wiki-core
Skill173 starsChanged 5 months agoMITView on GitHub
Core operating rules for the student knowledge wiki. This skill should be used whenever working inside an Obsidian vault that contains a wiki/ folder with hot.md, or when the user mentions ingest, lint, review, exam-prep, or asks to build/maintain their course knowledge base. Loads the three-layer architecture, token budget rules, and bilingual page formats. Always loads first.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- giovannifrontera/academic-research-prisma-wiki-ragtheir own version, first seen Sep 30, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
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 repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .claude/CLAUDE.mdCLAUDE.md
- .cursor/rules/wiki.mdcCursor rule
- exam-prepSkill
- wiki-diagramSkill
- wiki-ingestSkill
- wiki-lintSkill
- wiki-reviewSkill
- exam-prepSkill
- wiki-coreSkill
- wiki-diagramSkill
- wiki-ingestSkill
- wiki-lintSkill
- wiki-reviewSkill
The file
Files
Student LLM Wiki — Core
你是学生知识库维护者。读 raw/,写 wiki/。永不改 raw/。
You maintain a student knowledge wiki. Read raw/, write wiki/. Never modify raw/.
三层架构 Architecture
raw/{course}/ ← Layer 1: 只读课件 Read-only slides
wiki/ ← Layer 2: 你维护的知识 Knowledge you maintain
hot.md ← 上下文缓存(≤500词) Context cache, READ FIRST
index.md ← 总目录 Master catalog
overview.md ← 跨课程综述 Cross-course synthesis
log.md ← 操作日志(追加) Append-only log
courses/ ← 课程MOC
concepts/ ← 概念页(核心) Concept pages
sources/ ← 来源页
exam-prep/ ← 练习题
raw/.manifest.json ← 去重追踪 Dedup tracker
(这些skill是 Layer 3 — 操作规则)Token预算规则(最高优先级 TOP PRIORITY)
违反这些规则会浪费用户额度。它们高于所有其他规则。 Violating these wastes the user's quota. They override all other rules.
- 每次session首先只读
wiki/hot.md(≤500词)。若含所需上下文,不再读其他页 - 需要更多上下文才读
wiki/index.md,从中定位页面 - 每次ingest最多读3-5个已有页面。超过5个=读得太广
- 局部编辑,不全文重写。改一个字段不要重写整页
- wiki页面100-300行。超300行拆分
- 批量操作时 index/hot/log 最后更新一次,不是每源更新一次
语言 Language
中英双语。标题"中文 English"。[[链接]] 用英文+连字符(如 [[Gradient-Descent]])。
概念页格式 Concept Page Format
文件: wiki/concepts/{English-Name}.md
---
tags: [concept, {domain}]
courses: [COMP6713]
confidence: low|medium|high
last_reviewed: YYYY-MM-DD
created_at: YYYY-MM-DD
---章节: 直觉 Intuition(费曼风格)→ 详细解释 Detailed → [图解 Diagram,可选,见 wiki-diagram skill] → 为什么重要 Why → 连接 Connections(含跨课程)→ 来源 Sources → 待解决 Open Questions
课程总览页格式 Course Overview Format
文件: wiki/courses/{COURSE}-overview.md,frontmatter: tags: [course-overview, {course-code}] + course: {COURSE-CODE} + updated: YYYY-MM-DD
章节: 综述 Summary(一句话)→ 概念图谱 Concept Map([[链接]] 列表)→ 薄弱环节(Dataview: confidence=low)→ 来源(Dataview: sources by course tag)
来源页格式 Source Page
文件: wiki/sources/{name}.md
---
tags: [source, {course-code}]
course: {COURSE-CODE}
ingested: YYYY-MM-DD
source_file: raw/{course}/{filename}
---章节: 关键要点(3-5) → 图表描述(标注是否建议用 Mermaid 重绘,见 wiki-diagram skill) → 新概念列表 → 更新了哪些页面
Hot Cache 内容
每次操作完更新 wiki/hot.md(≤500词):最近3个来源 / 最近概念 / 当前薄弱概念 / 待办
域规则 Domain Rules
- 数学: LaTeX + 直觉解释每个符号
- 安全 COMP4337: 攻防配对,lab验证标 ✅
- ML COMP9417: 适用场景 + 对比 + bias-variance特性
- NLP COMP6713: 架构描述 + 预训练/微调/推理区分
- 分析 INFS5730: SAS VTA步骤 + 数据类型
绝对规则 Hard Rules
- 永不改 raw/
- 一概念一页,费曼风格
- 跨课程连接是最大价值
- 不确定标 confidence:low
- Token预算规则高于一切
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.

