"预估 review 结果"、"这篇能中 AAAI 吗"时加载。 与 self-review 的区别:self-review 是"我写完了,帮我找问题"(作者视角);review-simulator 是"如果我是 AAAI PC member,我会怎么写审稿意见"(审稿人视角)。 评分边界:0–6 是内部诊断分,0–5 是当前评估的 Confidence;二者相互独立,均不是 AAAI 官方量表或录用概率。 强制尾注:每份模拟审稿必须以 Final Overall Score 和 Assessment Confidence 作为最后两个非空行结束。

章节子 Skill(sections/)

文件用途加载时机
title.md标题:Title Case + 命名规律Phase 3 第 1 步
abstract.md摘要:五步弧线 + AAAI 约束Phase 3 第 2 步
introduction.md引言:范式→骨架→撰写Phase 3 第 3 步
related-work.md相关工作:组织策略 + 切割话术Phase 3 第 4 步
method.md方法:模块三元组 + 按类型变体Phase 3 第 5 步
experiments.md实验:设计 + 表格 + 分析框架Phase 3 第 6 步
conclusion.md结论:总结 + 局限性 + 展望Phase 3 第 7 步

论文类型专属(paper-types/)

文件适用类型
theory.md理论/算法型
model-method.md模型/方法型
benchmark-resource.md基准/资源型
application-driven.md应用驱动型

与现有 Skill 的协同

本 skill(aaai-writing)是编排层,负责:
  ├── 全程编排:大纲 → 逐节 → 打磨 → 终审
  ├── AAAI 场景聚焦:不涉及 CVPR/NeurIPS/ICML 等其他会议
  ├── 论文类型特化:4 类 × 各章节 = 精准指导
  └── 获奖论文蒸馏:50 篇 AAAI 2023-2026 获奖论文的规律提取

具体写作执行时,会引用/继承:
  ├── research-paper-writing → 全局写作方法论(三元组、reverse outlining 等)
  │     https://github.com/Master-cai/Research-Paper-Writing-Skills
  ├── paper-review → modules/self-review.md 的上游参考
  │     https://github.com/FanBroWell/AI-paper-reviewer
  ├── scripts/aaai27_check.py → Phase 5 确定性源文件预检
  │     规则源:rules/aaai27-format-rules.json
  └── aaai-paper → 规范速查(保持不变,独立使用)

执行规则

  1. 按需加载:每次只加载当前 Phase/章节需要的子模块,不提前加载全部
  2. 类型注入:Phase 1 确定论文类型后,该类型作为全局上下文贯穿后续所有步骤
  3. 来源标注:每条写作建议标注 📄(获奖论文实例)或 📋(Author Kit 规范)
  4. 先大纲后动笔:不跳过 Phase 2 直接写正文
  5. 每节自查:写完一节立即自查,不要等全文写完再回头修
  6. Claim 有据:Phase 4 claim-evidence 映射中任何 unsupported claim 必须削弱或删除
  7. 用户主导:写作方向和风格由用户决策;本 skill 提供结构化指导和规律参考,不替用户做学术判断
  8. 规则分层:Author Kit 的通用要求、具体 event 政策和经验性写作建议必须分开标注;缺少 event 政策时输出 NEEDS_POLICY,缺少检查工件时输出 NOT_CHECKED

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

and 97 more files.

Frequently asked questions

What does the Aaai Writing AI skill do?

AAAI 2027 论文写作全流程编排。从大纲生成到逐节撰写到终稿打磨,针对不同论文类型(理论/模型方法/基准资源/应用驱动)提供特化指导。所有规则蒸馏自 2023-2026 AAAI 获奖论文,格式约束对齐 Author Kit 2027。当用户要写 AAAI 论文、准备 AAAI 投稿、问"AAAI 论文怎么写"时使用。

Why use Aaai Writing on TypingMind?

Because you install it once and use it with any model. Aaai 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 Aaai Writing in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/HansonLegacy/Great-AAAI-Writing-Skills/tree/master. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Aaai 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 Aaai Writing?

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

Is the Aaai Writing AI skill free?

Yes. It is published on GitHub by HansonLegacy 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 👇