The separator row is what makes a Markdown table recognizable
A line such as | --- | ---: | under the header tells Markdown that this is a table. The converter uses that marker too, so unrelated prose containing a | character is not pulled into the CSV.
Outer pipes are optional
Both | A | B | and A | B can be read as rows. If a pipe is part of the cell value itself, escape it as \| so it is not treated as a column boundary.
Short rows are padded; extra columns stop the conversion
If the header has three columns and a data row has two, the third CSV cell is left blank. If a row has more columns than the header, the converter stops instead of silently discarding data.
Markdown formatting inside cells stays as text
Syntax such as **bold** or Markdown links is not rendered or stripped. This converts table structure to CSV; it is not a Markdown-to-plain-text cleaner.
A practical example
Convert a comparison table from GitHub into CSV for sorting, or move a specification table into Excel. After conversion, checking the header and a few rows is usually enough to spot a column mismatch.