Markdown is the format I write almost everything in - READMEs, plugin documentation, release notes, pull request descriptions, the changelog for WP Adminify, half my internal notes. It gets out of the way for every construct except one. The moment a table is involved, I stop writing and start counting pipe characters, and the flow is gone. Add a column to a five-row table and you are editing six lines by hand, plus the delimiter row, and you will not know whether you got it right until the page renders.
TL;DR: The Markdown Table Generator gives you a spreadsheet-style grid, generates valid GitHub Flavored Markdown as you type, and imports CSV, spreadsheet paste, or an existing Markdown table so you can restructure it. Set per-column alignment with one click, and export Markdown, HTML, or CSV. Everything runs in your browser.
The tool exists because table syntax is the one part of Markdown where the source format actively fights editing. This guide covers how GFM tables work, the rules that decide whether yours renders, the escaping traps that break rows, and how to use the generator for the workflows tables usually come from - a spreadsheet, a query result, or a table someone already wrote.
What is a Markdown table?
A Markdown table is three ingredients stacked in a fixed order: a header row, a delimiter row, and zero or more body rows. Tables are not part of core Markdown at all - they are an extension defined by the GitHub Flavored Markdown spec, which is why a table can render on GitHub and come out as literal pipes elsewhere. Cells are separated by pipe characters, and each row occupies exactly one line.
| Tool | Category | Offline |
| --- | --- | --- |
| JSON Formatter | Coding | Yes |
| YAML to JSON | Data | Yes |
The delimiter row is the part that makes it a table rather than a paragraph containing pipes. It must be the second line, it must consist of hyphens with optional colons, and it must have the same number of cells as the header. Get any of those wrong and the renderer treats the whole block as ordinary text - which is why "my table renders as plain text" is almost always a delimiter row problem.
Tables are not part of the original CommonMark specification. They arrived as a GitHub Flavored Markdown extension and were subsequently adopted almost everywhere: GitHub, GitLab, Bitbucket, Obsidian, Notion, Docusaurus, MkDocs, VitePress, Hugo, Jekyll, Discord, and most static-site generators. "Almost" is the operative word. A minimal CommonMark renderer with no extensions will show your table as literal pipes, and the fix is to emit HTML instead - which the generator can do with one click on the output tabs.
Why the syntax is harder to maintain than it looks
Reading a Markdown table is easy. Maintaining one is where the cost hides, and it comes from four places.
Structure lives in punctuation. There is no cell boundary except a character that also appears in regular expressions, shell commands, and TypeScript union types. Every one of those inside a cell must be escaped or the row silently gains a column.
Alignment is encoded in colon placement. A colon on the left means left-aligned, a colon on both sides means centered, a colon on the right means right-aligned. There is no keyword to remember it by, and the difference between ---: and :--- is one character in a row nobody reads closely.
Editing is multi-line. Adding a column means touching every row plus the delimiter. Reordering rows means cut and paste with no visual feedback. Deleting a column is a search for the nth pipe on every line.
Alignment of the source is manual. Padded columns are not required - renderers ignore extra spaces - but unpadded tables are miserable to read in a diff, so people pad them by hand and then re-pad them every time a cell's content grows.
A grid editor removes all four. You edit cells; the syntax is generated. That is the entire premise of the tool.
How to use the Markdown Table Generator
Step 1: Get your data into the grid
Three routes in, depending on where the data already lives.
Type it. The grid opens with a filled sample. Overwrite the cells, or hit Clear for an empty three-by-three and start fresh.
Paste from a spreadsheet. Open the Import panel and paste a copied range from Excel, Google Sheets, Numbers, or LibreOffice. Spreadsheets put tab-separated text on the clipboard, and the importer detects that automatically - no need to tell it which delimiter you are using.
Paste CSV or an existing table. The same panel accepts CSV, semicolon-separated exports (the default in many European locales), pipe-delimited text, and Markdown tables. If the text looks like a Markdown table - a header line followed by a delimiter row - it is parsed as one, including its column alignments, so you can load a table from a README, restructure it, and paste it back.
CSV import follows the RFC 4180 rules rather than splitting on commas naively. A quoted field can contain commas, line breaks, and doubled quotes that stand for a literal quote character. That matters more than it sounds: any export containing an address, a description, or a price in a locale that uses commas will break a naive split, and you will not notice until a column is one cell short. If your CSV is large enough that you want to inspect it before importing, the CSV Viewer opens it in a sortable grid first.
Step 2: Shape the table
The controls around the grid cover the operations that are painful in raw Markdown:
- Row and Column add to the end of the grid.
- The arrows beside each row move it up or down - useful for ordering a comparison table by whatever you decided matters most.
- The minus button on a row or column header deletes it, across every row at once.
- Transpose swaps rows and columns. Data pasted from a report is frequently oriented the wrong way for a document, and transposing is one click instead of a full retype.
Step 3: Set alignment per column
Each column header carries three alignment buttons: left, center, right. Click one to apply it, click it again to return to the renderer's default. The generator writes the corresponding colons into the delimiter row.
Alignment is worth using deliberately. Numeric columns - prices, counts, percentages, durations - read far better right-aligned, because the digits line up by place value. Short status values look tidy centered. Text belongs left. This is not decoration; a right-aligned column of numbers is genuinely faster to scan for the largest value.
Step 4: Choose the output format and copy
Three output tabs:
Markdown produces GFM. The Padded columns toggle controls whether cells are padded with spaces so the raw source lines up in a monospace editor. Padding changes nothing about how the table renders - it is purely for whoever reads the file in a diff. Outer pipes controls the leading and trailing pipe on each row; both styles are valid, and the leading-pipe style is by far the more common convention.
HTML produces a <table> with <thead> and <tbody> and inline text-align styles matching your column alignments. Use it for renderers that lack table support, for email templates, or when you need colspan and multi-line cells that Markdown cannot express. If you want to go the other direction and lift a table out of a web page, HTML to Markdown handles it.
CSV gives you the grid back as data, properly quoted, for pasting into a spreadsheet or feeding to a script.
Copy to the clipboard or download as a .md, .html, or .csv file.
The escaping rules that actually break tables
Two characters cause nearly every broken Markdown table, and both are handled automatically by the generator - but you should know the rules for the tables you edit by hand.
The pipe. A literal pipe inside a cell must be written as \|. Without the backslash, the renderer sees a column separator and your row gains a cell, throwing off every column after it. This bites hardest in technical documentation, where cells legitimately contain things like string | null, grep foo | wc -l, or a regex alternation. The escape works even inside backtick code spans - a rare case where Markdown's usual "code spans are literal" rule does not apply.
The newline. GFM tables are strictly one row per line, so a cell cannot contain a real line break. The universal workaround is an HTML <br> tag. When you paste multi-line content into the grid, the generator inserts the tag for you; when you import a table containing <br>, it converts back to a real line break in the editor so the cell is comfortable to edit.
Two more rules are worth committing to memory because they cause "my table does not render" reports that have nothing to do with the table itself. First, a table needs a blank line before it when it follows a paragraph. Second, inside a list item, the whole table must be indented to the list item's content column - otherwise the list swallows the first line and the delimiter row lands in the wrong place.
Where tables come from, and how each case flows
A spreadsheet you were sent
Copy the range, paste into the Import panel, load it into the grid, delete the columns nobody needs, right-align the numeric one, copy the Markdown. Under a minute for a table that would take ten to hand-format.
A query result or an API response
Data that starts as JSON usually needs one hop. Convert it with JSON to CSV, then paste the CSV into the importer. Going the other way - a Markdown table you need as data - export CSV from the output tabs and run it through CSV to JSON.
A table that already exists in a README
Paste it into the Import panel. It comes back into the grid with its alignments intact, so adding a column or reordering rows is a normal edit rather than a rewrite. This is the workflow I use most: documentation tables grow one row at a time for months, then need restructuring all at once.
A comparison table you are writing from scratch
Start with the grid, type as you think, and let the padding sort itself out. Writing a comparison in a grid instead of in pipe syntax keeps your attention on whether the comparison is fair rather than on whether the columns line up.
Documentation that has to work in several renderers
Generate the Markdown for the platforms that support GFM tables, then switch the output tab to HTML for the one that does not. Same grid, two exports, no retyping. If your pipeline converts Markdown to HTML anyway, Markdown to HTML shows you exactly what the conversion produces.
Markdown tables versus HTML tables
| Consideration | Markdown table | HTML table |
|---|---|---|
| Readability of source | High - scannable as plain text | Low - tags dominate the content |
| Multi-line cells | Only via <br> |
Native, any block content |
| Merged cells | Not supported | colspan and rowspan |
| Per-column alignment | Yes, via delimiter colons | Yes, via CSS or attributes |
| Styling control | None | Full CSS |
| Renderer support | GFM extension, widespread but not universal | Universal wherever HTML is allowed |
| Diff friendliness | Good, one line per row | Poor, many lines per row |
The practical rule: use Markdown until you need a merged cell, a block element inside a cell, or specific styling. Then switch that one table to HTML and leave the rest of the document alone. Most Markdown renderers pass raw HTML through, so mixing the two in one file is normal.
A note on privacy
The tables people build are often the ones they would rather not upload: pricing not yet announced, headcount, internal benchmarks, a client's data pasted from a spreadsheet.
The generator is client-side JavaScript. The grid, the import parser, and every export run in the page, and nothing you paste leaves your machine. Load it once and it keeps working offline. That is the same principle behind everything on toolz.dev, and the reasoning is worth reading in full in the guide on data privacy in online tools. If you are putting together a broader set of browser-based utilities for daily work, the coding tools guide and the web developer toolkit cover what else earns a bookmark.
Practical tips
Keep the header short. With padding on, the widest cell in a column sets the width of every row in the source. A verbose header makes the raw file noisy for no rendering benefit.
Do not fear unpadded output for generated files. If a table is written by a script and never read as source, turn padding off and save the bytes.
Sort before you paste. Reordering rows one arrow-click at a time is fine for five rows and tedious for fifty. Sort in the spreadsheet first, then import.
Watch for the empty header row. With "First row is header" turned off, the generator emits an empty header above your data, because GFM has no way to express a headerless table. Some renderers show that as a thin blank row. If it bothers you, export HTML instead, which needs no header at all.
Check the rendered result, not the source. Padded source that looks perfect can still render wrong if the delimiter row has the wrong number of cells. Paste into the preview of whatever platform you are publishing on before you call it done.
FAQ
How do I create a table in Markdown?
A Markdown table needs a header row, a delimiter row of hyphens, and one row per record, with cells separated by pipe characters. Fill in the grid in the generator and the tool writes that syntax for you, including the delimiter row and any alignment colons.
How do I align columns in a Markdown table?
Alignment is set with colons in the delimiter row: a colon on the left aligns left, colons on both sides center, and a colon on the right aligns right. Choose an alignment from a column header in the editor and the correct colon placement is generated automatically.
Can I convert an Excel or Google Sheets range into a Markdown table?
Yes. Copy the cells in your spreadsheet and paste them into the Import panel. Spreadsheets place tab-separated text on the clipboard, which the importer detects automatically and loads into the grid.
How do I put a pipe character inside a table cell?
A literal pipe must be escaped with a backslash or the renderer will read it as a column separator. Type the pipe normally in the editor and the generator escapes it in the output for you.
Can a Markdown table cell contain multiple lines?
Not with real newlines, because GFM tables are strictly one row per line. The usual workaround is an HTML line break tag inside the cell, which this tool inserts automatically when a cell you paste contains a line break.
Why does my table render as plain text on GitHub?
The delimiter row is almost always the cause: it must be the second line and it must have the same number of cells as the header. A missing blank line before the table can also stop it rendering when the table follows a paragraph or sits inside a list.
Do I have to pad the columns so the source lines up?
No. Renderers ignore extra spaces, so an unpadded table displays identically. Padding exists for humans reading the raw file in an editor or a diff. Turn it off with the padded output toggle if you prefer compact source.
Are Markdown tables supported everywhere?
Tables are a GitHub Flavored Markdown extension rather than part of the original CommonMark specification. GitHub, GitLab, Obsidian, Notion, Docusaurus, and most static-site generators support them, but some minimal renderers do not. Switch the output to HTML if yours is one of them.
