Markdown tables: generate them instead of counting pipes

The syntax, in one example

                | Tool          | Input       | Offline |
| :------------ | :---------: | ------: |
| YAML to JSON  | YAML        | yes     |
| JWT decoder   | JWT         | yes     |
              

The first row is the header, the second is the delimiter row, and everything after is data. Colons in the delimiter row set alignment: `:---` left, `:---:` centre, `---:` right. GitHub and most renderers accept a leading and trailing pipe as optional, but including them keeps the raw text readable in an editor.

The three things that break tables

• A pipe inside a cell — escape it as `\|`, or the row is split into one column too many.
• A delimiter row with the wrong number of columns — the header shifts and the table renders misaligned.
• A blank line inside the table — the renderer treats it as the end of the table and prints the rest as text.

Escaping matters most for code and regexes in cells: `a|b` and `\bfoo\b` both need attention before they can live happily inside a table.

Getting data in from a spreadsheet

The natural workflow is to paste a selection from a spreadsheet or a CSV export and let the generator do the quoting and alignment. That is faster and less error-prone than typing rows, and it also handles the boring cases — variable column widths, numbers that should be right-aligned, cells that contain commas. Trim trailing empty columns before pasting to avoid a table with phantom columns that render as gaps.

When not to use a table

Tables are for comparisons: feature matrices, flag references, compatibility lists, environment variables. They are poor for prose, bad for long descriptions, and painful on narrow screens because markdown has no responsive behaviour — the renderer either scrolls horizontally or squeezes columns into unreadable slivers. If a cell needs a paragraph, use a definition list, a heading per entry, or a bulleted list instead. A table nobody can read on a phone is worse than the list it replaced.

Generate, then paste anywhere

The output is plain markdown, so it drops straight into a README, a pull request description, a wiki page, a Jira ticket or a comment. The YamlBox markdown table generator runs in the browser, aligns columns for you, escapes pipes, and keeps working offline — so you can paste internal data without publishing it anywhere.