Create Processor

Create a new processor in Extend, optionally cloning from an existing processor.

Create Processor

cURL

curl --request POST \
  --url https://api-prod.extend.app/v1/processors \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --data '
{
  "name": "<string>",
  "type": "<string>",
  "cloneProcessorId": "<string>",
  "config": {
    "type": "<string>",
    "baseProcessor": "<string>",
    "baseVersion": "<string>",
    "schema": {},
    "fields": [
      {
        "id": "<string>",
        "name": "<string>",
        "type": "<string>",
        "description": "<string>",
        "schema": [
          {}
        ],
        "enum": [
          {
            "value": "<string>",
            "description": "<string>"
          }
        ]
      }
    ],
    "extractionRules": "<string>",
    "advancedOptions": {
      "fixedPageLimit": 123,
      "splitMethod": "<string>",
      "splitIdentifierRules": "<string>",
      "splitExcelDocumentsBySheetEnabled": true
    },
    "classifications": [
      {
        "id": "<string>",
        "type": "<string>",
        "description": "<string>"
      }
    ],
    "classificationRules": "<string>",
    "splitClassifications": [
      {
        "id": "<string>",
        "type": "<string>",
        "description": "<string>"
      }
    ],
    "splitRules": "<string>"
  }
}
'

Example Success Response

{
  "success": true,
  "processor": {
    "object": "document_processor",
    "id": "processor_1234",
    "name": "New Invoice Processor",
    "type": "EXTRACT",
    "createdAt": "2024-03-01T12:00:00Z",
    "updatedAt": "2024-03-01T12:00:00Z"
  }
}

This endpoint allows you to create a new processor or clone an existing one. Typically processors are created and configured in the Extend Studio, but this endpoint can be used to create processors programmatically in order to sync ID’s across systems.

Body

name · string · required

The name of the new processor.

type · string · required

The type of the processor.

cloneProcessorId · string

The ID of an existing processor to clone. If provided, a new processor will be created that clones the config of this processor.

config · object

Optionally supply a config to be used when creating a processor. Any fields not supplied will use the listed defaults.

Extraction

Show properties

type · string · required

Must be "EXTRACT" for extraction processors.

baseProcessor · string

The base processor to use. For extractors, this is either "extraction_performance" or "extraction_light". See the base processor documentation for more details.

baseVersion · string

The version of the base processor to use (e.g. "4.0.0"). If this is provided, baseProcessor must be provided as well. See the processor changelog for available versions.

schema · object

The schema that defines the structure of data to extract from documents. One of schema or fields must be provided. We recommend using schema as fields is deprecated. See the extraction processor schema documentation for more details.

fields · array · deprecated

The schema that defines the structure of data to extract from documents. One of schema or fields must be provided. We recommend using schema as fields is deprecated. See the extraction processor schema documentation for more details on using the fields shape.

Show properties

id · string · required

Unique identifier for the field.

name · string · required

Human-readable name for the field.

type · string · required

Type of the field. Supported values:

• string: Text values
• number: Numeric values
• currency: Monetary values
• boolean: True/false values
• date: Date values
• array: Lists of values (requires schema)
• enum: Values from a predefined list (requires enum)
• object: Nested structure (requires schema)
• signature: Signature information

description · string · required

Detailed description of the field, including expected content and format.

schema · array

Required when type is “array” or “object”. Contains nested field definitions.

enum · array

Required when type is “enum”. List of allowed values.

Show properties

value · string · required

The enum value.

description · string · required

Description of the enum value.

extractionRules · string

Custom rules to guide the extraction process in natural language.

advancedOptions · object

Advanced configuration options.

Show properties

fixedPageLimit · number

Limit processing to a specific number of pages from the beginning of the document.

documentKind · string

Provide a hint about the document type (e.g. “invoice”, “receipt”, etc.).

keyDefinitions · string

Define specific key terms or concepts relevant to the document type.

modelReasoningInsightsEnabled · boolean

Enable model reasoning insights in the extraction results.

advancedMultimodalEnabled · boolean

Enable advanced multimodal processing for better handling of visual elements.

citationsEnabled · boolean

Enable citation information for extracted fields.

advancedFigureParsingEnabled · boolean

Enable advanced parsing of figures and diagrams in the document.

chunkingOptions · object

Options for controlling document chunking.

Show properties

chunkingStrategy · string

Strategy for chunking the document. Supported values:

• standard: Default chunking strategy
• semantic: Content-aware chunking based on document structure

customSemanticChunkingRules · string

Custom rules for semantic chunking in natural language.

pageChunkSize · number

Number of pages per chunk.

chunkSelectionStrategy · string

Strategy for selecting chunks. Supported values:

• intelligent: AI-based selection
• confidence: Select based on confidence score
• take_first: Always use first chunk
• take_last: Always use last chunk

Classification

Show properties

type · string · required

Must be "CLASSIFY" for classification processors.

baseProcessor · string

The base processor to use. For classifiers, must be "classification_performance" or "classification_light". See the base processor documentation for more details.

baseVersion · string

The version of the base processor to use (e.g. "3.2.0"). If this is provided, baseProcessor must be provided as well. See the processor changelog for available versions.

classifications · array · required

Array of possible classifications for the document.

Show properties

id · string · required

Unique identifier for the classification. We recommend lowercase, underscore-separated format.

type · string · required

Type identifier for the classification.

description · string · required

Detailed description of the classification.

classificationRules · string

Custom rules to guide the classification process in natural language.

advancedOptions · object

Advanced configuration options.

Show properties

fixedPageLimit · number

Limit processing to a specific number of pages from the beginning of the document.

context · string

The context to use for classification. Supported values: default max

advancedMultimodalEnabled · boolean

Enable advanced multimodal processing for better handling of visual elements during classification.

Splitter

Show properties

type · string · required

Must be "SPLITTER" for splitter processors.

baseProcessor · string

The base processor to use. For splitters, this can currently only be "splitter_performance". See the base processor documentation for more details.

baseVersion · string

The version of the base processor to use (e.g. "1.0.0"). If this is provided, baseProcessor must be provided as well. See the processor changelog for available versions.

splitClassifications · array · required

Array of classifications that define the possible types of document sections.

Show properties

id · string · required

Unique identifier for the split classification.

type · string · required

Type identifier for the split classification.

description · string · required

Detailed description of the document section type.

splitRules · string

Custom rules to guide the document splitting process in natural language.

advancedOptions · object

Advanced configuration options.

Show properties

fixedPageLimit · number

Limit processing to a specific number of pages from the beginning of the document.

splitMethod · string

Method to use for splitting. Supported values:

• high_precision: More accurate but potentially slower
• low_latency: Faster but potentially less precise

splitIdentifierRules · string

splitExcelDocumentsBySheetEnabled · boolean

For Excel documents, split by worksheet.

Response

success · boolean

A true or false value indicating whether the processor was created successfully or not.

processor · DocumentProcessor

A Processor object representing the newly created processor. See the DocumentProcessor object for more details.

Error Responses

success · boolean

Will be false if the request failed.

error · string

A description of the error that occurred.

Possible Common Errors

• 400 Bad Request: If the request body fails schema validation.
• 404 Not Found: If the processor to clone (cloneProcessorId) is not found.

Example Success Response

{
  "success": true,
  "processor": {
    "object": "document_processor",
    "id": "processor_1234",
    "name": "New Invoice Processor",
    "type": "EXTRACT",
    "createdAt": "2024-03-01T12:00:00Z",
    "updatedAt": "2024-03-01T12:00:00Z"
  }
}