Guide · Agent skill
How to use the ljg-learn agent skill: concept anatomy in markdown
Updated · written with AI from the project's own documentation and checked against our catalogue data
The ljg-learn skill is an agent skill for Claude-style agents that takes a concept, term or idea and examines it from eight directions: history, dialectics, phenomenology, linguistics, formalization, existentialism, aesthetics and meta-philosophy. It then compresses the results into a formula, a one-sentence insight and an ASCII structure diagram. This guide is for readers who have not used agent skills before. By the end you will know how to install the skill, trigger it with a natural request, and find the markdown note it writes.
Deep Concept Anatomylijigang/ljg-skills · MIT Download the skill Quick start
- 1Download the ZIP. Download the ljg-learn skill ZIP from this page and unzip it. Confirm the folder contains SKILL.md.
- 2Place the folder. Move the folder to ~/.claude/skills/ for personal use, or to a project's .claude/skills/ folder. Alternatively, upload it under Settings → Capabilities → Skills in the Claude apps.
- 3Check the notes folder. Make sure ~/Documents/notes exists and that your agent can write to it, because the skill saves files there automatically.
- 4Send a trigger request. Ask in a form the skill matches, such as /ljg-learn 熵 or explain concept entropy.
- 5Approve the shell command. The skill runs date +%Y%m%dT%H%M%S for a timestamp. Approve it only if you accept that step.
- 6Read the report. Open the markdown file the agent reports, which uses the pattern {timestamp}--概念解剖-{概念名}__concept.md in ~/Documents/notes.
What an agent skill is and what ljg-learn makes the agent do
An agent skill is a folder containing a SKILL.md file that gives an agent extra instructions for a specific task. When the request matches the skill's description, the agent follows those instructions instead of answering from scratch.
The ljg-learn skill is a deep concept anatomist. When you ask it to explain, dissect or deeply understand a concept, term or idea, the agent first pins down the common definition and the common misunderstandings. It then works through the eight dimensions, one short pass each, looks for the shared deep structure across them, and writes a compressed insight.
- Eight dimensions: history, dialectics, phenomenology, linguistics, formalization, existentialism, aesthetics, meta-reflection
- Compressed output: a formula written as concept = ..., a one-sentence insight, and an ASCII structure diagram
- Saved result: a markdown file written to ~/Documents/notes
Install ljg-learn: download the ZIP and unzip it into a skills folder
Download the ZIP of the skill from this page and unzip it. The unzipped folder should contain SKILL.md. For a personal setup, place the folder in ~/.claude/skills/. For a single project, place it in the project's .claude/skills/ folder instead.
If you use the Claude apps, you can instead upload the skill under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/ljg-learn/
- Project install: .claude/skills/ljg-learn/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload the skill
How to trigger the skill with a natural request
You can use the slash trigger or a plain request. The skill's description lists these triggers: '解剖概念', '概念解剖', 'explain concept', 'learn concept' and '/ljg-learn'. The source's usage example uses the slash form with a single term.
Ask for a concept by name and say you want it explained or dissected. Asking for a plain definition may not match the skill, because the skill is built for deep, multi-angle exploration.
- Slash form: /ljg-learn 熵
- Plain request: explain concept entropy
- Chinese request: 解剖概念 熵
- Learn request: learn concept opportunity cost
/ljg-learn 熵Examples of what the output looks like
The source's example runs the skill on the Chinese term 熵 (entropy) and produces a markdown report. The report is organised under the headings 定锚 (anchoring), 八刀 (eight cuts), 内观 (inner view) and 压缩 (compression). Each of the eight dimensions gets its own H2 heading, such as 历史 for history and 辩证 for dialectics.
Each dimension is limited to two or three sentences. The inner view asks the agent to speak as the concept in the first person, then names the cuts that point to the same deep structure. The compression step gives you the formula, the sentence and the diagram.
- Example request: /ljg-learn 熵
- Report headings: 定锚, 八刀 (历史, 辩证, 现象, 语言, 形式, 存在, 美感, 元反思), 内观, 压缩
- Compression parts: formula, one-sentence insight, ASCII structure diagram
What ljg-learn needs and what it does on your machine
The skill needs an agent with file-writing access to ~/Documents/notes and shell access so it can run the date command for a timestamp. The source says the skill runs scripts, and it writes files. It does not use the network, according to the listing.
The skill's instructions tell the agent to run date +%Y%m%dT%H%M%S, then write a file named with that timestamp and the concept name into ~/Documents/notes. The file uses a markdown front matter block with a title, a tag and a date. Review the file path before you approve any shell command, and make sure you are comfortable with files being written to that folder.
- Requires: file-writing access to ~/Documents/notes and shell access for the date command
- Runs scripts: yes
- Uses network: no
- File name pattern: {timestamp}--概念解剖-{概念名}__concept.md
- Output format: plain markdown; the instructions forbid Org syntax
date +%Y%m%dT%H%M%STips for getting useful results
Give the skill one concept at a time, and name the domain if the word has several meanings. The skill is designed to dig into a single term, so a clear term gives a clearer anatomy.
Treat the output as interpretation rather than reference. The skill relies on the model's own reasoning and does not cite sources, so check any factual claim, such as a historical date, against a proper reference before you rely on it.
- Use a single, precise term such as 熵 or opportunity cost
- Ask for a deep or multi-angle explanation, not a quick definition
- Check historical and factual claims elsewhere, since no sources are cited
- Keep your notes folder organised, since each run adds a new file
What people use it for
- Understanding a difficult term from several angles before a seminar or exam
- Exploring the philosophical framing of an idea for an essay
- Building a personal library of concept notes in markdown
- Compressing a complex idea into one formula and one sentence for your own reference
Frequently asked questions
How do I install the ljg-learn skill?
Download the ZIP from this page and unzip it so that SKILL.md sits in a ljg-learn folder. Place that folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder. In the Claude apps, you can upload it under Settings → Capabilities → Skills.
Is ljg-learn free to use?
The skill is published under the MIT licence, as stated in its listing. This page does not list prices for any agent product you might run it in, so check the terms of that product.
What does ljg-learn need to run?
It needs an agent with file-writing access to ~/Documents/notes and shell access to run the date command for a timestamp. The listing says it runs scripts and does not use the network.
Does ljg-learn use the internet or cite sources?
The listing says it does not use the network. It relies on the model's own reasoning and does not cite sources, so verify any facts yourself.
Can I use ljg-learn for quick definitions?
It is not the best choice for that. The output is long and interpretive, so a plain question will be faster for a simple definition.
How is this different from asking an agent to explain a concept normally?
The skill follows a fixed method of eight dimensions, a compression step and a saved markdown file. A normal answer does not follow that structure or write a file to your notes folder.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use ljg-learn if you want a deep, multi-angle, philosophical view of a hard term and you are happy to keep a folder of markdown notes. Skip it for quick factual definitions, for sourced research, or if you do not want files written to ~/Documents/notes.
Download the skill