LJG Skills - Setup Guide¶
Source: lijigang/ljg-skills (124.5K installs across 30 skills)
GitHub: lijigang/ljg-skills (7.2K stars)
Skills: 30 skills; top skills: ljg-card (6.7K), ljg-paper (6.6K), ljg-roundtable (6.5K)
Category: Productivity / Personal Knowledge Management
First Seen: skills.sh Mar 9, 2026; catalogued in the Aug 31, 2026 sweep
Quality Tier: 🟡 Trusted - publisher authority (Li Jigang, veteran Chinese tech blogger) with Gen Agent Trust Hub and Socket Pass, but Snyk carries a warning on the top skill (see Security Audit Status)
Li Jigang - a long-time Chinese tech blogger and community moderator - publishes a 30-skill personal knowledge-work suite. The skills encode his own working methods: ljg-card turns content into PNG share cards (long-form narrative, full-text layout, comic panels, and whiteboard argument modes via -l, -f, -c, -w flags), ljg-paper and ljg-paper-flow structure academic paper reading, ljg-writes and ljg-word-flow cover writing workflows, ljg-learn handles deliberate learning, ljg-think structures thinking, and ljg-invest supports investment research.
The suite is distinctive for its tightly parameterized interfaces: ljg-card declares its four modes as a parameter table with dimensions, image roles, and required read order - an unusually precise skill design. The skills are consistent bestsellers on skills.sh (top skills all 6-7K installs).
Installation¶
npx skills add lijigang/ljg-skills
Or per-skill:
npx skills add https://github.com/lijigang/ljg-skills --skill ljg-card
Prerequisites¶
| Requirement | Details |
|---|---|
| Node.js | Recent LTS for the npx skills installer |
| Language handling | SKILL.md content is Chinese; agents must process the original-language instruction text (see Limitations) |
Core Skills¶
| Skill | Installs | Use For |
|---|---|---|
| ljg-card | 6.7K | Content-to-PNG share cards: -l long narrative, -f full text, -c comic panels, -w whiteboard argument |
| ljg-paper | 6.6K | Academic paper reading and note-taking |
| ljg-roundtable | 6.5K | Multi-perspective discussion and debate structure |
| ljg-writes | 6.3K | Writing workflows |
| ljg-plain | 6.2K | Plain-text and minimal-format discipline |
| ljg-learn | 6.2K | Deliberate learning workflows |
| ljg-invest | 6.1K | Investment research structure |
| ljg-rank | 6.1K | Ranking and prioritization methods |
| ljg-travel | 6.1K | Travel planning workflows |
| ljg-word | 6.0K | Vocabulary and word-level craft |
Plus ljg-relationship, ljg-think, ljg-read, ljg-paper-flow, ljg-skill-map, ljg-paper-river, ljg-word-flow, ljg-present, ljg-qa, ljg-push, ljg-book, ljg-library, ljg-map, ljg-constraint, ljg-blind, ljg-x-download, ljg-structure, ljg-is, ljg-classic, ljg-teach.
Security Audit Status (skills.sh, verified Aug 31, 2026)¶
| Skill | Gen Agent Trust Hub | Socket | Snyk |
|---|---|---|---|
| ljg-card | Pass | Pass | Warn |
Gen Agent Trust Hub and Socket Pass; Snyk carries a warning - the reason this suite ships 🟡 Trusted rather than 🟢 Production.
CorpusIQ Use Cases¶
| Use Case | How |
|---|---|
| Share-card generation | ljg-card's parameterized card modes are a reference pattern for visual content produced from text |
| Research synthesis | ljg-paper + ljg-paper-flow structure reading pipelines for research sweeps |
| Skill design reference | The ljg-card parameter table (modes, dimensions, image roles, read order) is a model for writing precise, parameterized skills |
| Writing workflows | ljg-writes and ljg-word-flow complement CorpusIQ content production |
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations / Verification¶
- Source language is Chinese. The SKILL.md files ship in the original Chinese and are not translated; agents must handle the original-language instruction text, and downstream English output requires translation discipline. Named per the non-English-suite precedent (wecomteam, genshijin).
- The suite is one author's personal method collection; it is opinionated by design.
- Snyk warning is named in the tier above.
Verification after install:
npx skills add lijigang/ljg-skills --list # 30 skills discovered
Related¶
- Skills Catalog - full quality-tiered directory
- Skills Marketplace - more discovery batches
← Skills Catalog | Skills Marketplace → Powered by CorpusIQ