Interface RagIngestionOptions

Chunking options and ingestion-time overrides.

interface RagIngestionOptions {
    targetDataSourceId?: string;
    duplicateHandling?: "error" | "overwrite" | "skip";
    chunkingStrategy?: {
        type: "none" | "semantic" | "fixed_size" | "recursive_character";
        chunkSize?: number;
        chunkOverlap?: number;
        strategySpecificParams?: Record<string, any>;
    };
    embeddingModelId?: string;
    userId?: string;
    personaId?: string;
    batchSize?: number;
    processAsync?: boolean;
}

Properties

targetDataSourceId? duplicateHandling? chunkingStrategy? embeddingModelId? userId? personaId? batchSize? processAsync?

Properties

Optional targetDataSourceId

targetDataSourceId?: string

Explicit target data source ID. If omitted, the augmentor falls back to the document-specified dataSourceId, category behavior defaults, or system defaults.

Optional duplicateHandling

duplicateHandling?: "error" | "overwrite" | "skip"

Behavior when a document ID already exists.

  • overwrite: replace the existing document/chunks (default).
  • skip: ignore duplicate IDs.
  • error: surface a validation error.

Optional chunkingStrategy

chunkingStrategy?: {
    type: "none" | "semantic" | "fixed_size" | "recursive_character";
    chunkSize?: number;
    chunkOverlap?: number;
    strategySpecificParams?: Record<string, any>;
}

Chunking configuration. strategySpecificParams allows pluggable implementations to carry provider-specific hints without widening the base interface each time.

Type declaration

  • type: "none" | "semantic" | "fixed_size" | "recursive_character"
  • Optional chunkSize?: number
  • Optional chunkOverlap?: number
  • Optional strategySpecificParams?: Record<string, any>

Optional embeddingModelId

embeddingModelId?: string

Embedding model identifier used when generating embeddings for this ingestion request. When omitted the augmentor consults the service config / category defaults.

Optional userId

userId?: string

Optional user identifier for auditing and personalization.

Optional personaId

personaId?: string

Optional persona identifier for personalization.

Optional batchSize

batchSize?: number

Batch size for large ingestion jobs.

Optional processAsync

processAsync?: boolean

Whether to schedule ingestion asynchronously (future enhancement hook).

Settings

Member Visibility

Theme

On This Page

targetDataSourceIdduplicateHandlingchunkingStrategyembeddingModelIduserIdpersonaIdbatchSizeprocessAsync