Paste a CSV and get a Markdown table ready for a README, a pull request or documentation. Columns are padded to a common width so the source is readable where you will actually edit it, and pipes inside cells are escaped — that last one is the failure most converters miss, because an unescaped pipe ends the cell and silently splits your value into two columns. Alignment is set per table with left, center or right, written as the colons in the separator row. The input is parsed as real CSV, so a quoted field containing a comma stays one cell rather than becoming two. Drop a folder of files and get them all back as a ZIP; nothing is uploaded.
| Separator | Alignment |
|---|---|
| --- | Left — Markdown's default |
| :--- | Left, stated explicitly |
| :---: | Center |
| ---: | Right |
It is escaped as \|. Left alone, a pipe ends the cell, so a value like 'yes | no' becomes two columns and knocks the rest of the row out of alignment. This is the most common way a generated Markdown table breaks.
No parser needs it, but a person editing the source does. An unpadded table is unreadable in raw Markdown, and raw Markdown is where it gets edited.
Pick left, center or right, and the colons are written into the separator row. Left is Markdown's own default, which is why a plain --- separator behaves as left-aligned.
It becomes a space. Markdown tables have no multi-line cell, and a raw line break would end the row early. If you need a visible break, HTML's <br> works in most renderers.
No. Conversion happens in your browser.
Two details separate a Markdown table that works from one that looks right in the box and breaks when rendered. The first is the pipe: a pipe inside a cell ends that cell, so a value like 'yes | no' silently becomes two columns and pushes the rest of the row out of alignment. Every pipe is escaped on the way out. The second is the separator row, where colons set the alignment. Carving the colons out of a fixed-width run of dashes can leave only two, and several parsers reject a two-dash separator outright, so the colons are added to at least three dashes rather than taken from them. Cells are padded to a common column width, which no parser requires and every human editing the source does. A newline inside a cell is flattened to a space, because Markdown tables have no row-spanning cell and a raw line break would end the row. Input is parsed as full RFC 4180 rather than split on commas, so a quoted field containing a comma stays one cell. Everything runs in your browser.