knowledge-graph

Original：🇨🇳 Chinese

Translated

Organize research, discussions, and exploratory content into systematic knowledge documents, or collect and organize research information about companies/products. Use this skill when users request knowledge summarization, note organization, knowledge base document generation, or structuring discussion content into formal documents. It also applies to collecting and organizing research information about AI companies, startups, and products. Even if users don't explicitly mention "knowledge graph" or "knowledge base", this skill should be used for any workflow that involves sorting scattered information into systematic documents.

2installs

Sourcelumincui/skills

Added on2026-04-26

NPX Install

npx skill4agent add lumincui/skills knowledge-graph

SKILL.md Content (Chinese)

View Translation Comparison →

Knowledge Document Organization

Organize research, discussions, and exploratory content into systematic, long-maintainable knowledge documents.

Step 1: Determine Document Type

Based on the user's needs, select the appropriate document mode:

Mode	Applicable Scenarios	Reference File
Overview	Domain cognition, concept sorting, architectural knowledge; helps readers establish structured cognition	`references/overview.md`
Technical	Technical implementation, engineering details, code design, algorithm mechanisms; helps readers understand "how to do it"	`references/technical.md`
Research	Research reports on companies/products/individuals; collect and organize external information	`references/research.md`
Paper	Reading notes for academic papers; extract core arguments, reasoning processes, research contributions and limitations	`references/paper.md`

If unsure, first determine whether the user's goal is establishing cognition (Overview mode), understanding implementation (Technical mode), collecting external information (Research mode), or understanding a paper (Paper mode).

Read the corresponding reference file and follow the guidelines within to complete the document.

Step 2: Locate Before Adding Knowledge

Regardless of the mode, before adding content to the existing knowledge base, perform the location step:

Scan the existing directory structure: List the files and directories already in the knowledge base
Find the most relevant path: Based on topic similarity, identify the most suitable location
- If there is a highly relevant existing document → Add to the corresponding section of that document
- If there is a relevant directory but no corresponding document → Create a new document under that directory
- If it is a completely new domain → Create a new directory and document
Confirm before writing: If necessary, inform the user of your judgment before starting to write

This location step avoids duplicate document creation, maintains the aggregation of the knowledge base, and allows knowledge to accumulate over time instead of becoming fragmented.