What Is ljg-skills

One-sentence answer: ljg-skills is an open-source Claude Code custom skill set from "Prompt God" Li Jigang. It turns complex cognitive tasks into directly callable Skill units using 20 deeply condensed cognitive tools and 2 automated workflows. The repository entry point is https://github.com/lijigang/ljg-skills .

Product Positioning

It is more like an "executable encapsulation of cognitive methodologies" than a feature-packed toolbox. The underlying design philosophy is compression: complex prompts of hundreds of words have evolved into minimalist Lisp expressions—like a cup that can hold many kinds of drinks while the container structure remains stable.

Each skill therefore corresponds to a stable type of cognitive action—writing, deconstructing a book, tracing origins, simplifying, questioning—rather than to a button in a specific application.

Core Functions

According to the official description, ljg-skills includes 20 cognitive tools and 2 automated workflows. By purpose, they can be divided into the following groups.

Content Creation

• ljg-writes: Opens up an idea like a scalpel, thinking it through layer by layer during writing.
• ljg-card: Converts content into PNG visual cards, supporting long images, infographics, comics, whiteboards, and other formats.
• ljg-present: Generates presentation slides in Takahashi or slogan style—one keyword per page, or black-and-red assertion blocks.

Learning and Research

• ljg-learn: Dissects a concept from eight dimensions—history, dialectics, phenomena, language, form, existence, aesthetics, and meta-reflection—then compresses it into one moment of insight.
• ljg-paper: Extracts the core ideas of a paper for non-academic readers, valuing understanding over critique.
• ljg-paper-river: Uses reverse reading to recursively trace earlier literature and tell the evolution of a question from its source.
• ljg-book: Deconstructs a book around questions, sorting out the author's consensus, points of shift, conclusions, and spiritual core.
• ljg-library: Explains the unique framing of a book and generates a PNG library card dated 2050.
• ljg-word: Deeply deconstructs the core meaning of an English word to find the moment of understanding.
• ljg-read: Provides three-level English translation, structural annotation, deep questioning, and cross-domain asides as a reading companion.

Thinking Tools

• ljg-rank: Given a field, finds the irreplaceable independent generators behind it, reducing phenomena to their simplest form.
• ljg-think: Drills vertically into a viewpoint or phenomenon until reaching its indivisible essence.
• ljg-plain: Rewrites any content so a bright twelve-year-old can understand it, without losing information.
• ljg-invest: Judges whether a project is an "order-creating machine" from six dimensions.
• ljg-qa: Distills the core ideas of an article, paper, or book into a Q-A chain: Q cuts to the heart, A is structured in four parts.
• ljg-roundtable: Organizes truth-seeking, structured multi-person dialectical dialogue, generating an ASCII thinking framework diagram each round.

Relationship Analysis

• ljg-relationship: Uses five-layer structural diagnosis and psychological analysis to help users "see" the real structure of a relationship.

Practical Tools

• ljg-travel: Enter a city name to automatically generate an in-depth cultural document and portable card covering historical layers, museums, ancient architecture, and archaeological discoveries.
• ljg-x-download: Downloads images and videos from an X/Twitter post to the local ~/Downloads folder in one click.
• ljg-skill-map: Scans all installed skills and renders a visual overview.
• ljg-push: Syncs local skills to both the master and md branches of a GitHub repository in one click.

Automated Workflows

• ljg-paper-flow: Read a paper, generate analysis, and create a comic card in one continuous flow.
• ljg-word-flow: Automatically generate an infographic card after deep word analysis, in one continuous flow.

How to Use ljg-skills

Step 1: Clone the Repository

Run git clone https://github.com/lijigang/ljg-skills.git%60 to obtain all skill files.

Step 2: Choose an Output Format

The default branch uses org-mode, suitable for Emacs users; the Markdown branch is suitable for Obsidian, VSCode, Notion, and other mainstream note-taking tools. The two branches have exactly the same functionality and differ only in output format. Pick one based on your editor.

Step 3: Install Skills

Run npx skills add lijigang/ljg-skills -g --all to install all skills globally; or append --skill <name> to install a single skill. The npx-based skills CLI supports global/local installation, single/all skills, and multi-skill combinations, with almost no manual configuration required.

Step 4: Install Dependencies (Optional)

If you want to use ljg-card, enter the directory and additionally run npm install && npx playwright install chromium to install screenshot dependencies.

Step 5: Verify and Trigger

In Claude Code, entering a skill name such as ljg-card will trigger it. On platforms in China, you can adapt it by migrating the Skill files.

Note: The repository root is not the skill directory. After cloning, you still need to sync the skills/ folder to Codex.

Use Cases

• Deep writing: Use the writing engine to open up an idea, the plain-language engine to simplify expression, and the content-to-card tool to generate visual cards for direct publishing.
• Academic research: Use the paper reader to extract the core, the paper-river tool to trace literature evolution, and concept deconstruction to take apart key terms.
• Knowledge management: Use the book deconstruction skill to archive books, the information question generator to turn articles into a Q-A knowledge base, and the library card to visualize them.
• Investment decisions: Use the investment analysis to judge from six dimensions whether a project is an "order-creating machine" to assist decision-making.
• Travel planning: Enter a city name to automatically generate an in-depth cultural document and portable card covering history, museums, ancient architecture, and archaeology.

Core Advantages

1. Deeply refined: Each skill has been studied and polished for half a year, serving as an encapsulated cognitive methodology rather than a pile of tools.
2. Minimal compression: Complex prompts of hundreds of words have evolved into minimalist Lisp expressions with a stable structure.
3. Dual-format output: Supports both org-mode and Markdown, covering Emacs users and the mainstream note-taking ecosystem.
4. One-command installation: The npx skills CLI supports multiple installation granularities without manual configuration.
5. Workflow chaining: Skills can be combined into pipelines—for example, paper reading followed by card generation—to automate complex cognitive tasks.
6. Platform migration: The standard SKILL.md format has become a common industry specification and can be migrated to platforms such as OpenClaw and Antigravity.

Limitations and Cautions

• Ecosystem dependence: It is released as a Claude Code custom skill set, so the experience is directly related to the capabilities of the host AI platform.
• Heavier dependencies: ljg-card relies on Playwright screenshotting and requires an additional Chromium installation.
• Branch choice cost: You need to choose between org-mode and Markdown according to your editor; choosing the wrong one can affect later use.
• Installation path pitfall: The cloned repository root is not the skill directory; you still need to manually sync skills/ to Codex.
• Output depends on input: These skills encapsulate thinking frameworks, so output quality still depends on the content you provide and the model itself.
• Information freshness: The skill count, branch strategy, and dependency commands may change as the repository updates; refer to the official site.

Pricing and Availability

The source materials for this article do not provide any pricing, subscription fees, licensing terms, or availability commitments for ljg-skills. The project's public repository is its authoritative entry point. Pricing, licensing, branch strategy, dependency versions, and other changeable details should be checked against https://github.com/lijigang/ljg-skills in real time. Do not rely on this article or other summaries for decisions.

FAQ

Q1: Does ljg-skills require payment?

The source materials do not include pricing information. Check the official repository for any pricing statement or license terms.

Q2: Do I have to use Claude Code?

It is a custom skill set designed for Claude Code, but the standard SKILL.md format has become an industry-wide specification and can be migrated to platforms such as OpenClaw and Antigravity that support the Skill system.

Q3: What is the difference between org-mode and Markdown branches?

They have exactly the same functionality; only the output format differs. The default branch is org-mode (for Emacs), and the Markdown branch is suitable for Obsidian, VSCode, and Notion.

Q4: What if a skill doesn't work after installation?

First confirm that the corresponding dependencies are installed: ljg-card requires running npm install && npx playwright install chromium. Also note that the repository root is not the skill directory; after cloning, sync skills/ to Codex.

Q5: Can I install all skills at once?

Yes. npx skills add lijigang/ljg-skills -g --all installs all skills globally; or append --skill <name> to install a single skill.

Q6: What content does ljg-card generate?

PNG visual cards, supporting long images, infographics, comics, whiteboards, and other formats.

Alternatives

The source materials do not name specific competitors, so here are only mechanism-level alternative paths:

• Change the platform, not the skill: If your primary tool is not Claude Code, consider a platform that supports the standard SKILL.md format, such as OpenClaw or Antigravity, and migrate the skill files there.
• Build your own skills: Write your own skills according to the SKILL.md specification, reproducing only the one or two workflows you actually use frequently, rather than migrating the entire set.
• Use prompts only: If you only need this occasionally, manually reproduce the thinking framework with ordinary prompts. This costs less, but loses the benefits of reusability and automated chaining.
• Single-point replacement: If you need only one type of capability, install only the corresponding single skill—for example, only ljg-writes or only ljg-card.

Official Address and Primary Sources

• GitHub repository (official): https://github.com/lijigang/ljg-skills
• Clone address: https://github.com/lijigang/ljg-skills.git