Home / ly8427 / gan-zhong-xue · GitHub

gan-zhong-xue SKILL.md skillA

gan-zhong-xue from ly8427/gan-zhong-xue is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.

Install it, pinned
mdr add ly8427/gan-zhong-xue/gan-zhong-xue@git:20261021.d58560c

Pins git:20261021.d58560c in mdr.lock. Pin the fingerprint instead.

Audit grade
A, 17 of 17 checks passed
Kind
SKILL.md skill
Repository
ly8427/gan-zhong-xue
Path
skills/gan-zhong-xue/SKILL.md
Versions recorded
6
Latest version
git:20261021.d58560c, committed Oct 21, 2026
First version recorded
Jul 8, 2026
Size
66,991 bytes
Repository stars
11
License
MIT

在你干真实开发时（边干边挖：搞懂 AI 正在做的这步，掌控方向）或干完提交后（事后挖： 回头搞懂一个甩手没懂的改动），就地陪你走一轮"先让你猜、再揭示、逼到原理"的回合制， 把黑盒变成你真懂、能复述、能迁移的东西。治的是 vibe-coding 的病：交付物在增加， 理解没跟上。触发词："干中学"、"帮我搞懂 AI 正在做的这步"、"我刚才那段没懂"、 "这个 commit 我说不清"、"帮我真正搞懂刚才做的"；掌舵（任务中途）："等等这步我没懂"、 "为什么这么改"、"这步我没跟上"。 English triggers: "learning by doing", "help me actually understand what you just did", "I didn't really get that last change", "I can't explain this commit"; mid-task: "wait, I didn't follow that step", "why did you change it this way". Output…

Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.

What the file says

# 干中学 (Learning by Doing) · v1.0.8（真机实证版）

> 把"一次真实的、你在场的深渊穿越"复现出来的 skill。
> 引擎在 2026-07-03~06 五个晚上里被真人逐轮压测 + 四轮独立审查打磨；
> **v1 是把 07-06 真机实测暴露的问题补齐后的硬化版；v1.0.4 修了"挖太浅 / 问三下就收尾"——引入议程图 + 逐节点下潜到可迁移定律（借 gstack 的"先分解再挖 / 议程不许静默跳过 / 推一次再推一次"）。**
> **v1.0.5 修"挖到底仍是模型自报、没客观裁判"+ "边挖边冒的新问题只能丢 pending 断心流"——给到底判据接客观裁判（用户答底部验证题）、把学习型岔路从"中断"变成"当场深度验证"、并加详略判据（铁律9）。**
> **v1.0.6 修"代码事实虚构"——主 agent 没读过源码行就凭通用知识断言"代码怎么工作"，被用户当场拆穿。新增铁律10（代码断言铁律）：没读过就闭嘴去读，断言必须带路径+行号+原文秒验锚。真机实证：这是 skill 迄今最严重的信任崩坏。**
> **v1.0.7 修"越问越不懂"（issue #2）——挖着挖着前置递归（架构→模板→语法→指针→∞），用户被拖离主线。A 类岔路加第二判两步测试（浅层语义够撑当前靶+深入不改变对靶的解释→外部前置搭桥放行；否则进图挖），map 新增 ⏸外部前置格（靶ID+关联节点，关系记录、先按靶调度、靶ID字面对应、随靶失效），补全学习边界（纵向别钻到晶体管、横向别追到整个语言体系）。**
> **v1.0.8 修"委托型深渊"（issue #3）——用户只下高层指令、规则/协议/裁决文本由 agent 之间生成时，"先猜"的地基（用户在场过：写过/看过）塌了：让用户猜一本没读过的书 = 撞墙（真机复现：三轮两撞墙，用户原话"完全让我在乱猜啊"；仅补上材料原文，用户立即自己产出核心因果）。新增铁律11（委托靶，内容先行）+ 第 2 步入口发牌门（先问"原文当时看过吗"，宁可问不可猜；没看过先展示+共读原文再开问，发牌只给原文不预讲因果）+ 第 0 步协作工件靶类（无 diff，冷读输入换材料文本）。**
> **本文件自包含，工作机上只需要这一个文件**（地图 `~/.gan-zhong-xue/` 首次跑自建）。
> **哲学锚**：「知之为知之，不知为不知，是知也」/「绝知此事要躬行」/「少就是多」。
> **语言**：输出跟随用户——中文用户用中文、英文用户用英文；规则本体与触发逻辑不变。

---

## 隐私铁则（最优先，违反即失败）

这个 skill 跑在用户**真实工作代码**上（可能是公司代码）。硬规则：
…

Read the whole file at its exact version.

How to install

The exact contents, by fingerprint
mdr add ly8427/gan-zhong-xue/gan-zhong-xue@sha256:99618c09b461af81

The line at the top of this page pins the label git:20261021.d58560c, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.

Badge

mdr badge

[![mdr](https://markdownregistry.com/badge/art_7j7mgphksgao7tno.svg)](https://markdownregistry.com/a/art_7j7mgphksgao7tno)

2 badge image fetches in 30 days. What the badge shows.

Versions

The registry has recorded 6 versions of this file, between Jul 8, 2026 and Oct 21, 2026, about one every 21 days.

versioncommittedcommitsizeaudit
git:20261021.d58560c latest2026-10-21 d58560c 66,991 BA view · diff
git:20260913.62d3d2d2026-09-13 62d3d2d 62,658 BA view · diff
git:20260809.df88b032026-08-09 df88b03 52,881 BA view · diff
git:20260803.fb79fa22026-08-03 fb79fa2 47,804 BA view · diff
git:20260723.f4c3f662026-07-23 f4c3f66 27,258 BA view · diff
git:20260708.2dbdaf42026-07-08 2dbdaf4 22,155 BA view

Audit of the latest version

A  17 of 17 checks passed. Fixed automatic checks with no AI involved, so the same checks give the same answer every run.
  • pass: Frontmatter block present
  • pass: Frontmatter declares a name
  • pass: Frontmatter declares a description
  • pass: Size between 200 bytes and 200 KB (66991 bytes)
  • pass: No zero-width or bidi control characters
  • pass: No instruction hidden inside an HTML comment
  • pass: No link to an exfiltration or paste host
  • pass: No credential-shaped string
  • pass: No instruction to send local credentials anywhere
  • pass: No text hidden with inline styles
  • pass: No prompt-injection phrasing
  • pass: No curl or wget piped into a shell
  • pass: No recursive delete of root, home or parent
  • pass: No instruction to read or print local credentials
  • pass: No base64 blob over 200 characters
  • pass: No link to a raw IP address
  • pass: No script tag

Source

GitHub

ly8427/gan-zhong-xue · 11 stars · license MIT · pushed 2026-10-21 · branch main

API

GET https://markdownregistry.com/api/v1/artifacts/art_7j7mgphksgao7tno
GET https://markdownregistry.com/api/v1/resolve?ref=ly8427/gan-zhong-xue/gan-zhong-xue
GET https://markdownregistry.com/api/v1/blob/99618c09b461af8120618842388f5d95b2cc618d08ef55afd22d61e478234e6b

Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.

Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.

Browse by kind, by grade A, or by owner.