Interface CorpusChunk

A chunk of corpus content with optional pre-computed embedding. Used during corpus ingestion into the vector store.

interface CorpusChunk {
    id: string;
    content: string;
    heading: string;
    sourcePath: string;
    embedding?: number[];
}

Properties

id: string

Unique identifier for the chunk.

content: string

The text content of the chunk.

heading: string

Section heading or title the chunk belongs to.

sourcePath: string

File path or document source path this chunk was extracted from.

embedding?: number[]

Pre-computed embedding vector. When present, the ingestion pipeline can skip embedding generation for this chunk.