diff --git a/.jules/sentinel.md b/.jules/sentinel.md
index 5e0d967e..78d69e6b 100644
--- a/.jules/sentinel.md
+++ b/.jules/sentinel.md
@@ -14,3 +14,7 @@
 **Vulnerability:** Several `escapeHtml` implementations used DOM manipulation (`document.createElement('div').innerHTML`) or incomplete string replacement, failing to escape single and double quotes.
 **Learning:** Incomplete escaping allows XSS payloads to break out of HTML attributes (e.g., `<input value="${escapeHtml(userInput)}">`).
 **Prevention:** Always use `String(text)` cast combined with a comprehensive replace chain for `&`, `<`, `>`, `"`, and `'` (e.g., `&#039;`) in custom string escaping functions.
+## 2026-06-23 - Missing Sanitization in Markdown Rendering in site/app.js
+**Vulnerability:** Untrusted user input via markdown notes could be rendered into HTML and appended to the DOM via `innerHTML` without sanitization when `marked.parse` is available.
+**Learning:** Even when using a markdown parser, the resulting HTML can contain malicious script tags or attributes if the original markdown contains raw HTML. The markdown-to-HTML step is not an automatic sanitization step.
+**Prevention:** Always sanitize the resulting HTML from markdown parsers (e.g., using DOMPurify) before inserting it into the DOM via `innerHTML`.
diff --git a/site/app.js b/site/app.js
index 788bd7e7..03e6993e 100644
--- a/site/app.js
+++ b/site/app.js
@@ -2724,7 +2724,7 @@ function renderSpecsPanel() {
 
       if (typeof marked !== 'undefined') {
         const rendered = marked.parse(processedNote);
-        html += rendered;
+        html += window.DOMPurify ? window.DOMPurify.sanitize(rendered, { ADD_ATTR: ['data-note-node', 'data-node'] }) : `<pre>${processedNote.replace(/</g, '&lt;').replace(/>/g, '&gt;')}</pre>`;
       } else {
         html += `<pre>${processedNote.replace(/</g, '&lt;').replace(/>/g, '&gt;')}</pre>`;
       }