chunkAndEmbedAllMessages

chunkAndEmbedAllMessages(ctx: StorageOperationsContext, options: MemoryRetrievalEmbeddingOptions & ChunkingOptions, filter?: object): Promise<number>

Defined in: src/lib/memoryRetrieval/embeddings.ts:331

Chunk and embed all messages without embeddings/chunks in the database. Uses chunking for long messages, whole-message embedding for short ones.

Parameters

Parameter	Type	Description
`ctx`	`StorageOperationsContext`	Storage operations context
`options`	`MemoryRetrievalEmbeddingOptions` & `ChunkingOptions`	Embedding and chunking options
`filter?`	`object`	Optional filter for which messages to embed
`filter.conversationId?`	`string`	Only embed messages from this conversation
`filter.minContentLength?`	`number`	Minimum content length to embed (default: 30). Shorter messages are skipped.
`filter.rechunkExisting?`	`boolean`	Re-chunk messages that have whole-message embeddings but no chunks
`filter.roles?`	(`"user"` \| `"assistant"`)[]	Only embed messages with these roles

Returns

Promise<number>

Number of messages embedded