markitdown

name	markitdown
description	文档转Markdown转换工具(markitdown)。当用户需要将各种文件格式转换为Markdown(PDF转Markdown/Word转Markdown/Excel转Markdown/PPT转Markdown/图片转Markdown/音频转Markdown/HTML转Markdown/CSV转Markdown/JSON转Markdown/XML转Markdown/ZIP转Markdown/YouTube视频转Markdown/EPub电子书转Markdown)、使用Azure Document Intelligence进行OCR识别、使用LLM生成图片描述、通过CLI命令行或Python API进行文档转换、处理可选依赖插件等相关任务时立即使用。支持pip install 'markitdown[all]'安装，Python >= 3.10。
trigger	["转Markdown","PDF转Markdown","Word转Markdown","Excel转Markdown","文档转换","markitdown","文件转MD","PPT转Markdown","图片转文字","视频转文字"]

MarkItDown Skill

当前版本: 0.1.5 | Python >= 3.10

v0.1.5 新特性 (2026-02-20)

• PDF 表格优化：支持对齐的 Markdown 表格格式提取
• 宽表格支持：改进 PDF 转换中对宽表格的处理
• 列表解析修复：正确处理部分编号列表
• HTTP 头部：Accept headers 添加 text/markdown 内容类型

详细更新日志见 references/CHANGELOG.md

Quick Start

Installation

pip install 'markitdown[all]'  # All features (recommended)
# Or install only what you need:
pip install 'markitdown[pdf,docx,pptx,xlsx]'  # Common formats

CLI Usage

# Convert file to stdout
markitdown path-to-file.pdf > document.md

# Specify output file
markitdown path-to-file.pdf -o document.md

# Pipe content
cat path-to-file.pdf | markitdown

# List installed plugins
markitdown --list-plugins

# Enable plugins
markitdown --use-plugins path-to-file.pdf

# Use Azure Document Intelligence
markitdown path-to-file.pdf -o document.md -d -e "<document_intelligence_endpoint>"

# Version check
markitdown --version

Python API

from markitdown import MarkItDown

# Basic usage
md = MarkItDown()
result = md.convert("test.xlsx")
print(result.text_content)

# Enable plugins
md = MarkItDown(enable_plugins=True)
result = md.convert("document.pdf")

# With Azure Document Intelligence
md = MarkItDown(docintel_endpoint="<endpoint>")
result = md.convert("document.pdf")

# With LLM for image descriptions (OpenAI)
from openai import OpenAI
client = OpenAI()
md = MarkItDown(llm_client=client, llm_model="gpt-4o")
result = md.convert("example.jpg")
print(result.text_content)

# Streaming API (binary mode required)
import io
with open("document.pdf", "rb") as f:
    stream = io.BytesIO(f.read())
    result = md.convert_stream(stream)
    print(result.text_content)

Supported Formats

Format	Install Command	Notes
PDF	[pdf]	Text extraction, table support
PowerPoint	[pptx]	Slides, notes, images
Word	[docx]	Document text, tables
Excel	[xlsx]	Spreadsheet data
Old Excel	[xls]	Legacy format
Images	Built-in	EXIF + OCR
Audio	[audio-transcription]	EXIF + speech transcription
YouTube	[youtube-transcription]	Video transcription
HTML	Built-in	Web page conversion
CSV/JSON/XML	Built-in	Text-based formats
ZIP	Built-in	Iterates contents
EPub	Built-in	Ebook conversion
Outlook	[outlook]	Email messages
Azure Doc Intel	[az-doc-intel]	Cloud OCR

Optional Dependencies

pip install 'markitdown[all]'           # Everything
pip install 'markitdown[pptx]'           # PowerPoint only
pip install 'markitdown[docx]'           # Word only
pip install 'markitdown[xlsx]'           # Excel only
pip install 'markitdown[pdf]'            # PDF only
pip install 'markitdown[xls]'            # Old Excel
pip install 'markitdown[outlook]'        # Outlook emails
pip install 'markitdown[audio-transcription]'  # Audio transcription
pip install 'markitdown[youtube-transcription]' # YouTube transcription
pip install 'markitdown[az-doc-intel]'  # Azure Document Intelligence

Plugins

Using Plugins

# List installed plugins
markitdown --list-plugins

# Enable plugins for conversion
markitdown --use-plugins path-to-file.pdf

Developing Plugins

See packages/markitdown-sample-plugin in the MarkItDown repository for plugin development guide.

Azure Document Intelligence

For complex PDF layouts, use Azure Document Intelligence:

markitdown document.pdf -o output.md -d -e "<endpoint>"

Python API:

md = MarkItDown(docintel_endpoint="your-endpoint")
result = md.convert("document.pdf")

Setup guide: https://learn.microsoft.com/en-us/azure/ai-services/document-intelligence/how-to-guides/create-document-intelligence-resource

LLM Image Descriptions

For richer image descriptions, provide an LLM client:

from markitdown import MarkItDown
from openai import OpenAI

client = OpenAI()
md = MarkItDown(llm_client=client, llm_model="gpt-4o")
result = md.convert("image.jpg")

Streaming API

import io
from markitdown import MarkItDown

md = MarkItDown()

# Convert from bytes stream
with open("document.pdf", "rb") as f:
    stream = io.BytesIO(f.read())
    result = md.convert_stream(stream)
    print(result.text_content)

Note: convert_stream() requires a binary file-like object (not text mode like io.StringIO).

Troubleshooting

Common Issues

"No converter found": Install the optional dependency for that format (e.g., pip install 'markitdown[pdf]')

OCR fails on images: Ensure Tesseract is installed or use LLM integration for image descriptions

Azure Document Intelligence errors: Verify endpoint URL and authentication credentials

Plugin not loading: Use --use-plugins flag or set enable_plugins=True in Python

Streaming error: Use io.BytesIO() (binary mode), NOT io.StringIO() (text mode)

Installation Check

markitdown --help  # Verify CLI is installed
pip show markitdown  # Check installed version
markitdown --version  # Show version

Docker Usage

# Build image
docker build -t markitdown:latest .

# Convert file using Docker
docker run --rm -i markitdown:latest < input.pdf > output.md
docker run --rm -i markitdown:latest input.pdf -o output.md

Use Cases

LLM Data Preparation

# Convert documents for LLM ingestion
from markitdown import MarkItDown

md = MarkItDown()
files = ["report.pdf", "notes.docx", "data.xlsx"]

for f in files:
    result = md.convert(f)
    print(f"=== {f} ===")
    print(result.text_content[:500])  # Preview

Batch Processing

import os
from markitdown import MarkItDown

md = MarkItDown()
output_dir = "markdown_output"
os.makedirs(output_dir, exist_ok=True)

for filename in os.listdir("."):
    if filename.endswith((".pdf", ".docx", ".pptx")):
        result = md.convert(filename)
        output_path = os.path.join(output_dir, filename.rsplit(".", 1)[0] + ".md")
        with open(output_path, "w", encoding="utf-8") as f:
            f.write(result.text_content)