@vivantel/virage-core
    Preparing search index...

    Interface EmbeddedChunk

    interface EmbeddedChunk {
        commitHash: string;
        denseText: string;
        denseTextHash: string;
        denseVector: number[];
        embeddedAt: number;
        metadata: ChunkMeta;
        metadataGeneratorId: string;
        sourceFile: string;
        sparseText: string;
        sparseTextGeneratorId: string;
    }

    Hierarchy (View Summary)

    Index
    commitHash denseText denseTextHash denseVector embeddedAt metadata metadataGeneratorId sourceFile sparseText sparseTextGeneratorId
    commitHash: string

    Git commit hash when this chunk was generated.

    denseText: string

    Breadcrumb prefix + full chunk body — text sent to the embedding model.

    denseTextHash: string

    sha256(denseText) truncated to 16 hex chars — primary cache key.

    denseVector: number[]

    Dense embedding vector of denseText.

    embeddedAt: number

    Unix timestamp (ms) when embedding was computed.

    metadata: ChunkMeta

    Enrichment metadata.

    metadataGeneratorId: string

    Method fingerprint for metadata assembly (${name}@${version}:meta:${optsFp}). Stored per-chunk. If changed on a subsequent run, metadata is re-enriched for the affected fileset.

    sourceFile: string

    Original source file path.

    sparseText: string

    Raw chunk body without breadcrumb — used for BM25/FTS lexical search.

    sparseTextGeneratorId: string

    Method fingerprint for sparseText generation (${name}@${version}:sparse:${optsFp}). Stored per-chunk. If changed on a subsequent run, sparseText is regenerated and the FTS index is rebuilt for the affected fileset.

    Settings
    Member Visibility

    On This Page
    Properties