JSON to Markdown

JSON to Markdown, in whatever shape fits

JSON is great for machines and awkward to read in a pull request, a wiki page, or an LLM prompt. This JSON to Markdown converter turns it into something people can scan, and picks the layout from the shape of the data:

• Array of objects → table. One row per object, columns from the union of all keys (first-seen order), numbers right-aligned, pipes escaped.
• Flat object → key/value table. A two-column table of settings, a config file, or a single API record.
• Nested object → document. Simple fields become a bullet list, nested objects become headings, and arrays of objects inside become tables — ideal for turning an API response into readable docs.
• Anything → nested bullet list or code block. Force the layout when you want a compact outline, or a fenced ```json block for documentation.

When the table you want is buried inside a response — say { "data": [ … ], "meta": { … } } — the converter finds every array of objects and lets you convert just that one.

A JSON to Markdown table example

[
  { "id": 1, "name": "ada", "owner": { "team": "web" }, "tags": ["bug", "auth"] },
  { "id": 2, "name": "grace", "owner": { "team": "api" }, "tags": [] }
]

|  id | name  | owner.team | tags      |
| --: | ----- | ---------- | --------- |
|   1 | ada   | web        | bug, auth |
|   2 | grace | api        |           |

The --: under id right-aligns the numeric column; owner.team is a flattened nested field. The CSV to Markdown converter explains the full Markdown table syntax and has a formatter for hand-written tables.

Convert a Markdown table to JSON format

Paste a Markdown table — from a README, a spec, or a chatbot answer — and get a JSON array of objects, one per row, keyed by the header. It's built to round-trip with the JSON → Markdown direction:

• Numbers, true/false, and null become real JSON types; 007 and very large integers stay strings so IDs aren't mangled.
• Dotted headers like owner.name are nested back into objects, and cells holding inline JSON (`["a","b"]`) are parsed back into arrays and objects.
• Empty cells can become "", null, or be left out; keys can stay as written or become camelCase or snake_case; duplicate headers get a suffix instead of overwriting each other.
• Escaped pipes (\|) are unescaped and <br> becomes \n. Output as pretty JSON, compact JSON, or JSON Lines.

Need the same table in a spreadsheet instead? The Markdown to Excel converter downloads it as an .xlsx or copies it straight into Google Sheets.

JSON to Markdown in Python and on the command line

import json
import pandas as pd            # pip install pandas tabulate

data = json.load(open("data.json"))
print(pd.json_normalize(data).to_markdown(index=False))

jq -r '(.[0] | keys_unsorted) as $k
  | ($k | "| " + join(" | ") + " |"),
    ($k | map("---") | "| " + join(" | ") + " |"),
    (.[] | [.[$k[]] | tostring] | "| " + join(" | ") + " |")' data.json

Both are fine for scripts, with caveats: the jq version takes its columns from the first object only, prints nested objects as raw JSON, and doesn't escape pipes, and pandas needs a Python environment. For Markdown to JSON in Python, read the table with pd.read_csv(io.StringIO(md), sep="|"), drop the empty edge columns and the --- row, then call df.to_json(orient="records").

Frequently asked questions