Quickstart

pip install markdrop
markdrop convert report.pdf --output_dir out

Fast mode on CPU (no Docling/Torch):

markdrop convert report.pdf --output_dir out --fast

Generated files:

• out/report-markdroped.md
• out/report-markdroped.html
• out/images/ and out/tables/
• out/manifest.json

Optional table downloads in HTML:

markdrop convert report.pdf --output_dir out --add_tables

AI enrichment (requires API key setup):

markdrop setup gemini
markdrop describe out/report-markdroped.md --ai_provider gemini --output_dir out/enriched

Python API:

import asyncio
from markdrop import convert_document, MarkDropConfig, process_markdown, ProcessorConfig, AIProvider

result = convert_document("report.pdf", "out", MarkDropConfig())
print(result.markdown_path, result.html_path)


async def enrich():
    config = ProcessorConfig(
        input_path=str(result.markdown_path),
        output_dir="out/enriched",
        ai_provider=AIProvider.GEMINI,
    )
    return await process_markdown(config)


asyncio.run(enrich())