diff --git a/docs/development/agent-workflow.md b/docs/development/agent-workflow.md
index 459d081db..9ad64eb05 100644
--- a/docs/development/agent-workflow.md
+++ b/docs/development/agent-workflow.md
@@ -58,7 +58,7 @@ Explicit HTML anchors use `parse5`, excluding comments, scripts, styles and temp
 Rendered HTML anchor and image-map area hrefs and image sources use the same local-reference checks
 as Markdown links, including decoded attributes and fragment validation.
 URL attributes discard surrounding ASCII control/space characters before resolution.
-Iframe sources are document references and retain Markdown-target anchor checks.
+Iframe/embed sources and object data attributes are document references and retain Markdown-target anchor checks.
 Image and media references must resolve to files, not directories.
 Image references check file existence without interpreting image fragments as
 Markdown headings; document links keep anchor checks even when sharing a target.
@@ -85,7 +85,7 @@ separate package mentions from adjacent prose.
 Markdown destinations decode HTML entities before URI parsing, matching rendered links.
 Heading-anchor lookup is limited to Markdown targets. Source-file line fragments,
 PDF page fragments and other non-Markdown fragments retain file-existence checks.
-The import scan includes visible HTML text and literal backticks; it excludes parsed code nodes and non-rendered containers.
+The import scan separates HTML block/table elements and includes visible text and literal backticks; it excludes parsed code nodes and non-rendered containers.
 Navigation uses the parsed rendered tree too. Temporary in-memory markers retain
 definition, unresolved-reference and literal-path metadata, so Markdown inside
 inert templates is excluded consistently with raw HTML navigation.
diff --git a/tools/skills/agent-guidance-markdown.mjs b/tools/skills/agent-guidance-markdown.mjs
index 997b96476..166c608f5 100644
--- a/tools/skills/agent-guidance-markdown.mjs
+++ b/tools/skills/agent-guidance-markdown.mjs
@@ -78,16 +78,20 @@ function htmlNavigation(html, inspect = () => {}) {
                     'source',
                     'track',
                     'iframe',
+                    'embed',
                 ].includes(node.tagName) &&
                     attribute.name === 'src') ||
-                (node.tagName === 'video' && attribute.name === 'poster')
+                (node.tagName === 'video' && attribute.name === 'poster') ||
+                (node.tagName === 'object' && attribute.name === 'data')
             )
                 references.push({
                     target: attribute.value.replace(
                         /^[\u0000-\u0020]+|[\u0000-\u0020]+$/gu,
                         ''
                     ),
-                    image: !['a', 'area', 'iframe'].includes(node.tagName),
+                    image: !['a', 'area', 'iframe', 'object', 'embed'].includes(
+                        node.tagName
+                    ),
                 });
             if (
                 ['img', 'source'].includes(node.tagName) &&
@@ -113,6 +117,37 @@ export function guidanceProse(markdown) {
         if (node.nodeName === '#text') return node.value;
         const content = (node.childNodes ?? []).map(text).join('');
         return [
+            'address',
+            'article',
+            'aside',
+            'details',
+            'summary',
+            'dialog',
+            'dl',
+            'dt',
+            'dd',
+            'fieldset',
+            'legend',
+            'figure',
+            'figcaption',
+            'footer',
+            'form',
+            'header',
+            'hgroup',
+            'hr',
+            'main',
+            'nav',
+            'ol',
+            'ul',
+            'section',
+            'table',
+            'caption',
+            'thead',
+            'tbody',
+            'tfoot',
+            'tr',
+            'td',
+            'th',
             'p',
             'li',
             'blockquote',
@@ -125,7 +160,7 @@ export function guidanceProse(markdown) {
             'h5',
             'h6',
         ].includes(node.tagName)
-            ? content + '\n'
+            ? '\n' + content + '\n'
             : content;
     }
     return text(parseFragment(new Marked().parse(markdown)));
diff --git a/tools/skills/validate-agent-guidance.test.mjs b/tools/skills/validate-agent-guidance.test.mjs
index 63e36d28f..909594c2a 100644
--- a/tools/skills/validate-agent-guidance.test.mjs
+++ b/tools/skills/validate-agent-guidance.test.mjs
@@ -1334,3 +1334,39 @@ for (const reference of [
         );
     });
 }
+
+for (const html of [
+    '<details><summary>Use @angular/core</summary><p>for Angular</p></details>',
+    '<table><tr><td>Use @angular/core</td><td>for Angular</td></tr></table>',
+    '<section>Use @angular/core</section><aside>for Angular</aside>',
+]) {
+    test(`HTML blocks separate package prose: ${html}`, async (t) => {
+        assert.deepEqual(
+            await diagnostics(t, {
+                'AGENTS.md': html,
+                'package.json': JSON.stringify({
+                    dependencies: { '@angular/core': '*' },
+                }),
+            }),
+            []
+        );
+    });
+}
+for (const tag of ['object', 'embed']) {
+    test(`embedded document targets are checked: ${tag}`, async (t) => {
+        const markup = (target) =>
+            `<${tag} ${tag === 'object' ? 'data' : 'src'}="${target}"></${tag}>`;
+        for (const target of ['docs/missing.pdf', 'docs/example.md#missing']) {
+            assert.ok(
+                (await diagnostics(t, { 'AGENTS.md': markup(target) })).length >
+                    0
+            );
+        }
+        assert.deepEqual(
+            await diagnostics(t, {
+                'AGENTS.md': markup('docs/example.md#repeat'),
+            }),
+            []
+        );
+    });
+}