Free JSON to Markdown Converter
Convert JSON objects/arrays into nested Markdown headings + lists or key-value tables. Useful for feeding API responses, config files into ChatGPT/Claude.
Benefits
Nested objects → nested headings + bullet lists. AI reads structure more easily than raw JSON brackets.
Array of objects (uniform keys) → GFM table. Easy to spot patterns.
Type info kept: strings quoted, numbers unquoted, booleans italicised. AI understands data types.
How to use
- 1Paste JSON into the input box (or drop a .json file).
- 2Syntax is auto-checked — invalid JSON shows line/column.
- 3Choose output style: heading-list (nested objects) or table (arrays).
- 4Copy MD or 'Open in Claude' for AI Q&A.
What is JSON to Markdown?
JSON (JavaScript Object Notation) is the most common data format for APIs and configs. Pasting JSON into AI works but the brackets and commas force AI to re-parse — costing tokens and sometimes misreading structure. Converting to Markdown with headings + lists helps AI 'see' the structure visually.
We use native JSON.parse plus our custom transformer. Object → heading; key → bullet; nested object → nested heading; array of primitives → bullet list; array of objects (uniform schema) → GFM table.
Use cases: API engineers converting JSON responses to MD before pasting into Claude for documentation; devs converting configs (package.json, tsconfig.json, .eslintrc) for AI to explain settings; data engineers converting NoSQL documents for AI schema analysis.
- ✓Object → heading (level matches nesting depth)
- ✓Array of primitives → bullet list
- ✓Array of objects → GFM table with columns from keys
- ✓Nested object → nested heading + indented list
- ✓Boolean, null, number keep their type (italicised or plain)
- ✓Long strings (>100 chars) wrap automatically
- ✓Auto pretty-print with 2-space indent
When to use it
Take JSON from your API → MD → paste Claude to write docs or generate TypeScript types.
package.json/tsconfig.json/eslintrc → MD → ChatGPT explains each setting.
JSON Schema or OpenAPI spec → MD → Claude reviews, suggests improvements.
MongoDB document export → MD → AI analyses structure, suggests indexes.
Convert i18n JSON to MD so translators (or Claude) work more easily.
How it works
Parsing uses native JSON.parse — fast, supports JSON5 if user enables option (allows comments, trailing commas, single quotes). Syntax errors are caught and shown with line/column for quick fixes.
Tree transform uses recursive walk: for each node, check type. Object → heading + walk children. Array → check: primitives → bullet list; uniform-schema objects (90%+ matching keys) → table; mixed → list of nested.
Output has 2 modes: 'Verbose' keeps every heading even at level 1 (good for documentation), 'Compact' merges levels with <2 entries inline (denser, better for AI feed). Default 'Compact' to save tokens.
JSON → Markdown FAQ
Does JSON with comments (like tsconfig) parse?
Default JSON.parse is strict — no comments. Toggle 'JSON5 mode' to allow comments, trailing commas, unquoted keys.
What about huge JSON (10MB+)?
Browser parses up to ~50MB. Larger may slow. Phase 2 will add streaming.
How are circular references handled?
JSON spec disallows circular. If your data has them (like JS objects), use JSON.stringify with a replacer first.
What about arrays of mixed types?
Rendered as a bullet list, each item formatted by type (string/number/object separately).
JSON Schema validation supported?
Currently syntax-check only. Phase 2 will add schema validation if user provides a schema.