references/owasp-xss.md

GitHub

OWASP Cross-Site Scripting Prevention

Use this reference when untrusted data reaches HTML, attributes, JavaScript, CSS, URLs, templates, or browser DOM operations.

Review Checks

  • Identify the exact output context before evaluating the encoding control.
  • Prefer framework auto-escaping and safe DOM sinks, but verify escape hatches and custom rendering.
  • Treat HTML, attribute, JavaScript, CSS, and URL contexts as different sinks with different encoding needs.
  • Review rich-text features for dedicated sanitization rather than ad hoc filtering.
  • Check that CSP is defense in depth, not the only XSS control.

Pattern Examples

Unsafe DOM sink

// Vulnerable
preview.innerHTML = comment;
// Safer for plain text
preview.textContent = comment;

Raw template rendering

<!-- Vulnerable when `bio` contains untrusted HTML -->
<div>{{{ bio }}}</div>
<!-- Safer when the template engine escapes by default -->
<div>{{ bio }}</div>

Attribute injection

// Vulnerable: attacker controls both value and execution context.
node.innerHTML = `<img src="${avatarUrl}" onerror="${handler}">`;
// Safer: hardcode the attribute name and assign a validated URL.
node.setAttribute("src", validateImageUrl(avatarUrl));

Rich HTML

// Vulnerable
article.innerHTML = markdownToHtml(userMarkdown);
// Safer when HTML is intentionally allowed
article.innerHTML = htmlSanitizer.sanitize(markdownToHtml(userMarkdown));

Review Prompts

  • What parser context receives the value: HTML body, attribute, JS string, CSS value, or URL?
  • Does the code use an unsafe sink such as innerHTML, outerHTML, or document.write?
  • Are attributes hardcoded and non-executable?
  • Is user-controlled HTML sanitized with a maintained library?
  • Are JSON responses served with the correct content type instead of being rendered as HTML?

Source

Local summary based on the OWASP Cross Site Scripting Prevention Cheat Sheet: https://cheatsheetseries.owasp.org/cheatsheets/Cross_Site_Scripting_Prevention_Cheat_Sheet.html