> ## Documentation Index
> Fetch the complete documentation index at: https://docs.abbyy.com/llms.txt
> Use this file to discover all available pages before exploring further.

# JSON Output

> JSON output file types produced by ABBYY Vantage skills: file names, export options, schemas, and which Document, OCR, or Process skill creates each.

The following table describes the available JSON output file types. The **Skill type** column lists the skill types that can generate each file type. For the full list of Vantage output formats, see [Output file types](/vantage/developer/output/output-file-types).

| File type         | Export option                                               | Description                                                                                                                                                                                                                                                                                                                                                                                                                                                   | Skill type                                | File name                                                                                   | JSON schema                                                               |
| :---------------- | :---------------------------------------------------------- | :------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ | :---------------------------------------- | :------------------------------------------------------------------------------------------ | :------------------------------------------------------------------------ |
| **Fields (JSON)** | **Values, metadata, and field structure for each document** | A file containing classification results (if Classification skill was applied), full data extraction results (if Document skill was applied), rule errors, document and transaction registration parameters (if available), and other metadata.                                                                                                                                                                                                               | Document<br />Classification<br />Process | `<Applied_skill_name>.json` \*                                                              | [JSON schema](/vantage/developer/output/json/json-schema)                 |
| **Fields (JSON)** | **Values only**                                             | A file containing field values and rule errors.                                                                                                                                                                                                                                                                                                                                                                                                               | Document<br />Process                     | `<Applied_skill_name>_fields.json` \*                                                       | Depends on the Document skill used. Currently not available for download. |
| **Text (JSON)**   | **Text only**                                               | This mode is suitable for extracting all text from the input image, including small text areas of low quality. The document appearance and structure are ignored, pictures and tables are not detected. It is designed for the situations when you need to retrieve the data from the image for some further processing on your side, such as extracting data from bills, receipts or invoices. Selecting this mode makes export to DOCX and XLSX impossible. | OCR<br />Process                          | OCR: `<First_source_file_name>.json` \*\*<br />Process: `<Applied_skill_name>_text.json` \* | [OCR skill schema](/vantage/developer/output/json/ocr/ocr-json-schema)    |
| **Text (JSON)**   | **Preserve document structure**                             | This mode is focused on retaining the original document structure and appearance, including font styles, pictures, background color, etc., and is more focused on documents like agreements, contracts, specifications.                                                                                                                                                                                                                                       | OCR<br />Process                          | OCR: `<First_source_file_name>.json` \*\*<br />Process: `<Applied_skill_name>_text.json` \* | [OCR skill schema](/vantage/developer/output/json/ocr/ocr-json-schema)    |

**`* <Applied_skill_name>`** will be one of the following:

* The name of the skill if a transaction was created for a Document or Classification skill.
* If a transaction was created for a Process skill:
  * the name of the last Document skill applied to the document
  * the name of the last Classification skill applied to the document if no Document skills were applied
  * "Unknown" if no Document and Classification skills were applied, while at least one of them exists in the Process skill flow.

`** <First_source_file_name>` will be the name of the first file used to assemble the document, such as "IMG\_12234".

The Document skill always generates both **Fields (JSON)** files. Process skill settings allow you to select one **Fields (JSON)** file and/or one **Text (JSON)** file. OCR skill settings allow you to select one **Text (JSON)** file.

If a transaction fails, you can get information about the error in JSON format via the Vantage API. The JSON string contains information about the transaction, the error message, as well as names and identifiers of all source files in the transaction. This string will also be exported to the *Error.json* file in the output shared folder if you configure the Output activity of a Process skill accordingly.