diff --git a/scripts/content.js b/scripts/content.js
index fde202a..05757d2 100644
--- a/scripts/content.js
+++ b/scripts/content.js
@@ -78,6 +78,9 @@
     "style",
     "noscript",
     "iframe",
+    "embed",
+    "object",
+    "frame",
     "nav",
     "aside",
     "form",
@@ -85,6 +88,37 @@
     "input",
   ];
 
+  /**
+   * Subtrees to omit from plain-text collection. `textContent` includes script/style/template
+   * bodies, which pulls in Astro/React/Vite inline bundles when walking `main div` etc.
+   * Also skip embed-like tags (iframe/object/embed) so news players do not dump URLs or attrs.
+   */
+  const TEXT_SUBTREE_EXCLUDE_TAGS = new Set([
+    ...EXCLUDE_TAGS,
+    "template",
+  ]);
+
+  function shouldExcludeTextSubtree(tagName) {
+    return TEXT_SUBTREE_EXCLUDE_TAGS.has(tagName.toLowerCase());
+  }
+
+  /**
+   * Strip embed markup that appears as literal text in article bodies (CMS/oEmbed fallbacks).
+   * Complements subtree skipping — some sites still surface tags as visible copy.
+   */
+  function sanitizeLiteralEmbedMarkup(text) {
+    if (!text || typeof text !== "string") {
+      return text;
+    }
+    let t = text;
+    t = t.replace(/<iframe\b[\s\S]{0,20000}?<\/iframe>/gi, "\n");
+    t = t.replace(/<iframe\b[^>]{0,8000}\/?>/gi, "\n");
+    t = t.replace(/<\/iframe>/gi, "");
+    t = t.replace(/<embed\b[^>]{0,8000}\/?>/gi, "\n");
+    t = t.replace(/<object\b[\s\S]{0,20000}?<\/object>/gi, "\n");
+    return t;
+  }
+
   function extractWithReadability() {
     const documentClone = document.cloneNode(true);
     const reader = new Readability(documentClone);
@@ -126,6 +160,7 @@
     }
 
     let content = article.textContent || "";
+    content = sanitizeLiteralEmbedMarkup(content);
 
     content = content
       .replace(/[^\S\n]+/g, " ")
@@ -134,7 +169,7 @@
 
     extractedText += content;
 
-    const bodyTextLen = (article.textContent || "").trim().length;
+    const bodyTextLen = content.trim().length;
 
     let wasTruncated = false;
     if (extractedText.length > MAX_LENGTH) {
@@ -177,7 +212,7 @@
       try {
         const elements = document.querySelectorAll(selector);
         for (const el of elements) {
-          const content = el.textContent.trim();
+          const content = getTextContent(el).trim();
           if (content.length < 20 || seen.has(content.substring(0, 100))) continue;
 
           const style = window.getComputedStyle(el);
@@ -200,7 +235,7 @@
     if (text.length < 500 && !wasTruncated) {
       const allParagraphs = document.querySelectorAll("p");
       for (const p of allParagraphs) {
-        const content = p.textContent.trim();
+        const content = getTextContent(p).trim();
         if (content.length > 30 && !seen.has(content.substring(0, 100))) {
           const style = window.getComputedStyle(p);
           if (style.display === "none" || style.visibility === "hidden") continue;
@@ -404,6 +439,10 @@
       } else if (node.nodeType === Node.ELEMENT_NODE) {
         const tagName = node.tagName.toLowerCase();
 
+        if (shouldExcludeTextSubtree(tagName)) {
+          continue;
+        }
+
         if (["br", "p", "div", "li"].includes(tagName)) {
           text += " " + getTextContent(node) + " ";
         } else {
@@ -416,7 +455,8 @@
   }
 
   function cleanExtractedText(text, shouldTruncate = true) {
-    let cleaned = text
+    let cleaned = sanitizeLiteralEmbedMarkup(text);
+    cleaned = cleaned
       .replace(/[^\S\n]+/g, " ")
       .replace(/\n{3,}/g, "\n\n")
       .replace(/^\s+|\s+$/g, "");