Learn Skill — AI skill for Claude Code
Agent skill that distills AI study chats into a causal-chain learning wiki — mechanical checks, a regression eval gate, evidence-driven self-iteration (RSI).
How to install Learn Skill
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open Shen-An/learn-skill and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Learn Skill does
Agent skill that distills AI study chats into a causal-chain learning wiki — mechanical checks, a regression eval gate, evidence-driven self-iteration (RSI). DSH / Codex / Claude Code. | 把 AI 学习问答沉淀为因果链学习 Wiki:机械校验 + 回归门禁 + 证据驱动的自迭代。
Alternatives in AI
- Observal — Observal is an AI agent registry with first in class observabilty and eval framework 679 ★
- Memory Forge Rs — Stop resetting satisfying AI chats — edit the memory instead 474 ★
- Prismor — Self-hosted runtime control plane for AI agents 334 ★
README
learn-skill
一个通用的 **Agent Skill**:把 AI 问答学习对话沉淀为结构化学习 Wiki。 同一份 `SKILL.md` 在 **DSH / OpenAI Codex / Claude Code** 三个 harness 通用(它们共享目录 bundle 格式)。
它解决什么问题
频繁开对话学习一个主题时,知识散落在几十个会话窗口里,回头找不到"当时是怎么想通的"。 安装本 skill 后,任何时候说一句 **"总结这次学习,做成 wiki"**,它就会:
- 按因果链组织章节(每个机制由上一个机制的局限逼出来),而不是百科式词条
- 还原你的真实提问作小节标题,保留你踩过的误解(❌ 直觉版 → ✅ 修正版)
- 术语首现必释 + 全表速查、每篇 Mermaid 流程图、自测题配推演式参考答案
- 支持增量续写:下次学到新内容,它会找到已有 wiki 继续生长,不推倒重来
产出长什么样
learning-<主题>/
├── README.md # 因果链一图流 + 目录 + 术语速查 + 复习建议
├── 01-....md # 一章 = 一个"局限 → 机制 → 代价"循环
├── 02-....md
└── ... # 每篇:真实问题标题、Mermaid 流程图、一句话总结、自测题+参考答案
参考实例:本仓库未附带,可在你自己的项目里试跑一次——用对话学完一个主题后调用即可。
安装
三个平台都遵循 `/SKILL.md` 目录 bundle 格式,把 `learning-wiki/` 整个文件夹放进对应扫描目录即可,无需编译安装:
DSH
# 用户级(所有项目可用;热发现,无需重启)
cp -r learning-wiki ~/.dsh/skills/ # 或 ~/.agents/skills/
# 项目级
cp -r learning-wiki <项目>/.dsh/skills/ # 或 <项目>/.agents/skills/
Claude Code
# 个人级(所有项目可用)
cp -r learning-wiki ~/.claude/skills/
# 项目级(可提交进仓库共享给团队)
cp -r learning-wiki <项目>/.claude/skills/
Codex
# 全局
cp -r learning-wiki ~/.codex/skills/
# 项目级见官方文档:https://learn.chatgpt.com/docs/build-skills
使用
在任意学习对话里(哪怕对话已经进行了很久):
/learning-wiki
或者直接说:`总结这次学习,做成学习 wiki` / `把这次问答整理成 wiki`。
可选指定:输出路径(默认 `./learning-<主题>/`)、只要某一部分、增量更新已有目录。
仓库结构
.
├── README.md
├── LICENSE # MIT
└── learning-wiki/ # ← 把这一层作为 skill 目录安装
├── SKILL.md # 主指令:Step 1–5 生成流程 + Step 6 触发式自迭代
├── CHANGELOG.md # 版本变更账(每次 bump 须过门禁 + 人工确认)
├── scripts/
│ └── check_wiki.py # 机械校验 E1–E5 / W1–W3:链接/围栏/编号/必备小节/悬空引用
├── evals/
│ ├── run_evals.py # 回归门禁:正样本 0 错 + 负样本 8 类埋错逐类被抓
│ └── bad-wiki/ # 负样本 fixture(故意不合格的 wiki,勿修)
├── feedback/
│ └── ledger.example.jsonl # 证据账本格式:失败信号先记账,攒够证据才迭代
├── sample-wiki/ # 示例产出 = 回归正样本(3 章,真实通过校验器)
Related Skills
Ab900
AB-900 exam prep + AI study buddy O'Reilly Live Learning course materials, distilled Microsoft Learn content,
Knowledge Distiller
AI agent skill that distills documents into temporally-aware structured knowledge graphs. Dual output: .knowle
Rsi
by ddisisto - Reads all commands and key project files to optimize AI-assisted development by streamlining the
Easel
Easel is a Codex and Claude macOS workspace for AI-assisted product design and frontend iteration. It brings a
Genesys
Open-source causal graph memory for AI agents. 89.9% on LoCoMo. MCP server with ACT-R scoring, spreading activ
Netlog AI
v0.6.0 causal console. AI-powered network log analyzer with LLM-assisted RCA (Junos / EOS / FRR / SONiC / Cumu
Related Agents
Ds Flash
DSH (DeepSeek Harness) worker on the flash tier - fast and cheap. Delegate simple, well-scoped subtasks: mecha
Ai901 Cert Buddy
AI-901 study buddy. Generates exam-realistic practice questions, brief hands-on Microsoft Foundry labs, and pe
Professor X
Use this agent when user wants to learn new concepts, understand patterns, study best practices, or get deeper