Document¶

Document ¶

Document(source=None, text=None, **kwargs)

Natural language interface for documents.

Supports PDF, DOCX, PPTX, XLSX, HTML, JSON, MD, TXT, XML, YAML, RTF formats. Documents are automatically chunked for efficient processing.

PARAMETER	DESCRIPTION
`source`	Path to document file. Mutually exclusive with text. TYPE: `str` DEFAULT: `None`
`text`	Document content as text. Mutually exclusive with source. TYPE: `str` DEFAULT: `None`

ask(prompt, model=None, context=None, output_type=None, stream=False)

Ask natural language questions and get structured responses.

PARAMETER	DESCRIPTION
`prompt`	Natural language question or instruction. TYPE: `str`
`model`	AI model to use (e.g., 'openai:gpt-4', 'anthropic:claude-3-5-sonnet'). TYPE: `str` DEFAULT: `None`
`context`	Additional business context for better responses. TYPE: `str` DEFAULT: `None`
`output_type`	Pydantic model for structured responses. TYPE: `BaseModel` DEFAULT: `None`
`stream`	Show live AI reasoning in terminal. TYPE: `bool` DEFAULT: `False`

RETURNS	DESCRIPTION
`Any`	Response matching the requested output type.

tools()

Available tool methods for document operations.

RETURNS	DESCRIPTION
`list[callable]`	Methods for document reading with intelligent chunking.

instructions(context=None)

Generate system instructions for AI agents.

PARAMETER	DESCRIPTION
`context`	Additional business context to include in instructions. TYPE: `str` DEFAULT: `None`

RETURNS	DESCRIPTION
`str`	System instructions for AI interaction with this datasource.