Document Loader Node

Dynamically parses and structuralizes flat file formats natively inside the state execution graphs.

Functionality

Unlike the Retriever Node (which reads from indexed Document Stores), the Document Loader parses raw files in the flow run—for example a PDF attached in chat or passed through flow state. It turns those files into text your downstream nodes can use.

Configuration Parameters

• Document Parsing Engine: Select the underlying loader adapter (e.g., PDF Loader, Docx Loader, Text Loader).
• Source Path / Stream: Maps the document's dynamic binary input or localized URL utilizing {{ $flow.state.uploadedFile }}.
• Update Flow State: Append the final aggregated parsed PlainText strings into $flow.state effectively allowing downstream Agents to semantically summarize or scrape it iteratively.

Inputs & Outputs

• Inputs: Inherits runtime parameters containing binary file references.
• Outputs: Emits fully stringified textual extraction arrays explicitly ready for LLM consumption paths.