Top 200 ReposEXPLORER
Browse agents, categories and pages
Back to results
Reported by sourceThe source says this happened. Top 200 Repos has not reproduced it.

Convert PDFs and Office files to Markdown for an LLM

Run MarkItDown on PDFs, Word, PowerPoint and Excel files to get structure-preserving Markdown an LLM can read cheaply.

Try the promptRead the original (opens in a new tab)

The problem

How it works

MarkItDown is a Python utility with a CLI and an API that turns PDF, PowerPoint, Word, Excel, HTML, CSV, JSON, EPUB and other inputs into Markdown, keeping headings, lists, tables and links. You install it with optional extras per format, then call `markitdown file -o out.md` or the `convert()` method. Optional settings add LLM image descriptions or Azure Document Intelligence for layout extraction.

Results and limitations

The README positions the output as suited to text analysis and LLM pipelines, and says it keeps document structure intact.

Before you try it

The maintainers note it is not aimed at high-fidelity conversions for human readers, that Azure features can be billable, and that it runs with the current process's privileges, so untrusted files need care. Results depend on your setup and were not independently reproduced by Top 200 Repos.

Tools and repositories

Tools
Python 3.10+ · markitdown CLI
Prerequisites
No API key for local conversion; Azure Document Intelligence or an LLM client are optional and need their own credentials.

Try it yourself: Batch-convert a folder to Markdown

Fill the fields to adapt the prompt. Empty fields stay as visible [placeholders].

Preview
Convert every file in [Folder with source files] with the extensions [File types] to Markdown using MarkItDown. Before converting, list the files you found with their sizes and tell me which optional extras need installing; ask before installing anything. Write each result to [Markdown output folder], keeping the original file name with a .md extension, and never overwrite an existing file without asking. Do not enable Azure or LLM-based features unless I confirm, since they send content to a cloud service. When done, report files converted, files that failed with the error, and any tables or images that looked lost so I can spot-check them.

Only convert files you trust, and keep cloud conversion off for confidential documents unless approved.

Related videos

About microsoft/markitdown

About microsoft/markitdown

Independent creators. Listing a video implies no endorsement, sponsorship or partnership.

Sources

  1. microsoft/markitdown README (opens in a new tab)

    Project documentation · github.com · Undated · Source checked 9 Nov 2026, 22:21 UTC

    What this source supports
    • Converts PDF, Office, HTML, CSV, JSON, EPUB and more to Markdown
    • Preserves headings, lists, tables and links
    • CLI and Python API usage; Python 3.10 to 3.14
    • Not intended for high-fidelity human-facing conversion
    • Azure Document Intelligence and Content Understanding are optional and may be billable
Report a problem

Reported by source

Parse scanned PDFs with tables into Markdown and JSON

Run PaddleOCR's PP-StructureV3 pipeline on scans to recover layout, tables, formulas and reading order as Markdown or JSON.

github.comPaddlePaddle/PaddleOCR
Planned, not done

Auto-tag receipts and bills in Paperless-ngx with paperless-ai

paperless-ai suggests titles, tags, correspondents and dates for documents in a Paperless-ngx archive, using a local Ollama model or a cloud LLM, with manual review and undo.

github.comollama/ollama
Reported by source

Extract invoice fields locally with a vision model on Ollama

Stephen Sebastian replaced a paid cloud OCR API with Gemma 4 E4B on Ollama to pull date, amount and description from contractor invoice photos on his laptop.

dev.toollama/ollama
Reported by source

Get a new OCR model running on unfamiliar GPU hardware with Claude Code in a sandbox

Simon Willison gave Claude Code full permissions inside a Docker container on an NVIDIA Spark and had it get DeepSeek-OCR working; it took four prompts and under 40 minutes, with one nudge from him.

simonwillison.net · 20 Oct 2025anthropics/claude-code, pytorch/pytorch, huggingface/transformers
Reported by source

Build a formula-based analysis workbook with Agent Mode in Excel

Agent Mode in Excel plans formulas, sheets and charts from a prompt, checks and fixes its own results, and reports its validation steps; Microsoft reports 57.2% accuracy on SpreadsheetBench in its own evaluation.

microsoft.com · 29 Sept 2025
Planned, not done

Build a formula-driven Excel model with the Claude xlsx skill

Install Anthropic's xlsx skill in Claude Code so the agent writes live formulas, recalculates the workbook with LibreOffice and checks it for formula errors.

github.comanthropics/skills, anthropics/claude-code