diff --git a/dist/index.js b/dist/index.js
index 32d5ca33adadd3928983ae76d78949190320cdab..056ab06ceb047e4bb08c2b0dacfd8b2a41540ec1 100644
--- a/dist/index.js
+++ b/dist/index.js
@@ -8,7 +8,7 @@ import {
 } from "@chat-adapter/shared";
 import { createAppAuth } from "@octokit/auth-app";
 import { Octokit } from "@octokit/rest";
-import { ConsoleLogger, convertEmojiPlaceholders, Message } from "chat";
+import { ConsoleLogger, convertEmojiPlaceholders, Message, toPlainText } from "chat";

 // src/cards.ts
 import { renderGfmTable } from "@chat-adapter/shared";
@@ -144,6 +144,55 @@ var GitHubFormatConverter = class extends BaseFormatConverter {
   }
 };

+// GitHub comments commonly represent uploaded files as Markdown images. The
+// shared plain-text converter preserves their label but drops the destination,
+// which makes the uploaded artifact impossible to recover downstream. Append
+// a bounded set of safe HTTPS destinations without treating them as fetchable
+// attachments or retaining credential-bearing URLs.
+function safeGitHubMarkdownUrl(value) {
+  if (typeof value !== "string" || value.length === 0 || value.length > 2048) {
+    return null;
+  }
+  let parsed;
+  try {
+    parsed = new URL(value);
+  } catch {
+    return null;
+  }
+  if (parsed.protocol !== "https:" || parsed.username || parsed.password) {
+    return null;
+  }
+  return parsed.href;
+}
+function normalizedGitHubCommentText(formatted) {
+  const plainText = toPlainText(formatted);
+  const urls = [];
+  const seen = /* @__PURE__ */ new Set();
+  const stack = [formatted];
+  let visited = 0;
+  while (stack.length > 0 && urls.length < 32 && visited < 1e4) {
+    const node = stack.pop();
+    visited += 1;
+    if (node && typeof node === "object") {
+      if (node.type === "link" || node.type === "image" || node.type === "definition") {
+        const url = safeGitHubMarkdownUrl(node.url);
+        if (url && !seen.has(url)) {
+          seen.add(url);
+          urls.push(url);
+        }
+      }
+      if (Array.isArray(node.children)) {
+        for (let index = node.children.length - 1; index >= 0; index -= 1) {
+          stack.push(node.children[index]);
+        }
+      }
+    }
+  }
+  const missingUrls = urls.filter((url) => !plainText.includes(url));
+  if (missingUrls.length === 0) return plainText;
+  return [plainText, ...missingUrls].filter(Boolean).join("\n");
+}
+
 // src/index.ts
 var REVIEW_COMMENT_THREAD_PATTERN = /^([^/]+)\/([^:]+):(\d+):rc:(\d+)$/;
 var ISSUE_THREAD_PATTERN = /^([^/]+)\/([^:]+):issue:(\d+)$/;
@@ -677,11 +726,12 @@ var GitHubAdapter = class {
    */
   parseIssueComment(comment, repository, prNumber, threadId, threadType = "pr") {
     const author = this.parseAuthor(comment.user);
+    const formatted = this.formatConverter.toAst(comment.body);
     return new Message({
       id: comment.id.toString(),
       threadId,
-      text: this.formatConverter.extractPlainText(comment.body),
-      formatted: this.formatConverter.toAst(comment.body),
+      text: normalizedGitHubCommentText(formatted),
+      formatted,
       raw: {
         type: "issue_comment",
         comment,
@@ -709,11 +759,12 @@ var GitHubAdapter = class {
    */
   parseReviewComment(comment, repository, prNumber, threadId) {
     const author = this.parseAuthor(comment.user);
+    const formatted = this.formatConverter.toAst(comment.body);
     return new Message({
       id: comment.id.toString(),
       threadId,
-      text: this.formatConverter.extractPlainText(comment.body),
-      formatted: this.formatConverter.toAst(comment.body),
+      text: normalizedGitHubCommentText(formatted),
+      formatted,
       raw: {
         type: "review_comment",
         comment,