About Coze
Coze (扣子) is a no-code AI application development platform by ByteDance. It lets anyone — regardless of coding experience — rapidly build chatbots, AI agents, and plugins, then deploy them across social and messaging platforms. The SoMark plugin is available in the Coze plugin store and brings powerful document parsing capabilities to your agents and workflows.Adding SoMark to an Agent
Create an agent
Go to your workspace, click Project Development → + Project (top right) → Create Agent. 
Add the SoMark plugin
In the agent editor, locate the Skills section and click to add a plugin. 
Search for somark, then find
and add the somark_doc_parse plugin. 
Search for somark, then find
and add the somark_doc_parse plugin. Configure the API key
In the plugin parameter editor, enter your SoMark API key.
After entering the API key, toggle the switch on the right to enable it by default. This ensures every
request automatically uses the configured API key without requiring users to enter it manually.
No API key yet? Go to the SoMark API workbench to get your API key. Free quota does not need to be claimed and is automatically
credited to your account (500 pages/day, 2000 pages/month).
After entering the API key, toggle the switch on the right to enable it by default. This ensures every
request automatically uses the configured API key without requiring users to enter it manually.Configure other parameters
In addition to the API key, you can configure other optional parameters based on your needs, such as Output formats, Image format, Formula format, Element
structure format, and Table format. You can also use 
feature_config to control options such as enable text cross page, enable table cross page, enable
title level recognition, enable inline image, enable table image,enable image understanding, and keep header footer. If you are unsure how to configure
these options, start with the defaults and tune them one by one based on your workflow results. See Plugin Parameters below for detailed configuration descriptions.
Plugin Parameters
| Parameter | Type | Required | Description |
|---|---|---|---|
file | file | ✅ | File to parse. Supports PDF, images, Word, PPT, and Excel formats |
api_key | string | ✅ | API key in the format sk-*** |
output_formats | array | ❌ | Output formats. You can pass multiple values. If omitted, the default is ["markdown", "json"]. Supported values: json, markdown |
element_formats | object | ❌ | Element format settings that control the return format for images, formulas, tables, and chemical structures |
feature_config | object | ❌ | Feature settings. This parameter was renamed from extract_config to feature_config |
element_formats
element_formats
| Field | Type | Description |
|---|---|---|
image | string | Image return format. Supported values: url, base64, file, none |
formula | string | Formula return format. Supported values: latex, mathml, ascii |
table | string | Table return format. Supported values: markdown, html, image |
cs | string | Chemical structure return format. Currently supports image |
feature_config
feature_config
| Field | Type | Description |
|---|---|---|
enable_text_cross_page | boolean | Whether to enable cross-page text stitching. Supported values: true / false |
enable_table_cross_page | boolean | Whether to enable cross-page table stitching. Supported values: true / false |
enable_title_level_recognition | boolean | Whether to enable heading level recognition. Supported values: true / false |
enable_inline_image | boolean | Whether to return inline images. Supported values: true / false |
enable_table_image | boolean | Whether to return images inside tables. Supported values: true / false |
enable_image_understanding | boolean | Whether to enable image understanding. Supported values: true / false |
keep_header_footer | boolean | Whether to keep headers and footers. Supported values: true / false |