How to get your docs into markdown

Convert one small group of documents, check the result by eye, and add a short label to the top of each file. Claude Code, Copilot, Cursor, Codex, Gemini CLI, Windsurf and local models can all read the plain text that comes out.

1. Start with what agents can't read

Start with SOPs, policies, product docs and support docs. An agent can open your code by itself. It usually can't get at these. Leave out anything that is out of date, because an agent will repeat a stale document as fact.

2. Convert them

Google Docs: choose File, then Download, then Markdown.
Notion: use the page menu's Export command and choose Markdown.
Word and HTML files: Pandoc runs on your machine.
```
pandoc input.docx -t gfm -o output.md
```
PDFs and Office files: MarkItDown also runs on your machine.
```
markitdown input.pdf -o output.md
```
Scanned PDFs: these are pictures of text, so run OCR on them first.

We don't make these tools and have no tie to them. JidoSeal doesn't convert files. It reads the markdown once you have it.

3. Check the result by eye

Conversion gets most of the text right and trips on layout. Look at tables, headings, numbered steps and anything that sat in a column or a text box. Fix those by hand in your first few files, so you know what to expect from the rest.

4. Label each file

A converted file still doesn't say what it is. Add a short block of frontmatter at the very top. The Open Knowledge Format asks for a type, and the other fields tell an agent who owns the file and when it was last reviewed.

---
type: Policy
title: Refund policy
owner: Support team
---

5. Pick a folder, scan it

To find every file still missing a label, run the free scan. Your files stay on your machine. Scan a folder, free.