html-parser-triage

name	html-parser-triage
description	Triage HTML parser issues by isolating tokenization/tree-construction failures with minimal scope.

Skill: html-parser-triage

Use when:

• DOM structure is incorrect
• nodes are missing/misplaced
• a site failure suggests tokenizer/tree-builder issues
• raw source and DOM dump do not match

Goal

Find the first structural divergence and fix the owning parser stage.

Workflow

1. Compare raw source to parsed DOM/tree output.
2. Find the first divergence, not the noisiest later symptom.
3. Classify the issue:
   • tokenization
   • insertion mode / tree construction
   • attribute handling
   • raw-text/RCDATA/script/style handling
   • parser error recovery
4. Inspect the smallest relevant parser files.
5. Patch the earliest broken rule.
6. Verify with:
   • smallest repro HTML if possible
   • focused parser/tree-builder tests if available
   • then broader slice only if needed

Do not

• patch layout if DOM is already wrong
• use site-specific string hacks unless explicitly requested
• rewrite parser subsystems for a localized bug

Output contract

1. first divergence
2. owning subsystem
3. minimal correction
4. verification

Token discipline

• quote only the tiny source fragment that matters
• keep explanation tied to the first divergence