Content to Markdown (Kreuzberg)

The Content to Markdown plugin converts documents into markdown text using Kreuzberg, a Java library for document extraction. It supports a wide range of formats including PDF, DOCX, HTML, images (with OCR), PPTX, XLSX, and more.

Java 25 required

Kreuzberg requires Java 25 (or later) to run. If Otoroshi is running on an earlier JDK, Kreuzberg extraction will fail at runtime. Make sure your Otoroshi instance uses JDK 25+.

How it works

Kreuzberg extracts text content from documents and converts it to markdown format. For images and scanned PDFs, it uses Tesseract OCR as the OCR backend.

The plugin can process documents from three sources:

• URL — fetches the document from a remote URL and converts it
• JSON body — accepts url, content (base64), or raw document bytes
• Raw body — sends the document directly as the request body

Plugin configuration

• Plugin name: Cloud APIM - Content to Markdown
• Step: Backend call

Parameter	Type	Default	Description
maxBodySize	number	10485760 (10MB)	Maximum request body size in bytes

Usage

Via URL query parameter

curl "http://myroute.oto.tools:8080?url=https://example.com/document.pdf"

The response is the markdown content with Content-Type: text/markdown:

HTTP/1.1 200 OK
Content-Type: text/markdown; charset=utf-8
X-Source-Content-Type: application/pdf

# Document Title

This is the extracted content...

Via JSON body

curl -X POST http://myroute.oto.tools:8080 \
  -H "Content-Type: application/json" \
  -d '{
    "url": "https://example.com/document.pdf",
    "method": "GET",
    "headers": {
      "Authorization": "Bearer xxx"
    }
  }'

Or with base64-encoded content:

curl -X POST http://myroute.oto.tools:8080 \
  -H "Content-Type: application/json" \
  -d '{
    "content": "JVBERi0xLjQK...",
    "content_type": "application/pdf"
  }'

Via raw body

curl -X POST http://myroute.oto.tools:8080 \
  -H "Content-Type: application/pdf" \
  --data-binary @document.pdf

Supported formats

Kreuzberg supports all formats handled by the underlying Java libraries:

Format	MIME type	Notes
PDF	application/pdf	Text extraction + OCR for scanned pages
DOCX	application/vnd.openxmlformats-officedocument.wordprocessingml.document
PPTX	application/vnd.openxmlformats-officedocument.presentationml.presentation
XLSX	application/vnd.openxmlformats-officedocument.spreadsheetml.sheet
HTML	text/html	Converts to clean markdown
Plain text	text/plain	Passed through
Images	image/png, image/jpeg, image/tiff, etc.	OCR via Tesseract
And more...	Various	Any format supported by Kreuzberg

Workflow function

Kreuzberg is also available as a workflow function for use in Otoroshi workflows:

• Function name: extensions.com.cloud-apim.llm-extension.content_to_markdown

{
  "kind": "call",
  "function": "extensions.com.cloud-apim.llm-extension.content_to_markdown",
  "args": {
    "url": "https://example.com/document.pdf",
    "method": "GET",
    "headers": {
      "Authorization": "Bearer xxx"
    }
  },
  "result": "markdown_content"
}

Parameters

Parameter	Type	Description
url	string	URL of the document to fetch and convert
method	string	HTTP method for fetching (default: GET)
headers	object	HTTP headers as key-value pairs
content	string	Base64-encoded document content (alternative to url)
content_type	string	MIME type of the content (required when using content)

Output

{
  "content": "# Document Title\n\nExtracted markdown content...",
  "source_type": "application/pdf"
}

Agent built-in tool

Kreuzberg is also available as an agent built-in tool called content_to_markdown. Enable it in the agent's built_in_tools configuration:

{
  "built_in_tools": {
    "content_to_markdown": true
  }
}

The agent can then call the tool with either a url or content + content_type parameters. See the AI Agent node documentation for details.