PDF to Markdown converter for docs, wikis, notes and AI tools
You want a PDF's content in your wiki, docs site, Obsidian or Notion vault, or ready for an AI tool, and plain text loses all structure. Your todo.is agent turns the PDF into clean Markdown with real headings, lists, tables, code blocks and links, and extracts images into a folder.
The prompt
- Convert the attached PDF to Markdown: [ATTACH THE PDF]. Use # for the title and ## and ### for sections, matching the PDF's headings. Keep lists, bold and italic, links and footnotes. Turn tables into Markdown tables, and if a table is too complex, [COMPLEX TABLES]. Put code in fenced blocks with the language. Remove page headers, footers and page numbers, and re-join paragraphs split by page breaks. Save images to an images/ folder and link them. Output [OUTPUT] and zip it with the images.
What to change
- [ATTACH THE PDF]: Attach the PDF. Scanned PDFs are run through OCR first.
- [COMPLEX TABLES]: E.g. "use HTML tables", "simplify them" or "save them as CSV and link them".
- [OUTPUT]: E.g. "one .md file", "one .md file per chapter" or "Obsidian notes with front matter".
Example result
- PDF to Markdown: Device_Manual_v3.pdf
- 48 pages · 6 chapters · output: one .md file per chapter + images/ folder, zipped
- Structure detected
- • Title (largest font) → #
- • Chapter titles → ## (one file each: 01-getting-started.md ... 06-troubleshooting.md)
- • Sub-sections → ###
- • Bold labels like Warning: kept in bold
- Example: 02-installation.md
- Installation
- Requirements
- • Power supply: 12 V DC, 2 A
- • Firmware 3.1 or later
- Connect to the network
- 1. Hold the Reset button for 5 seconds.
- 2. Open the app and choose Add device.
- 3. Run this to check the connection:
- ping 192.168.4.1
- | LED color | Meaning |
- |---|---|
- | Green | Connected |
- | Blinking blue | Pairing mode |
- | Red | No network |
- 
- Clean-up
- • Running header "Device Manual v3" and page numbers removed from every page
- • 31 paragraphs that were split across pages were joined
- • Hyphenated words at line ends re-joined
- • Footnotes became Markdown footnotes [^1]
- • Internal "see page 23" references became links to the right section
- Tables and images
- • 7 tables as Markdown tables; 2 with merged cells as HTML tables (as you chose)
- • 22 images extracted as PNG, named by page
- Ready for
- GitHub, GitLab, Obsidian, docs generators such as MkDocs or Docusaurus, and pasting into AI tools that read Markdown.
How to do it with todo.is
- Copy the prompt and choose how to handle complex tables and how to split the output.
- Attach the PDF in todo.is, or send it to your agent on Discord or Telegram.
- Your agent maps the heading levels, converts each element and extracts the images.
- Download the zip. Ask for front matter, a different split or a single combined file if needed.
Tips for a better result
- Ask for one file per chapter for long PDFs. Wikis and docs sites handle small pages better.
- For Obsidian or a static site generator, ask for YAML front matter (title, tags, source) at the top of each file.
- Markdown tables cannot merge cells. Choose HTML tables or a simpler layout for complex ones.
- For AI tools and search indexes, keep page markers as comments, so answers can point back to the page.
PDF to Markdown converter: FAQ
- Why convert a PDF to Markdown? Markdown keeps the structure (headings, lists, tables, links) in plain text. It works well in wikis, Git, note apps and AI tools, and is easy to edit.
- Does it keep tables? Simple tables become Markdown tables. Tables with merged cells can be kept as HTML tables or saved as CSV files.
- Can it convert a scanned PDF to Markdown? Yes. Your agent runs OCR first, then rebuilds the structure. Check headings and numbers on poor scans.
- Is my PDF kept private? Your file stays in your own agent workspace and only your agent uses it.
JavaScript is required to use the todo.is app.