vault-ops
Skill2 starsChanged 32 days agoMITView on GitHub
维护 Obsidian 长期知识库("大脑"):目录结构与页面规范、Ingest/Query/Lint(体检)/新任务 四类工作流、git 提交协议、沉淀边界与经验总结。当用户要求整理/查询/体检知识库、写概念页/项目页/问答、沉淀本次会话知识时使用。
3 other projects use a version of this.
Adopted by
Most recently taken up by
- linxule/carreltheir own version, first seen Nov 8, 2026
- olafgeibig/skillstheir own version, first seen May 16, 2026
- kv0906/pm-kittheir own version, first seen Apr 5, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- kv0906/pm-kit★ 138same name
- linxule/carrel★ 2same name
- olafgeibig/skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
- research-workflowSkill
- doc-to-mdSkill
- figure-pipelineSkill
- zotero-opsSkill
The file
Files
Obsidian 知识库维护
vault:<vault>(取值见 LOCAL-CONFIG.md;vault 本身是 git 仓库)。
权威规则 = vault 根目录的 CLAUDE.md,动手前先读它;本文件是其操作摘要。
三层架构
- Raw(只读资料层)——
Raw/目录(文献 PDF 等)与散落各处的资料。只读取,绝不修改 - Wiki(vault 的 md 页面)——AI 全权编写与维护
- Schema(规则文件)——vault 根
CLAUDE.md
目录结构
| 目录/文件 | 用途 |
index.md | 全库目录:每页一行链接 + 一句话摘要;更新页面后同步 |
log.md | 追加式流水,条目 ## [YYYY-MM-DD] ingest|query|lint|fix|setup|review | 主题(可 grep) |
收件箱/ | 快速捕捉的碎片,稍后整理 |
概念/ | 跨领域通用概念页(一概念一页) |
项目/ | 每任务/项目一个 hub 页 + 同名子目录放子页 |
文献/ | 文献卡片库(每篇一页)+ 可选的过滤视图(.base 或 Dataview) |
Raw/文献PDF/ | 只读层:ZotMoov 移入的文献 PDF(git 排除 PDF 本体) |
问答/ | 有长期价值的问答 |
模板/ | 概念页 / 项目页 / 文献卡片模板(本仓库 templates/ 有副本) |
全局矛盾清单.md | 来源冲突与错误记录 |
每次会话工作流(核心)
- 开始:读
index.md了解现状;任务前先查相关页面,避免重复劳动 - 进行中:新知识随做随记;发现来源矛盾记入
全局矛盾清单.md - 结束前三件事 + 经验总结 + git:
- 写回持久知识(新建概念页/问答页/文献卡片,或更新项目页/试验页)
- 同步
index.md摘要 + 页面 frontmatterupdated+log.md追加一行 - 经验总结(必须):有效做法 / 踩坑与修复 / 迁移提示;分流——工具·网络·自动化坑 → AI 记忆系统;领域知识与长期方法论 → vault;两条都追加
log.md(类型review) - git 提交(身份取自
LOCAL-CONFIG.md,用-c内联传入,不改动全局 git 配置):
cd "<vault>" && git add -A && \
git -c user.name="<git-name>" -c user.email="<git-email>" commit -m "<当日 log 条目主题>"(建议 .gitignore 排除 .DS_Store、.obsidian/、Raw/**/*.pdf 等大文件与本地状态)
四类专项工作流
- Ingest(消化新资料):读资料 → 提取要点 → 新建/更新概念页、项目页 → index +
log追加ingest→ 与旧页矛盾则更新 + 记矛盾清单 - Query(回答提问):先查 index 定位 → 读页面与底层资料,综合回答并标注出处 → 有长期价值则写回
问答/或概念页 - Lint(体检):用户说"体检/整理知识库"时执行(或积累约 10 次操作后主动提议):矛盾修正 / 陈旧更新 / 孤儿页补链 / 缺口补充 → 结果写入 log 并告知用户
- 新任务:建
项目/<名称>.mdhub 页(目标、状态、关键链接),过程中持续更新
页面格式约定
- YAML frontmatter:
tags、aliases、updated、source [[wikilink]]互链;正文末尾"相关:"小节;遵循 Obsidian 原生格式(callout / properties / wikilink)- 宁短勿冗:概念页约 30~60 行
- 效率铁律:用 Edit 局部改,禁止整文件重写;大动作前先用 grep/diff 低成本定位
沉淀边界
- 要沉淀:领域知识、概念、问答;试验发现、假说与验证方案;方法论教训
- 不沉淀:临时任务细节、会话过程、纯工具操作;用户偏好与跨会话行为约定(属 AI 记忆系统)
- 分工:记忆系统回答"用户是谁、怎么配合";vault 回答"世界是什么、任务学到了什么"
版本与多端
- vault 是 md 文件集合,核心流程不挑 Obsidian 版本;差异点:
.base(Bases 视图)需 Obsidian 1.9+——旧版本改用 Dataview 插件或直接不用过滤视图,其余不受影响;- properties 的图形化编辑需 1.4+(写入 YAML 本身不受限)
- 打开
.base文件后 Obsidian 会按自己规范重写该文件,编辑前先重新读取
- 多端使用(多台电脑)注意 git 分叉:每端各自提交、定期合并;PDF 等大文件不进 git
- 用户的多个独立 vault / 子库:规范相同、独立维护,不要跨库搬运
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

