Interface: ExtractImageRequest
@kortexya/reasoninglayer / ExtractImageRequest
Interface: ExtractImageRequest
Section titled “Interface: ExtractImageRequest”Defined in: src/types/image-extraction.ts:69
Request to extract concepts from an image using a vision LLM.
Remarks
Section titled “Remarks”Sends the image directly to a vision-capable model (e.g., Qwen3-VL)
for concept extraction. If persist is true (default), the extracted
concepts are persisted as PsiTerms and synced to the homoiconic inference stores.
Properties
Section titled “Properties”filename
Section titled “filename”filename:
string
Defined in: src/types/image-extraction.ts:71
Original filename (used for media type detection via extension).
image:
string
Defined in: src/types/image-extraction.ts:73
Base64-encoded image data.
instruction?
Section titled “instruction?”
optionalinstruction?:string|null
Defined in: src/types/image-extraction.ts:75
Optional domain-specific instruction for the vision LLM.
ownerId
Section titled “ownerId”ownerId:
string
Defined in: src/types/image-extraction.ts:77
Owner ID (user who owns the extracted data).
persist?
Section titled “persist?”
optionalpersist?:boolean
Defined in: src/types/image-extraction.ts:79
Whether to persist extracted terms to the knowledge base (default: true).