MergeGlassDiff clearly. Merge confidently.

Structured data comparison

Compare JSON files by structure, not key order

Normalize JSON objects before comparing them so indentation and property order do not hide the value changes you actually need to review. Resolve the differences and export a clean combined result in your browser.

Compare JSON nowHow the tool works

Separate structural changes from formatting changes

Two JSON documents can represent the same object while using different indentation or object-key order. A purely line-oriented diff may then mark most of the document as changed. MergeGlass JSON mode parses both inputs, recursively sorts object keys, and pretty-prints a stable representation before comparison. Arrays retain their order because moving an array element can change meaning.

Normalization makes changed values, added properties, removed properties, and array edits easier to isolate. It is especially helpful for API fixtures, package metadata, exported settings, feature-flag snapshots, and machine-generated configuration. If either document is invalid JSON, the tool reports the parse problem instead of silently treating malformed data as structured input.

The merged output in semantic mode is the normalized representation. That is useful for producing consistent JSON, but comments cannot be preserved because standard JSON does not support them, and original spacing or property order will not survive. Use plain-text mode when exact formatting is part of what you need to review.

Choose the right JSON comparison approach

Use JSON mode for values

Choose semantic JSON when the meaning of objects matters more than the original formatting and object-key sequence.

Use text mode for formatting

Choose plain text if you are auditing generated output, intentional key ordering, or an exact formatting convention.

Treat arrays as ordered

If an array is logically a set, sort it using a domain-aware script before comparison. MergeGlass does not guess how elements should be matched.

Check types, not just text

The JSON values 1, "1", true, and "true" are different types. Review the full changed token before resolving it.

A repeatable four-step workflow

  1. Add both JSON documentsSelect .json files or paste each document. Automatic mode uses the file extension and content to detect JSON.
  2. Confirm semantic JSON modeThe interface identifies normalized semantic output. If parsing fails, fix the invalid document or switch deliberately to text mode.
  3. Resolve meaningful changesWork through additions, removals, and replacements while checking their containing object or array.
  4. Validate the resultDownload the merged JSON, parse it in the consuming application, and run schema validation when a schema exists.

Important limits

  • Duplicate object keys are ambiguous and parsers may keep only one value.
  • Array order is preserved and remains significant.
  • JSON normalization changes whitespace and object-key order in the exported result.