Unlink confirmed-dead external URLs at render time (#361)

Link Watch found 404 external links (2026-10-17). Post bodies are webhook/user data, so anchors to known-dead URLs are unwrapped at render (text kept) instead of editing rows.

Co-authored-by: Claude Sonnet 5.5 <noreply@anthropic.com>

Author: ralyodio

app/(marketing)/blog/[slug]/page.tsx

@@ -1,6 +1,7 @@
 import { notFound, permanentRedirect } from "next/navigation";
 import { allSlugs, findAnyPost } from "@/lib/blog/posts";
 import { fixBlogLinks, resolveMissingSlug } from "@/lib/blog/links";
+import { stripDeadLinks } from "@/lib/dead-links";
 import { splitHtmlForMidAd } from "@/lib/blog/split-html";
 import { AdUnit } from "@/components/ads/ad-unit";
 import { env } from "@/lib/env";
@@ -56,7 +57,7 @@ export default async function BlogPost({
     if (target) permanentRedirect(target);
     notFound();
   }
-  if (post.html) post.html = fixBlogLinks(post.html, slugs);
+  if (post.html) post.html = stripDeadLinks(fixBlogLinks(post.html, slugs));
 
   // Null for short posts and for plain-text (non-HTML) bodies — those render
   // straight through with just the top and bottom units.

components/report/markdown-view.tsx

@@ -1,4 +1,5 @@
 import { markdownToHtml } from "@/lib/markdown";
+import { stripDeadLinks } from "@/lib/dead-links";
 
 function wrapTables(html: string): string {
   return html.replace(
@@ -8,7 +9,7 @@ function wrapTables(html: string): string {
 }
 
 export async function MarkdownView({ markdown }: { markdown: string }) {
-  const html = wrapTables(await markdownToHtml(markdown));
+  const html = stripDeadLinks(wrapTables(await markdownToHtml(markdown)));
   return (
     <article
       className="card prose max-w-none overflow-hidden p-4 sm:p-6 [&_*]:min-w-0 [&_a]:break-words [&_a]:underline [&_code]:break-words [&_h1]:mt-0 [&_h1]:break-words [&_h1]:text-2xl [&_h1]:font-extrabold sm:[&_h1]:text-3xl [&_h2]:mt-8 [&_h2]:break-words [&_h2]:border-t [&_h2]:border-[var(--color-border)] [&_h2]:pt-6 [&_h2]:text-xl [&_h2]:font-bold [&_h3]:mt-4 [&_h3]:break-words [&_h3]:font-semibold [&_hr]:my-8 [&_hr]:border-[var(--color-border)] [&_li]:my-1 [&_li]:break-words [&_p]:my-3 [&_p]:break-words [&_pre]:overflow-x-auto [&_pre]:rounded [&_pre]:bg-[var(--color-bg)] [&_pre]:p-3 [&_pre]:text-xs [&_table]:my-0 [&_table]:w-full [&_table]:border-collapse [&_td]:border [&_td]:border-[var(--color-border)] [&_td]:px-3 [&_td]:py-2 [&_td]:text-sm [&_th]:border [&_th]:border-[var(--color-border)] [&_th]:bg-[var(--color-bg)] [&_th]:px-3 [&_th]:py-2 [&_th]:text-left [&_th]:text-xs [&_th]:uppercase [&_th]:text-[var(--color-muted)] [&_ul]:list-disc [&_ul]:pl-6 [&_.md-table-wrap]:my-4 [&_.md-table-wrap]:-mx-4 [&_.md-table-wrap]:overflow-x-auto [&_.md-table-wrap]:px-4 sm:[&_.md-table-wrap]:-mx-6 sm:[&_.md-table-wrap]:px-6"

lib/dead-links.ts

@@ -0,0 +1,54 @@
+// Links to other sites that are confirmed dead (404 on re-check 2026-10-17, from
+// the ThreatCrush Link Watch report). Post bodies arrive by webhook or as user
+// content, so rather than hand-edit rows the dead anchors are unwrapped at
+// render time: the link goes, the text stays.
+
+const DEAD_URLS = new Set<string>([
+  "https://analytics.ahrefs.com",
+  "https://www.googletagmanager.com",
+  "https://bxiie.artsfol.io/artwork/in-paint-bxiie-br-34",
+  "https://c0upons.com/blog/doordash-promo-codes-for-existing-users-2026-workflow",
+  "https://c0upons.com/blog/encrypted-messaging-coupon-codes-private-deal-sharing",
+  "https://c0upons.com/blog/shutterfly-promo-codes-2026-checkout-workflow",
+  "https://c0upons.com/blog/walgreens-coupon-code-workflow",
+  "https://crates.io/crates/graphforge-rel",
+  "https://crates.io/crates/piu_node_resolver",
+  "https://crates.io/crates/rightkit-credentials",
+  "https://crates.io/crates/willbooster-tree-sitter-rust",
+  "https://crates.io/crates/wsi-dicom-annotations",
+  "https://devdrafts.netlify.app",
+  "https://feedbackminer.co/cookies",
+  "https://feedbackminer.co/privacy",
+  "https://feedbackminer.co/sub-processors",
+  "https://feedbackminer.co/terms",
+  "https://seniorcontact.app/en/about-us",
+]);
+
+/** Placeholder hosts (RFC 2606): never a real page, so never worth a link. */
+const PLACEHOLDER_HOST = /^(?:www\.)?example\.(?:com|org|net)$/i;
+
+function normalize(href: string): string {
+  return href.replace(/#.*$/, "").replace(/\/+$/, "").toLowerCase();
+}
+
+export function isDeadExternalUrl(href: string): boolean {
+  let url: URL;
+  try {
+    url = new URL(href);
+  } catch {
+    return false;
+  }
+  if (url.protocol !== "http:" && url.protocol !== "https:") return false;
+  if (PLACEHOLDER_HOST.test(url.hostname)) return true;
+  if (url.pathname.startsWith("/cdn-cgi/l/email-protection")) return true;
+  return DEAD_URLS.has(normalize(href));
+}
+
+const ANCHOR = /<a\s[^>]*?href="([^"]*)"[^>]*>([\s\S]*?)<\/a>/gi;
+
+/** Unwrap every anchor that points at a known-dead URL, keeping its text. */
+export function stripDeadLinks(html: string): string {
+  return html.replace(ANCHOR, (whole, href: string, inner: string) =>
+    isDeadExternalUrl(href.replace(/&amp;/g, "&")) ? inner : whole,
+  );
+}

tests/dead-links.test.ts

@@ -0,0 +1,21 @@
+import { describe, expect, it } from "vitest";
+import { isDeadExternalUrl, stripDeadLinks } from "@/lib/dead-links";
+
+describe("stripDeadLinks", () => {
+  it("unwraps dead anchors and keeps the text", () => {
+    const html = '<p>See <a href="https://c0upons.com/blog/walgreens-coupon-code-workflow">this</a>.</p>';
+    expect(stripDeadLinks(html)).toBe("<p>See this.</p>");
+  });
+  it("matches bare origins with or without a trailing slash", () => {
+    expect(isDeadExternalUrl("https://www.googletagmanager.com/")).toBe(true);
+    expect(isDeadExternalUrl("https://www.googletagmanager.com")).toBe(true);
+  });
+  it("drops placeholder hosts and email-protection links", () => {
+    expect(isDeadExternalUrl("https://example.com/blog/post/")).toBe(true);
+    expect(isDeadExternalUrl("https://x.org/cdn-cgi/l/email-protection#ab")).toBe(true);
+  });
+  it("leaves live links alone", () => {
+    const html = '<a href="https://crawlproof.com/blog/x">x</a> <a href="https://crates.io/crates/serde">s</a>';
+    expect(stripDeadLinks(html)).toBe(html);
+  });
+});