Markdown cleanup checklist after conversion

Start with headings. A converted file should normally have one main `#` title, then clear `##` sections, then lower levels only when they describe real sub-sections. Do not use heading levels to make text look bigger. Use them to describe the document structure.

Promote real section titles that stayed as paragraphs. Demote labels that became headings only because they were bold in the source. This pass improves navigation, table-of-contents generation, search snippets, and RAG chunking.

Check numbered lists for missing steps, early restarts, and lost indentation. Procedures, policies, and requirements can change meaning when nesting breaks. Join list items that were split by line wrapping or page breaks.

Paragraphs from PDFs often break at visual line endings. Join lines that belong together, but keep real paragraph breaks. Around page boundaries, read the last sentence before the break and the first sentence after it.

Link text should explain the destination. `Click here` and `read more` are weak when the content leaves its original layout. Replace them with descriptive anchors where possible.

For images, decide whether the image is essential, decorative, or missing. Add alt text or captions for important images. For tables, check rows, columns, missing cells, and whether the table is too wide for Markdown.

Remove repeated headers, footers, page numbers, cookie banner text, navigation menus, slide numbers, watermarks, and template boilerplate when they do not help the final reader. These artifacts are common in converted PDFs, HTML pages, and slide decks.

Also search for private placeholders, comments, internal URLs, tracked-change remnants, and hidden template instructions. Conversion can expose text that was easy to ignore in the original application.

Preview the Markdown where it will actually live: repository, static site, CMS, docs platform, or knowledge-base system. Valid Markdown can still be hard to read if tables are too wide, headings are misleading, or links lack context.

For important files, record the source filename, conversion date, reviewer, and manual edits that changed meaning. This creates a lightweight audit trail without turning cleanup into bureaucracy.