Editing guide

How to clean up converted Markdown

Converted Markdown is a draft assembled from structural clues in another format. Work in passes: remove extraction noise, repair the outline, normalize block syntax, verify references, and only then polish the prose.

10 minute read

Preserve the raw conversion before editing

Keep the converted result as its own document and retain the source file until review is complete. This gives you a reference when text appears missing or out of order and prevents an unrelated unsaved draft from being overwritten.

Use undo after automated cleanup actions and make one category of change at a time. A broad rewrite can accidentally remove intentional spacing inside code or change list nesting that was already correct.

Remove extraction noise

PDF headers, page numbers, repeated footers, slide labels, empty table rows, and duplicated blank lines often add no meaning. Search for repeated fragments and remove them consistently, taking care not to delete a phrase that is also part of the real content.

  • Collapse runs of three or more blank lines outside code blocks.
  • Join words split only by PDF line-end hyphenation.
  • Delete navigation or legal boilerplate repeated on every page.
  • Remove empty headings and empty list items.

Rebuild a logical heading outline

Use one H1 for the document title. Major sections generally use H2, subsections H3, and so on. Do not skip levels merely to obtain a smaller visual style; presentation belongs to the renderer.

Converted PDFs may mark bold lines as headings too aggressively, while Word documents with manual formatting may not mark them at all. Read the outline as a table of contents and adjust levels based on meaning.

Before and afterMarkdown
Before:
# Deployment
#### Prerequisites
## Linux

After:
# Deployment
## Prerequisites
### Linux

Normalize lists and spacing

Use one marker style for unordered items and consistent indentation for nesting. Keep a blank line before and after a list, but avoid blank lines between every short item unless each item contains multiple paragraphs.

A converter can mistake numbered headings or table cells for list items. Confirm that ordered numbering carries meaning rather than reflecting a source page label.

Simplify tables for narrow screens and source readability

Check that every row has the same number of cells and that pipe characters inside content are escaped. Shorten verbose headers, move long explanations below the table, and split unrelated column groups.

Merged cells, nested tables, charts, and spreadsheet formulas have no reliable Markdown-table equivalent. Replace them with a list, several small tables, a text summary, or a link to maintained source data.

Verify links, images, and code

Relative links inherited from HTML or EPUB depend on their original location. PDF extraction can split URLs, and Office documents can contain display text whose target changed. Open every important link in the final destination.

Add descriptive alternative text to retained images. Confirm fenced code has the correct language and that special characters were not typographically transformed. Never execute copied commands until they have been reviewed as untrusted content.

Finish with meaning, not appearance

Compare names, dates, quantities, code, and decisions with the source. Rewrite sentences that depended on page position, slide visuals, or spreadsheet color. Then preview at desktop and narrow widths, download the .md file, and validate it in the publishing system.

A clean conversion should read naturally as a document that was written in Markdown—even when the original remains the authoritative record.

Clean the next conversion in one workspace

Open a converted draft, compare source and preview, then copy or download the reviewed Markdown.

Convert a supported file