Skip to content
← All posts

Building with Claude

Building with Claude: Don't Delete Your Old Site — Redirect It

June 19, 2026 · 4 min read

I have an old freelancer portfolio living at akashmony01.github.io. Now that I’ve built a new site, I wanted to clean up the old one — and I was about to do it the wrong way twice. This is the conversation that saved me from both mistakes.

The problem we were solving

I needed a GitHub profile README, and I thought the old portfolio repo was in the way. My plan, roughly: make the old site’s homepage look exactly like my new homepage, link everything over to the new domain, and call it a “branded entry point.” Failing that, just delete the old site entirely.

What Claude suggested

First, Claude untangled a misconception I didn’t know I had. The profile README lives in a repo named exactly after your username (akashmony01/akashmony01) and renders on your GitHub profile. The old site lives in a different repo (akashmony01.github.io). They’re unrelated — I didn’t need to touch the old site to get a README at all.

Then came the part that mattered. My “make it look like the new homepage” idea was a duplicate-content trap. If two domains serve the same page, Google has to pick one to rank — and the older, higher-authority github.io domain could win, burying my actual site with a copy of itself. You’d be competing against yourself, and losing.

The right move was a clean redirect. Don’t copy the content — point at it:

<!doctype html>
<html lang="en">
<head>
  <meta charset="utf-8" />
  <title>Mohammed Akash — now at mohakash.xyz</title>
  <link rel="canonical" href="https://mohakash.xyz/" />
  <meta http-equiv="refresh" content="0; url=https://mohakash.xyz/" />
  <script>window.location.replace("https://mohakash.xyz/");</script>
</head>
<body>
  <p>This site has moved to <a href="https://mohakash.xyz/">mohakash.xyz</a>.</p>
</body>
</html>

The rel="canonical" line is the important one: it tells Google “the real version lives at mohakash.xyz,” so the old page passes its authority along instead of competing. And because github.io is a high-trust domain, that redirect is actually a good backlink for the new site.

For the cleanup itself, Claude suggested preserving the old site on a branch and stripping main down to just the redirect:

git checkout main
git checkout -b old-portfolio   # keep the old site safe
git push -u origin old-portfolio
git checkout main
git rm -rf .                    # strip main to nothing
# add the redirect index.html, then:
git commit -m "Redirect old site to mohakash.xyz"
git push origin main

What I got wrong

Two things. First, I assumed the old site was clutter — something to erase. It’s not; it’s equity. It’s older than my new domain and probably already indexed, so it carries SEO weight I’d be throwing away by deleting it.

Second, my “make it identical” instinct felt clever but was actively harmful. Cloning the homepage would have split my search signals across two domains instead of consolidating them onto one. The thing that felt like more presence was really self-sabotage.

What Claude got wrong

Before handing me redirect steps, Claude assumed it was a plain static site. It wasn’t — it was a Jekyll site, which changes how GitHub Pages serves files (an index.html can clash with Jekyll’s index.md). Claude only caught this after I pushed back and it actually inspected the repo. It should have checked the repo structure first instead of giving generic instructions and correcting them after. The lesson cuts both ways: verify the thing in front of you before prescribing steps for it.

What we solved

  • The old portfolio now instantly redirects to mohakash.xyz, passing its authority along instead of competing with it.
  • The old site is preserved on an old-portfolio branch — nothing lost.
  • A profile README went up in the correct repo, with a real backlink to the new site.
  • Zero Cloudflare involvement: the old site is on GitHub Pages, so it was all GitHub, no DNS changes.

Key lessons learned

  1. An old site is an asset, not trash. Redirect it to pass authority and capture its existing traffic; deleting throws that away.
  2. Never duplicate a page across two domains. It’s a canonicalization gamble you can lose — the wrong copy ranks. Link, don’t clone.
  3. rel="canonical" is how you redirect a static host. GitHub Pages can’t do a real 301, but a meta-refresh plus a canonical tag passes the signal cleanly.
  4. A no-click redirect beats a “click here to continue” page. It forwards instantly and does the same SEO work, with less friction for the visitor.