How do I install the Markitdown skill?

Run npx skillmds@latest add officebeats/markitdown in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Markitdown safe to use?

This skill has not completed SkillMD's automated safety review yet. Capability flags: docs only. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Markitdown?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Markitdown free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Markitdown?

officebeats (@officebeats) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Markitdown by officebeats — Agent Skill

Name: Markitdown
Rating: 4.2 (5 reviews)
Author: officebeats

Markitdown

Convert local files into token-efficient Markdown with Microsoft MarkItDown. Use when the user asks to convert, extract, normalize, read, or ingest PDF, Word, PowerPoint, Excel, Outlook, HTML, CSV, JSON, XML, EPUB, ZIP, image, or audio content as Markdown, and by default when supported non-Markdown files enter the Beats PM intake staging lane.

officebeats Updated Sep 10, 2026

File contents

MarkItDown Intake

Use the repo wrapper so conversion stays local, repeatable, and bounded:

python3 system/scripts/markdown_intake.py "<input-file>" --output "<output-file>.md"

Workflow

Preserve the source file. Accept local files only; never pass an untrusted URL or enable remote fetching.
Run python3 system/scripts/markdown_intake.py --check. If MarkItDown is unavailable, explain that Python 3.10+ is required and ask before installing system/requirements-markitdown.txt.
Convert supported staged files to a sibling .md file unless the user names another local destination. Never overwrite an existing file without the user's explicit instruction.
Verify that the Markdown file exists and is non-empty, then use that file as the primary extraction and reasoning input. Keep the original as evidence.
Report scanned PDFs, image-only pages, complex tables, or unsupported media as conversion limitations. Do not enable plugins, LLM vision, Azure services, or other billable/networked processing without explicit approval.

Intake Defaults

Automatically convert .pdf, .docx, .pptx, .xlsx, .xls, .msg, .html, .htm, .csv, .json, .xml, and .epub files during file intake.
Keep screenshots and pasted images on the existing visual-extraction path unless the user explicitly requests Markdown conversion.
Convert ZIP, image, or audio files only when explicitly requested because expansion, OCR, or transcription may be expensive or incomplete.
Treat existing .md files as ready. Convert .txt only when normalization is requested.
Reject inputs larger than the wrapper's safety limit instead of silently consuming them.

MarkItDown output is optimized for LLM analysis and structure preservation, not pixel-perfect reproduction. Prefer the original file for visual or formatting QA.

---name: markitdowndescription: Convert local files into token-efficient Markdown with Microsoft MarkItDown. Use when the user asks to convert, extract, normalize, read, or ingest PDF, Word, PowerPoint, Excel, Outlook, HTML, CSV, JSON, XML, EPUB, ZIP, image, or audio content as Markdown, and by default when supported non-Markdown files enter the Beats PM intake staging lane.---# MarkItDown IntakeUse the repo wrapper so conversion stays local, repeatable, and bounded:```bashpython3 system/scripts/markdown_intake.py "<input-file>" --output "<output-file>.md"```## Workflow1. Preserve the source file. Accept local files only; never pass an untrusted URL or enable remote fetching.2. Run `python3 system/scripts/markdown_intake.py --check`. If MarkItDown is unavailable, explain that Python 3.10+ is required and ask before installing `system/requirements-markitdown.txt`.3. Convert supported staged files to a sibling `.md` file unless the user names another local destination. Never overwrite an existing file without the user's explicit instruction.4. Verify that the Markdown file exists and is non-empty, then use that file as the primary extraction and reasoning input. Keep the original as evidence.5. Report scanned PDFs, image-only pages, complex tables, or unsupported media as conversion limitations. Do not enable plugins, LLM vision, Azure services, or other billable/networked processing without explicit approval.## Intake Defaults- Automatically convert `.pdf`, `.docx`, `.pptx`, `.xlsx`, `.xls`, `.msg`, `.html`, `.htm`, `.csv`, `.json`, `.xml`, and `.epub` files during file intake.- Keep screenshots and pasted images on the existing visual-extraction path unless the user explicitly requests Markdown conversion.- Convert ZIP, image, or audio files only when explicitly requested because expansion, OCR, or transcription may be expensive or incomplete.- Treat existing `.md` files as ready. Convert `.txt` only when normalization is requested.- Reject inputs larger than the wrapper's safety limit instead of silently consuming them.MarkItDown output is optimized for LLM analysis and structure preservation, notpixel-perfect reproduction. Prefer the original file for visual or formattingQA.

Frequently asked questions