fix(seo): escape JSON-LD script content to prevent breakout (#208)

* fix(security): harden JSON-LD script serialization

Co-authored-by: Sebastien Tardif <sebtardif@ncf.ca>

* chore: drop PR changelog entry

---------

Co-authored-by: Peter Steinberger <steipete@gmail.com>

Author: SebTardif

src/layouts/Layout.astro

@@ -1,5 +1,6 @@
 ---
 import Analytics from '@vercel/analytics/astro';
+import { serializeJsonLd } from '../lib/serialize-json-ld';
 import '@openclaw/design-system/tokens.css';
 import '@openclaw/design-system/themes.css';
 import '@openclaw/design-system/typography.css';
@@ -76,7 +77,7 @@ const structuredDataItems = structuredData ? Array.isArray(structuredData) ? str
     <title>{title}</title>
 
     {structuredDataItems.map((item) => (
-      <script type="application/ld+json" set:html={JSON.stringify(item)} />
+      <script type="application/ld+json" set:html={serializeJsonLd(item)} />
     ))}
 
     <script is:inline>

src/lib/serialize-json-ld.ts

@@ -0,0 +1,5 @@
+export function serializeJsonLd(value: Record<string, unknown>): string {
+  // Script contents are HTML raw text: removing every tag opener prevents an
+  // embedded </script> from ending the element while preserving JSON.parse.
+  return JSON.stringify(value).replaceAll('<', '\\u003c');
+}

tests/assert-built-assets.mjs

@@ -18,6 +18,14 @@ function readText(relativePath) {
   return readFileSync(repoPath(relativePath), 'utf8');
 }
 
+function listHtmlFiles(relativeDir) {
+  return readdirSync(repoPath(relativeDir), { withFileTypes: true }).flatMap((entry) => {
+    const relativePath = path.join(relativeDir, entry.name);
+    if (entry.isDirectory()) return listHtmlFiles(relativePath);
+    return relativePath.endsWith('.html') ? [relativePath] : [];
+  });
+}
+
 function assertFileExists(relativePath) {
   assert.ok(existsSync(repoPath(relativePath)), `Expected ${relativePath} to exist`);
 }
@@ -32,6 +40,13 @@ function assertContains(text, expected, context) {
   assert.ok(text.includes(expected), `Expected ${context} to contain ${expected}`);
 }
 
+function assertJsonLdIsParseable(html, context) {
+  const scripts = [...html.matchAll(/<script type="application\/ld\+json">([\s\S]*?)<\/script>/g)];
+  for (const [, contents] of scripts) {
+    assert.doesNotThrow(() => JSON.parse(contents), `${context} must contain parseable JSON-LD`);
+  }
+}
+
 assertCopiedByteForByte('public/openclaw-logo-text-dark.png', 'dist/openclaw-logo-text-dark.png');
 assertCopiedByteForByte('public/logo.png', 'dist/logo.png');
 assertCopiedByteForByte('public/granola.png', 'dist/granola.png');
@@ -54,6 +69,10 @@ const homePage = readText('dist/index.html');
 const integrationsPage = readText('dist/integrations/index.html');
 const ecosystemPage = readText('dist/ecosystem/index.html');
 
+for (const relativePath of listHtmlFiles('dist')) {
+  assertJsonLdIsParseable(readText(relativePath), relativePath);
+}
+
 assertContains(homePage, siX.path, 'dist/index.html');
 assertContains(homePage, siGooglechrome.path, 'dist/index.html');
 assertContains(homePage, siGmail.path, 'dist/index.html');

tests/serialize-json-ld.test.ts

@@ -0,0 +1,46 @@
+import { describe, expect, test } from 'bun:test';
+import { serializeJsonLd } from '../src/lib/serialize-json-ld';
+
+const breakoutPayloads = [
+  '</script><script id="breakout">alert(1)</script>',
+  '</ScRiPt><script>alert(1)</script>',
+  '</script\n><script>alert(1)</script>',
+  '<!--</script><script>alert(1)</script>-->',
+  '<script><!--</script><img src=x onerror=alert(1)>',
+  '<\\/script></script>',
+  'plain text & > text — 日本語',
+];
+
+describe('serializeJsonLd', () => {
+  test.each(breakoutPayloads)('keeps script-breakout payload parseable: %s', (payload) => {
+    const value = { nested: { payload }, list: [payload, { payload }] };
+    const serialized = serializeJsonLd(value);
+
+    expect(serialized).not.toContain('<');
+    expect(JSON.parse(serialized)).toEqual(value);
+  });
+
+  test('emits a literal JSON Unicode escape instead of decoding it to a less-than sign', () => {
+    expect(serializeJsonLd({ value: '<' })).toBe('{"value":"\\u003c"}');
+  });
+
+  test('round-trips deterministic property payloads without raw tag openers', () => {
+    let state = 0x5eed1234;
+    const alphabet = '</script>ScRiPt!&—日本語\\u003c\n\t"';
+
+    for (let sample = 0; sample < 512; sample += 1) {
+      let payload = '';
+      for (let index = 0; index < 96; index += 1) {
+        state ^= state << 13;
+        state ^= state >>> 17;
+        state ^= state << 5;
+        payload += alphabet[Math.abs(state) % alphabet.length];
+      }
+
+      const value = { sample, payload, nested: [{ payload }] };
+      const serialized = serializeJsonLd(value);
+      expect(serialized).not.toContain('<');
+      expect(JSON.parse(serialized)).toEqual(value);
+    }
+  });
+});