Read XML as a tree
Open an element using its disclosure arrow, then choose Inspect to see its exact source. Attributes appear first, followed by child content in document order. Text, CDATA, comments and processing instructions remain distinct. Whitespace-only text is hidden initially to make the tree easier to scan; turn it on when spacing matters. Each container and search page loads 80 entries at a time.
Understand values and namespaces
The exact fragment preserves quotes, entity spelling and original line breaks. Decoded values resolve XML predefined entities and numeric references, with XML line-ending and attribute-whitespace normalization. The namespace URI shows the expanded namespace of an element or attribute. Default namespaces apply to elements but not unprefixed attributes. Namespace declarations have their own entry type.
Find the original source
Search names, decoded values, namespaces or navigation paths. Show in source selects the original span, with line and column feedback. Paths use sibling positions to distinguish repeated entries; they are descriptive navigation labels, not XPath queries. A selected fragment may depend on inherited namespaces or be an attribute or text node. It downloads as a text fragment; download the original XML for a complete document.
Recover work and download faithfully
Original downloads preserve imported UTF-8 bytes, BOM and line endings until editing. Example, Clear and successful import save one Undo step including the source, selected entry, expanded tree, search page and whitespace option. Ordinary source edits invalidate the view. Invalid UTF-8 and oversized imports preserve the accepted view. Syntax errors stay editable with source selection.
Local processing and limits
Files are read locally, with no automatic storage or uploads. XML is shown as text; embedded markup, scripts and stylesheet instructions are never executed. Only XML 1.0 with UTF-8 declarations is supported. DTDs and custom/external entities are rejected. This viewer checks well-formedness and namespace rules; it does not validate schemas or signatures. Limits are 256 KiB input, 20,000 entries including attributes and whitespace, 100 element levels and 4 MiB of combined navigation paths. Worker processing and file reading have a ten-second cancellation deadline.