https://cygnusyang.github.io/posts/01-ai-%E5%B7%A5%E5%85%B7%E4%B8%8E%E6%99%BA%E8%83%BD%E4%BD%93/08-academic-research-skills/03-architecture/技术 × 产品 × 机器人Hugo 0.160.1 & FixIt v0.4.6-20260517062719-ed8ec5c2zh-CNruohuyang@163.com (Cygnus Yang)ruohuyang@163.com (Cygnus Yang)Sat, 23 May 2026 00:00:00 +0800https://cygnusyang.github.io/posts/01-ai-%E5%B7%A5%E5%85%B7%E4%B8%8E%E6%99%BA%E8%83%BD%E4%BD%93/08-academic-research-skills/03-architecture/05-pipeline-architecture/Sat, 23 May 2026 00:00:00 +0800ruohuyang@163.com (Cygnus Yang)https://cygnusyang.github.io/posts/01-ai-%E5%B7%A5%E5%85%B7%E4%B8%8E%E6%99%BA%E8%83%BD%E4%BD%93/08-academic-research-skills/03-architecture/05-pipeline-architecture/<p>ARS 的核心不是任何一个单独的 Skill，而是 <strong>Academic Pipeline</strong>——一个 10 阶段的全流程编排器，将 Deep Research（深度研究）、Academic Paper（论文撰写）、Academic Paper Reviewer（论文审查）三个独立技能串联成一个端到端的学术生产管线。</p> <p>本文从宏观到微观，解析 pipeline 的流程、阶段矩阵、数据访问层级和 Skill 依赖关系。</p> <h2 class="heading-element" id="10-阶段全流程"><span>10 阶段全流程</span> <a href="#10-%e9%98%b6%e6%ae%b5%e5%85%a8%e6%b5%81%e7%a8%8b" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><pre><code>用户输入 ↓ Stage 1: RESEARCH → RQ Brief &#43; 方法论蓝图 ↓ Stage 2: WRITE → 大纲 &#43; 初稿 ↓ Stage 2.5: INTEGRITY → 7 类失败模式检查（诚信闸门 ①） ↓ Stage 3: REVIEW → 5 位审稿人 &#43; 编辑决定 ↓ ├── Accept ──────────────────────→ Stage 4.5 ├── Minor/Major → Revision Coaching → Stage 4: REVISE │ ↓ │ Stage 3&#39;: RE-REVIEW │ ↓ │ Stage 4&#39;: RE-REVISE（终版） │ ↓ └── Reject ──────────────────────→ 结束 Stage 4.5: FINAL INTEGRITY → 零容忍重新验证（诚信闸门 ②） ↓ Stage 5: FINALIZE → 格式转换（MD/DOCX/LaTeX/PDF） ↓ Stage 6: PROCESS SUMMARY → 过程记录 &#43; AI 自我反思 ↓ 完成</code></pre><h3 class="heading-element" id="关键设计特征"><span>关键设计特征</span> <a href="#%e5%85%b3%e9%94%ae%e8%ae%be%e8%ae%a1%e7%89%b9%e5%be%81" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h3><p><strong>两条回路</strong>：</p>https://cygnusyang.github.io/posts/01-ai-%E5%B7%A5%E5%85%B7%E4%B8%8E%E6%99%BA%E8%83%BD%E4%BD%93/08-academic-research-skills/03-architecture/06-four-skills-overview/Sat, 23 May 2026 00:00:00 +0800ruohuyang@163.com (Cygnus Yang)https://cygnusyang.github.io/posts/01-ai-%E5%B7%A5%E5%85%B7%E4%B8%8E%E6%99%BA%E8%83%BD%E4%BD%93/08-academic-research-skills/03-architecture/06-four-skills-overview/<p>ARS 由 4 个独立的 Claude Code Skill 组成，每个都可以单独使用，也都可以被 Academic Pipeline 编排为端到端流程。本文逐一介绍它们的定位、核心能力、输入输出和协作关系。</p> <h2 class="heading-element" id="概览"><span>概览</span> <a href="#%e6%a6%82%e8%a7%88" class="heading-mark"> <svg class="octicon octicon-link" viewBox="0 0 16 16" version="1.1" width="16" height="16" aria-hidden="true"><path d="m7.775 3.275 1.25-1.25a3.5 3.5 0 1 1 4.95 4.95l-2.5 2.5a3.5 3.5 0 0 1-4.95 0 .751.751 0 0 1 .018-1.042.751.751 0 0 1 1.042-.018 1.998 1.998 0 0 0 2.83 0l2.5-2.5a2.002 2.002 0 0 0-2.83-2.83l-1.25 1.25a.751.751 0 0 1-1.042-.018.751.751 0 0 1-.018-1.042Zm-4.69 9.64a1.998 1.998 0 0 0 2.83 0l1.25-1.25a.751.751 0 0 1 1.042.018.751.751 0 0 1 .018 1.042l-1.25 1.25a3.5 3.5 0 1 1-4.95-4.95l2.5-2.5a3.5 3.5 0 0 1 4.95 0 .751.751 0 0 1-.018 1.042.751.751 0 0 1-1.042.018 1.998 1.998 0 0 0-2.83 0l-2.5 2.5a1.998 1.998 0 0 0 0 2.83Z"></path></svg> </a> </h2><table> <thead> <tr> <th>Skill</th> <th>版本</th> <th>Agent 数</th> <th>模式数</th> <th>一句话定位</th> </tr> </thead> <tbody> <tr> <td><code>deep-research</code></td> <td>v2.9.2</td> <td>13</td> <td>7</td> <td>上游研究引擎：从问题到证据</td> </tr> <tr> <td><code>academic-paper</code></td> <td>v3.1.1</td> <td>12</td> <td>10</td> <td>下游出版引擎：从证据到论文</td> </tr> <tr> <td><code>academic-paper-reviewer</code></td> <td>v1.9.0</td> <td>7</td> <td>6</td> <td>多视角同行评议：从论文到决策</td> </tr> <tr> <td><code>academic-pipeline</code></td> <td>v3.7.0</td> <td>4</td> <td>1 (+1 resume)</td> <td>全流程编排器：从调度到品质保障</td> </tr> </tbody> </table> <p>25 种模式，统一由 <a href="https://github.com/Imbad0202/academic-research-skills/blob/main/MODE_REGISTRY.md" target="_blank" rel="external nofollow noopener noreferrer">MODE_REGISTRY.md<i class="fa-solid fa-external-link-alt fa-xs ms-1 text-secondary" aria-hidden="true"></i></a> 管理。</p>