kreuzberg-crewai

Kreuzberg document extraction tools for CrewAI agents.

Extract text and metadata from 88+ document formats — PDF, DOCX, XLSX, HTML, images with OCR, and more — directly from your CrewAI agents.

Installation

pip install kreuzberg-crewai

Quick Start

from crewai import Agent, Crew, Task

from kreuzberg_crewai import KreuzbergExtractTool

tool = KreuzbergExtractTool()

agent = Agent(
    role="Document Analyst",
    goal="Extract and analyze document content",
    backstory="You are an expert at reading and understanding documents.",
    tools=[tool],
)

task = Task(
    description="Extract the content from report.pdf and summarize the key findings.",
    expected_output="A summary of the key findings in the report.",
    agent=agent,
)

crew = Crew(agents=[agent], tasks=[task])
result = crew.kickoff()

Tools

KreuzbergExtractTool

Extracts text content from a document file.

Parameters:

from kreuzberg_crewai import KreuzbergExtractTool

tool = KreuzbergExtractTool()

# The agent calls this automatically, but you can also call it directly:
content = tool._run(file_path="report.pdf", output_format="markdown")

KreuzbergExtractMetadataTool

Extracts metadata (title, authors, dates, page count, format-specific details) from a document file.

Parameters:

from kreuzberg_crewai import KreuzbergExtractMetadataTool

tool = KreuzbergExtractMetadataTool()

metadata = tool._run(file_path="report.pdf")
# title: Annual Report 2025
# authors: ['John Doe']
# page_count: 42
# pdf_version: 1.7

Agent Example

Using both tools together:

from crewai import Agent, Crew, Task

from kreuzberg_crewai import KreuzbergExtractMetadataTool, KreuzbergExtractTool

extract_tool = KreuzbergExtractTool()
metadata_tool = KreuzbergExtractMetadataTool()

agent = Agent(
    role="Research Assistant",
    goal="Read documents and extract useful information",
    backstory="You help researchers by reading and analyzing documents.",
    tools=[extract_tool, metadata_tool],
)

task = Task(
    description=(
        "First, check the metadata of research-paper.pdf to find the authors and date. "
        "Then extract the full content in markdown format and list the key conclusions."
    ),
    expected_output="Authors, date, and key conclusions from the paper.",
    agent=agent,
)

crew = Crew(agents=[agent], tasks=[task])
result = crew.kickoff()

Supported Formats

Kreuzberg supports 88+ file formats:

• Documents: PDF, DOCX, DOC, XLSX, XLS, PPTX, PPT, ODT, ODS, ODP, RTF, and more
• Text/Markup: TXT, MD, HTML, XML, JSON, YAML, LaTeX, Jupyter notebooks
• Images (OCR): PNG, JPEG, TIFF, GIF, BMP, WEBP, SVG
• Email: EML, MSG (with attachment extraction)
• eBooks: EPUB
• Archives: ZIP, RAR, 7Z, TAR, GZIP
• Data: CSV, DBF

Development

# Install dependencies
uv sync

# Run tests
uv run pytest

# Run linting
uv run ruff check src/ tests/
uv run ruff format --check src/ tests/

# Run type checking
uv run mypy src/

License

MIT