# XML > Parse, generate, and transform XML with correct namespace handling and encoding. - Skill: `modbender/xml` (Agent Skill) - Install (CLI): `npx skillmds add modbender/xml` - Raw SKILL.md: https://api.skillmd.com/api/skills/modbender/xml/raw - Safety review: pending - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: modbender (https://skillmd.com/u/modbender) - Updated: 2026-09-09 - Page: https://skillmd.com/skills/modbender/xml --- ## Namespaces - XPath `/root/child` fails if document has default namespace—use `//*[local-name()='child']` or register prefix - Default namespace (`xmlns="..."`) applies to elements, not attributes—attributes need explicit prefix - Namespace prefix is arbitrary—`` and `` are identical if both prefixes map to same URI - Child elements don't inherit parent's prefixed namespace—each must declare or use prefix explicitly ## Encoding - `` must match actual file encoding—mismatch corrupts non-ASCII - Encoding declaration must be first thing in file—no whitespace or BOM before it (except UTF-8 BOM allowed) - Default encoding is UTF-8 if declaration omitted—but explicit is safer across parsers ## Escaping & CDATA - Five entities always escape in text: `&` `<` `>` `"` `'` - CDATA sections `` for blocks with many special chars—but `]]>` inside CDATA breaks it - Attribute values: use `"` if delimited by `"`, or `'` if delimited by `'` - Numeric entities `<` and `<` work everywhere—useful for edge cases ## Whitespace - Whitespace between elements is preserved by default—pretty-printing adds nodes that may break processing - `xml:space="preserve"` attribute signals whitespace significance—but not all parsers respect it - Normalize-space in XPath: `normalize-space(text())` trims and collapses internal whitespace ## XPath Pitfalls - `//element` is expensive—traverses entire document; use specific paths when structure is known - Position is 1-indexed: `[1]` is first, not `[0]` - `text()` returns direct text children only—use `string()` or `.` for concatenated descendant text - Boolean in predicates: `[@attr]` tests existence, `[@attr='']` tests empty value—different results ## Structure - Self-closing `` and empty `` are semantically identical—but some legacy systems choke on self-closing - Comments cannot contain `--`—will break parser even inside string content - Processing instructions `` cannot have `?>` in data - Root element required—document with only comments/PIs and no element is invalid ## Validation - Well-formed ≠ valid—parser may accept structure but fail against schema - DTD validates but can't express complex constraints—prefer XSD or RelaxNG for new projects - XSD namespace `xmlns:xs="http://www.w3.org/2001/XMLSchema"` commonly confused with instance namespace