Writing built-in knowledge

Built-in knowledge lets the canvas explain the basics before any model is set up: what a core node does, what a link type carries, what a model family needs. It is written by hand in English and Russian.

#Where it is

#The shape

• A node or a link type: what (what it is), why (why it is used), connect (what goes in and out) and an optional risk (a common mistake).
• A family: what it is, which loaders it needs, and the compatibility pitfall.

#Rules

• Truth first. what and connect describe standard ComfyUI behaviour and are shown as fact, with the official documentation page when there is one (every address checked against docs.comfy.org). why and risk are advice and shown as inference. A family note is a fact only when an official tutorial backs the parts it names.
• Both languages, the same items. The Russian text in Russian, the English text without Cyrillic; the same optional risk notes in both.
• Short. A description card shows the first line; the explanations tab shows the rest.
• No secrets, no model names of chat assistants, no links outside docs.comfy.org.

#Adding a node, a type or a family

1. Add its id to the list in coreKnowledge.ts (the class name as ComfyUI spells it).
2. Add its texts to both coreKnowledge.en.ts and coreKnowledge.ru.ts.
3. Add its documentation link, or null when there is none.
4. Run npx vitest run tests/unit/coreKnowledge.test.ts: it checks every item in both languages, the language of each text, the links and how each text is marked.
5. If a wiki page explains the same idea (docs/wiki/concepts/), keep the two saying the same.

#See also

• For developers
• Writing a lesson
• Texts and languages