Actions6
Extract Structured Data
AI-generatedSummary
Extract structured data from a document using either an inline JSON schema or a saved LlamaExtract configuration. Supply the document as binary data or as a file ID; the node waits for extraction to finish and returns the result.
Inputs
- Configuration Mode (required) — Choose Schema (default) to provide a JSON schema, or Configuration ID to use a saved LlamaExtract configuration.
- Data Schema (required) — In Schema mode, provide the JSON schema that defines the structure for extracted data.
- Configuration ID (required) — In Configuration ID mode, provide the ID of a saved LlamaExtract configuration, which defines the schema and prompts.
- Input Type (required) — Choose Binary File (default) to use a file from the current item, or File ID to use a file previously uploaded to LlamaParse Platform.
- Input Data Field Name (required) — In Binary File mode, name the input item's binary property containing the document. Defaults to
data. - File ID (required) — In File ID mode, provide the ID of a previously uploaded file.
Output shape
A single item per input document, with a result field containing the extraction result as a pretty-printed JSON string.
The node uploads binary files before starting extraction and polls until the job completes. Extraction failures or a missing result raise an error; with Continue On Fail enabled, the item instead contains an
errorfield.