Information Extractor

Node catalog Artificial intelligence

On this page

Extract structured data from text

Usage and specifics

This recipe configures a model call for extraction, summary, classification, sentiment or contextual answers. Supply prompts and recipe fields; a retriever must provide context for document answers.

Technical type : ai_information_extractor. Runtime : nodulz.

Credentials and prerequisites

Create and configure the credential : ai-model

Create credentials in your nodulz instance’s Credentials area, then select them on the node. Availability depends on ownership and sharing. See the credentials guide for access and rotation.

Parameters

Names below are technical keys. Defaults are not a ready-to-run configuration: complete fields required by your operation. Paths containing [] describe list items; some fields depend on the selected mode.

Parameter	Type	Default	Values / limits
`credentialId`	`string`	-	-
`model`	`string`	`""`	-
`systemPrompt`	`string`	`""`	-
`userPrompt`	`string`	`""`	-
`temperature`	`number`	-	-; ≥ 0, ≤ 2
`maxTokens`	`number`	-	-; ≥ 1
`timeoutSeconds`	`number`	-	-; ≥ 1, ≤ 3600
`attachments`	`array`	`[]`	-
`attachments[].kind`	`enum`	`"text"`	`text`, `image_url`, `image_base64`, `document_url`, `document_base64`
`attachments[].sourceMode`	`enum`	`"manual"`	`manual`, `node`, `file`
`attachments[].source`	`string`	`""`	-
`attachments[].mimeType`	`string`	-	-
`attachments[].name`	`string`	-	-
`messages`	`array`	`[]`	-
`messages[].role`	`enum`	`"user"`	`user`, `assistant`, `system`
`messages[].content`	`string`	`""`	-
`responseFormat`	`enum`	`"text"`	`text`, `json`
`jsonSchema`	`string`	`""`	-
`categories`	`array`	`[]`	-
`contextPath`	`string`	`""`	-
`retrieverSource`	`string`	`""`	-
`minScore`	`number`	-	-
`targetLength`	`number`	-	-; ≥ 1, ≤ 100000
`targetLanguage`	`string`	`"French"`	-; max 100

Configuration help

contextPath Context source: Path to documents or excerpts with text and metadata, for example nodes.splitter.documents. Citations preserve URL, document and position. The answer abstains when context is insufficient.
retrieverSource Alternative retrieval source: Alternative context source used when the previous field is empty. Example: nodes.retriever.matches. Excerpts and metadata are preserved.
minScore Minimum score (optional): Excludes passages below this threshold. Scores depend on the retrieval method. Passages without a score are kept. Without sufficient evidence, the answer sets abstained and contains no text.
targetLanguage Target language: Translation language. Choose a suggestion, enter another language or use an expression. Instructions ask the model to preserve formatting and names.
targetLength Target length (words): Word limit requested from the model for the summary. Leave empty for a concise summary without a word limit. The exact length depends on the model.

Initial configuration

Initial catalog configuration example, to adapt before running. Keep secrets in credentials.

json

{
  "model": "",
  "systemPrompt": "",
  "userPrompt": "",
  "attachments": [],
  "messages": [],
  "responseFormat": "text",
  "jsonSchema": "",
  "categories": [],
  "contextPath": "",
  "retrieverSource": "",
  "targetLanguage": "French"
}

Verification

Test with representative data. Check node output in the inspector and logs, then use editor mapping for downstream fields. A trigger test with a simulated payload does not validate external service delivery.