Docling

DoclingTools enable an Agent to convert documents from multiple input formats (PDF, DOCX, PPTX, XLSX, HTML, images, audio, video, etc.) into output formats like Markdown, JSON, YAML, HTML, DocTags, and more using the Docling library.

Prerequisites

The following example requires the docling library.

uv pip install -U agno docling openai

# Required for the OCR example
uv pip install -U easyocr

# Optional audio processing
uv pip install -U "docling[asr]"

# Optional video processing
uv pip install -U "docling-slim[format-video]"

ffmpeg is also required for audio/video processing:

macOS: brew install ffmpeg
Ubuntu: sudo apt-get install ffmpeg
Windows: Download from ffmpeg.org

Example

The following agent converts a PDF to Markdown:

from agno.agent import Agent
from agno.tools.docling import DoclingTools

agent = Agent(
    tools=[DoclingTools(all=True)],
    description="You are an agent that converts documents from all Docling parsers and exports to all supported output formats.",
)

agent.print_response(
    "Convert to Markdown: cookbook/07_knowledge/testing_resources/cv_1.pdf",
    markdown=True,
)

OCR Configuration

Configure OCR settings for scanned PDFs or documents with embedded images.

cookbook/91_tools/docling_tools/ocr_example.py

from agno.agent import Agent
from agno.tools.docling import DoclingTools

ocr_tools = DoclingTools(
    pdf_enable_ocr=True,
    pdf_ocr_engine="easyocr",
    pdf_ocr_lang=["pt", "en"],
    pdf_force_full_page_ocr=True,
    pdf_enable_table_structure=True,
    pdf_enable_picture_description=False,
    pdf_document_timeout=120.0,
)

ocr_agent = Agent(
    tools=[ocr_tools],
    description="You are an agent that converts PDFs using advanced OCR.",
)

ocr_agent.print_response(
    "Convert to Markdown: cookbook/07_knowledge/testing_resources/cv_1.pdf",
    markdown=True,
)

Toolkit Params

Parameter	Type	Default	Description
`converter`	`Optional[DocumentConverter]`	`None`	Pre-configured Docling `DocumentConverter` instance
`max_chars`	`Optional[int]`	`None`	Maximum characters in the converted text payload; not a bound on the serialized JSON envelope
`allowed_input_formats`	`Optional[List[str]]`	`None`	Restrict accepted input formats (e.g. `["pdf", "docx"]`)
`format_options`	`Optional[Dict[Any, Any]]`	`None`	Custom format options passed to the converter
`pdf_pipeline_options`	`Optional[PdfPipelineOptions]`	`None`	Full PDF pipeline configuration object
`pdf_enable_ocr`	`Optional[bool]`	`None`	Enable OCR processing for PDFs
`pdf_ocr_engine`	`Optional[str]`	`None`	OCR engine: `auto`, `easyocr`, `tesseract`, `tesseract_cli`, `ocrmac`, `rapidocr`
`pdf_ocr_lang`	`Optional[List[str]]`	`None`	OCR language codes (e.g. `["en", "pt"]`)
`pdf_force_full_page_ocr`	`Optional[bool]`	`None`	Force OCR on every page regardless of text layer
`pdf_enable_table_structure`	`Optional[bool]`	`None`	Enable table structure recognition in PDFs
`pdf_enable_picture_description`	`Optional[bool]`	`None`	Enable picture description extraction
`pdf_enable_picture_classification`	`Optional[bool]`	`None`	Enable picture classification
`pdf_document_timeout`	`Optional[float]`	`None`	Timeout in seconds for PDF processing
`pdf_enable_remote_services`	`Optional[bool]`	`None`	Enable remote services for PDF processing
`enable_convert_to_markdown`	`bool`	`True`	Register the `convert_to_markdown` function
`enable_convert_to_text`	`bool`	`True`	Register the `convert_to_text` function
`enable_convert_to_html`	`bool`	`True`	Register the `convert_to_html` function
`enable_convert_to_html_split_page`	`bool`	`True`	Register the `convert_to_html_split_page` function
`enable_convert_to_json`	`bool`	`True`	Register the `convert_to_json` function
`enable_convert_to_yaml`	`bool`	`True`	Register the `convert_to_yaml` function
`enable_convert_to_doctags`	`bool`	`True`	Register the `convert_to_doctags` function
`enable_convert_to_vtt`	`bool`	`True`	Register the `convert_to_vtt` function
`enable_convert_string_content`	`bool`	`True`	Register the `convert_string_content` function
`enable_list_supported_parsers`	`bool`	`True`	Register the `list_supported_parsers` function
`all`	`bool`	`False`	Enable all conversion functions when set to `True`

Toolkit Functions

Function	Description
`convert_to_markdown`	Converts a document (file path or URL) to Markdown. Accepts `source`, optional `headers` for URL requests, `raises_on_error`, `max_num_pages`, and `max_file_size`.
`convert_to_text`	Converts a document to plain text. Same parameters as `convert_to_markdown`.
`convert_to_html`	Converts a document to HTML. Same parameters as `convert_to_markdown`.
`convert_to_html_split_page`	Converts a document to HTML with page-level splitting. Same parameters as `convert_to_markdown`.
`convert_to_json`	Converts a document to JSON. Same parameters as `convert_to_markdown`.
`convert_to_yaml`	Converts a document to YAML. Same parameters as `convert_to_markdown`.
`convert_to_doctags`	Converts a document to DocTags format. Same parameters as `convert_to_markdown`.
`convert_to_vtt`	Converts a document (including audio/video) to VTT subtitle format. Same parameters as `convert_to_markdown`.
`convert_string_content`	Converts raw string content (Markdown or HTML) to another format. Accepts `content`, `source_format` (default `"markdown"`), `output_format` (default `"markdown"`), and optional `name`.
`list_supported_parsers`	Lists all supported Docling input parsers and any active format restrictions.

Local inputs and optional media

Before running either PDF example, place your PDF at cookbook/07_knowledge/testing_resources/cv_1.pdf relative to the working directory, or replace the prompt's path with your own PDF path. The PDF recipe does not require the optional audio/video extras. OCR and media pipelines may download model weights on first use; see Docling media processing for the extras and model requirements.