refactor(docs): phase 2 — CommunityShare 下 AI 内容进一步归并到 ai/

延续 PR 初版的语义化重组思路，继续把散落在 CommunityShare 下的纯 AI 内容
移到 ai/ 主目录对应主题下。新增 4 个 docId 的迁移：

- CommunityShare/Geek/leworldmodel (JEPA 世界模型论文摘要)
    → ai/foundation-models/world-models/leworldmodel
- CommunityShare/Amazing-AI-Tools/perplexity-comet (AI 浏览器介绍)
    → ai/misc-tools/perplexity-comet
- CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms
  (arXiv prompt 技巧论文摘要) → ai/foundation-models/prompt-engineering/
- CommunityShare/Personal-Study-Notes/Reinforcement-Learning/ppo
    → ai/reinforcement-learning/ppo

新增 3 个 ai/ 下子目录：world-models / prompt-engineering / reinforcement-learning

顺手收拾的：
- ppo.md 原本是 tag-one 占位符 → [reinforcement-learning, ppo, policy-gradient, gae]
  （和 PR #308 里对 ppo 的 tag 选择完全一致，merge 顺序无所谓）
- leworldmodel 原本 tag 是中文 "世界模型" → kebab-case 英文
  [world-models, jepa, representation-learning, embodied-ai]
- 清理空目录 CommunityShare/Personal-Study-Notes/Reinforcement-Learning/
  和 Personal-Study-Notes/

next.config.mjs 补 4 条 301，累计 8 条。statusCode:301 继续用显式 301
（和 PR / commit 口径一致）。

与 PR #308 (feat/docs-tag-cleanup) 的冲突风险：
本 PR 已经移走 ppo.md 并应用了和 PR #308 完全相同的 tag 更新。merge 顺序：
- 若本 PR 先合：PR #308 rebase 时会发现 ppo 老路径不存在，它对 ppo 的
  tag 修改会变 "deleted by us / modified by them" 冲突，需要 drop
  PR #308 中 ppo 那条 hunk（因为本 PR 已处理）
- 若 PR #308 先合：本 PR rebase 时 ppo 在老路径已有新 tags，git mv
  自然工作，无冲突

推荐：**本 PR 先合**，然后 PR #308 rebase 时丢掉 ppo 相关改动。

docId 守恒（4 份原文 + 3 份翻译 = 7 个文件全部 grep 对齐）：
- boo70qqm8nos8b0q9h7zjrki (leworldmodel)
- eej2awin6irhbdgcy8vvs3xb (perplexity-comet)
- l6eepr5ctjgrhdgupy3twr1t (prompt-repetition)
- zf8zk108oqbsg56xjyqb5txk (ppo)

Reviewer subagent LGTM，无阻塞问题。

Author: github-actions[bot]

…epetition-improves-non-reasoning-llms.md …epetition-improves-non-reasoning-llms.mdapp/docs/CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms.md renamed to app/docs/ai/foundation-models/prompt-engineering/prompt-repetition-improves-non-reasoning-llms.md app/docs/CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms.md renamed to app/docs/ai/foundation-models/prompt-engineering/prompt-repetition-improves-non-reasoning-llms.md

@@ -3,7 +3,10 @@ title: LeWorldModel
 description: Stable End-to-End Joint-Embedding Predictive Architecture from Pixels
 date: "2026-04-08"
 tags:
-  - 世界模型
+  - world-models
+  - jepa
+  - representation-learning
+  - embodied-ai
 docId: boo70qqm8nos8b0q9h7zjrki
 ---
 

…tition-improves-non-reasoning-llms.zh.md …tition-improves-non-reasoning-llms.zh.mdapp/docs/CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms.zh.md renamed to app/docs/ai/foundation-models/prompt-engineering/prompt-repetition-improves-non-reasoning-llms.zh.md app/docs/CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms.zh.md renamed to app/docs/ai/foundation-models/prompt-engineering/prompt-repetition-improves-non-reasoning-llms.zh.md

@@ -3,7 +3,10 @@ title: LeWorldModel
 description: 从像素端到端稳定训练的联合嵌入预测架构
 date: "2026-04-08"
 tags:
-  - 世界模型
+  - world-models
+  - jepa
+  - representation-learning
+  - embodied-ai
 docId: boo70qqm8nos8b0q9h7zjrki
 lang: zh
 translatedFrom: en

…docs/CommunityShare/Geek/leworldmodel.md …tion-models/world-models/leworldmodel.mdapp/docs/CommunityShare/Geek/leworldmodel.md renamed to app/docs/ai/foundation-models/world-models/leworldmodel.md app/docs/CommunityShare/Geek/leworldmodel.md renamed to app/docs/ai/foundation-models/world-models/leworldmodel.md

@@ -3,7 +3,10 @@ title: PPO
 description: ""
 date: "2025-10-03"
 tags:
-  - tag-one
+  - reinforcement-learning
+  - ppo
+  - policy-gradient
+  - gae
 docId: zf8zk108oqbsg56xjyqb5txk
 ---
 

…s/CommunityShare/Geek/leworldmodel.zh.md …n-models/world-models/leworldmodel.zh.mdapp/docs/CommunityShare/Geek/leworldmodel.zh.md renamed to app/docs/ai/foundation-models/world-models/leworldmodel.zh.md app/docs/CommunityShare/Geek/leworldmodel.zh.md renamed to app/docs/ai/foundation-models/world-models/leworldmodel.zh.md

@@ -64,6 +64,30 @@ const config = {
         destination: "/docs/ai/projects/ai-town",
         statusCode: 301,
       },
+      // feat/docs-reorg-rag-projects phase 2：继续把 CommunityShare 下的 AI 内容归并到 ai/
+      {
+        source: "/docs/CommunityShare/Geek/leworldmodel",
+        destination: "/docs/ai/foundation-models/world-models/leworldmodel",
+        statusCode: 301,
+      },
+      {
+        source: "/docs/CommunityShare/Amazing-AI-Tools/perplexity-comet",
+        destination: "/docs/ai/misc-tools/perplexity-comet",
+        statusCode: 301,
+      },
+      {
+        source:
+          "/docs/CommunityShare/Amazing-AI-Tools/prompt-repetition-improves-non-reasoning-llms",
+        destination:
+          "/docs/ai/foundation-models/prompt-engineering/prompt-repetition-improves-non-reasoning-llms",
+        statusCode: 301,
+      },
+      {
+        source:
+          "/docs/CommunityShare/Personal-Study-Notes/Reinforcement-Learning/ppo",
+        destination: "/docs/ai/reinforcement-learning/ppo",
+        statusCode: 301,
+      },
     ];
   },
   async rewrites() {