How to move your blog from another website to TeleportHQ with ChatGPT or Claude
In TeleportHQ, your own AI tool can move a blog you publish elsewhere, such as WordPress, Medium, Substack, Ghost, Squarespace or Wix: copy the prompt below into an AI that can browse the web, such as Claude Code, Cursor or ChatGPT. It asks for your blog's web address, finds every post through the feed and the sitemap, reads each one, downloads its pictures, writes one Markdown file per post with its date, author, tags, main picture and old address, puts the posts in folders that become your categories, and zips the folder. Drop that ZIP on the Import Posts button in the Blog panel, review what will be created, and your posts are live at /blog, dated as they were, with the old addresses redirected to the new pages.
Before you start
- Your project needs the Blog switched on. See How to add a blog to your site.
- You need an AI tool that can open web pages and save files: for example Claude Code or Cursor, or ChatGPT with browsing. The old blog must be online and public while the AI reads it.
- The ZIP holds one
.mdfile per post with the pictures next to it; a folder is a category. The full format is in How to import Markdown files into your blog or help center, and the prompt below tells the AI all of it. - A blog that exports CSV can skip the AI: see How to import or export blog posts.
Steps
1. Copy the prompt
Copy the whole block below. It tells the AI to find every post at your address, what to keep from each one, where the pictures go, how to record the date, the author and the old address, and to end by zipping the folder.
You are moving my blog to TeleportHQ as Markdown files. First ask me, and wait for the answer: the web address of the blog to move (for example https://example.com/blog). Ask nothing else.
Then open that address and find every post: read the feed first (/feed, /rss.xml, /feed.xml, /atom.xml, or the feed link in the page's head), then /sitemap.xml and /sitemap_index.xml, then the blog's list pages with every "older posts", "next" and page-number link, until nothing new turns up. Tell me how many posts and categories you found before you write anything.
Read every post page and write it as one Markdown file, keeping what a reader sees: the title, the text, headings, lists, tables, quotes, code blocks with their language, embedded videos as their links, links and pictures. Do not rewrite, shorten or improve the text.
Deliver this folder, zipped as blog.zip:
blog/
news/
_category.md
our-new-office.md
pictures/
our-new-office-cover.jpg
our-new-office-1.jpg
guides/
choose-a-standing-desk.md
Rules:
- One folder per category of the old blog, one .md file per post. A post in several categories goes in the folder of its first one, with the others listed under tags. A blog without categories puts every post at the top of the folder.
- Name every folder and file with lowercase letters, digits and dashes only, no numbers in front: your site lists posts by date. Take the file name from the post's old address when it has a readable one: our-new-office.md becomes "Our new office" at /blog/our-new-office.
- File names must be unique across all folders: each is its post's web address.
- Each post starts with front matter between two --- lines: title (the old title), description (the old excerpt or subtitle, or the first sentence), date (the old published date, written 2026-03-15), last_verified (the old updated date when the page shows one), author (the old author's name), tags (the old tags, as [design, remote-work]), image (the main picture, as pictures/our-new-office-cover.jpg), image_alt (what the main picture shows), and redirect_from (a list of the post's old web address paths, such as [/2024/03/15/our-new-office]).
- Add a _category.md (name, description, redirect_from with the category's old address paths) when the folder's name is not the old category's name.
- Download EVERY picture of a post, its main picture included, into the pictures folder next to it, named after the post: pictures/our-new-office-cover.jpg for the main picture, then pictures/our-new-office-1.jpg, -2.jpg, and so on in reading order, keeping each picture's format (png, jpg, gif, webp). Put each one in the text where it was: .
- Link to another post of the blog with a Markdown link whose address is its file's path, relative to the current file: ../guides/choose-a-standing-desk.md. Keep links to other websites as they are.
- Leave out menus, headers, footers, share buttons, "related posts" boxes, comments, newsletter forms, cookie notices and anything else that is not the post.
- Finish by zipping the blog folder. Then tell me how many posts and pictures you wrote, and name any page you could not read or picture you could not download.
- If you cannot download files, keep each picture's full web address in the Markdown instead, as  and image: https://…, and tell me at the end that the pictures stayed on the old site.2. Give it to your AI and answer its question
- Open your AI tool where it can browse the web and save files: Claude Code or Cursor in an empty folder, or a ChatGPT conversation with browsing on.
- Paste the prompt and send it.
- Answer its one question with your blog's web address, for example
https://brightly.app/blog. - Let it work. It reads the posts, downloads the pictures, writes the files, then zips the
blogfolder intoblog.zip. If your tool can't make a ZIP, zip theblogfolder yourself with your computer's compress command.
3. Check the ZIP
Open the blog folder before you import it and check:
- The count matches: as many
.mdfiles as your old blog has posts. Open two or three and compare them with the old pages. - Every post starts with a
title, adateand aredirect_frombetween two---lines. The date is the old published date, written like2024-03-15; theredirect_frompaths are the old addresses without the domain. - The main picture is named under
imageand sits in thepicturesfolder next to the post, with the other pictures, at the paths the post points to. If the AI said the pictures stayed on the old site, the posts point to web addresses instead; they show as long as the old site is online. - Every file name is different from every other, in every folder: the file name is the post's address.
4. Import it
- Open your project in the TeleportHQ editor.
- In the left sidebar, click Blog.
- At the top right of the Blog panel, click the upload arrow (tooltip Import Posts). While the list is empty, the panel also offers Import posts below Create Post.
- Drop
blog.zipon the box, or click the box to choose it. "Reading the file…" shows while TeleportHQ reads it. - Read the plan. Nothing is written yet. It counts what the import will do, for example "120 articles", "5 categories from the folders", "300 pictures to upload to your files" and "120 old addresses to lead to their new pages, with permanent redirects". A box headed with a number of problems lists files that won't be imported, and a box of notes lists, file by file, what will be left out or changed. The messages are explained under Troubleshooting.
- Leave Publish imported articles on to put the posts live, or turn it off to save them as drafts.
- Click Import followed by the number, for example Import 120 articles. The bar shows Creating categories, Uploading pictures, then Writing articles, and a message sums it up, for example "Import finished: 120 posts added, 5 categories created, 300 pictures in your files, 120 old addresses redirected".
5. Switch the address and read it on your site
- Publish your site. See How to publish a web app. The redirects reach your live site when you publish; after that, new and edited posts show within seconds, without publishing again.
- Open
/blogon your site: the posts are listed newest first, with their old dates. Open an old address, such ashttps://your-site.com/2024/03/15/our-new-office: it lands on the new post. - Point your old blog's domain at your TeleportHQ site when you are ready, so the old links in search results and newsletters keep working. See How to connect your own domain.
- For a wrong title, picture or paragraph, fix the
.mdfile, or ask your AI to, then zip the folder and import it again. Posts are matched by their address: a changed file updates its post in place, a new file is added, and the rest stay as they are. See Import again to update.
Check that it worked
- The categories show as pills at the top of the Blog panel.
- Open a post in the panel: its date is the old one, its author is on your authors list, and its main picture, pictures and links are in place.
- In Project settings (cog, bottom left) → SEO & social → Search engines → Custom Redirects, the old addresses are listed, each leading to its new post.
- On your live site,
/bloglists the posts by date, and each opens at/blog/followed by its file name.
Troubleshooting
The AI found far fewer posts than your blog has A feed lists only the latest posts, and some blogs load older posts as you scroll. Give the AI the address of the sitemap (/sitemap.xml) or of the archive pages, and ask it to continue from there. If the old platform can export your posts (WordPress, Ghost, Substack and Medium can), give the AI the export too.
Every post shows today's date The posts have no date in their front matter, so each was dated at its import. Ask the AI to add date with each post's old published date and import the ZIP again: the posts are updated in place.
The authors are wrong or missing An author that matches a name on your authors list credits that author; another name becomes the post's own author. A post without an author gets your list's default author. See How to add authors to your posts and articles.
"No Markdown articles found" The ZIP holds no .md files outside left-out folders (those whose names start with _ or .). Open the ZIP and check that the posts are there and that their folders are named without a leading _ or ..
A note says a picture "is not in the bundle, dropped" The post points to a picture that isn't in the ZIP at that path, usually one the AI could not download. Save the picture from the old page yourself, put it next to the post at that path, and import again.
The plan says pictures are .avif (or .heic, .heif, .bmp, .tif, .tiff) "which TeleportHQ cannot upload" TeleportHQ doesn't upload pictures in those formats, so the import leaves them out of the posts. Convert them to .png, .jpg or .webp, change the picture paths in the posts to the new file names, zip the folder again and import it.
A category's old address is not redirected, and the import's notes say so A category's old addresses need category pages. Turn on Category Pages in the Blog's settings (Settings → Listing) and import again. See How to use blog categories, tags and related posts.
Good to know
- Keep the old site online until the switch is done. The AI reads it while it works, and posts whose pictures stayed on the old site keep showing from there.
- Old addresses become permanent redirects. Each path under
redirect_frombecomes a permanent redirect (301) to the post's new page, and importing again keeps them up to date. A*stands for any characters:/2024/*/our-new-officetakes every old address ending that way. See Old addresses become permanent redirects. - Comments don't move. The import brings the posts, not the comments under them.
- What a re-import replaces. A post's title, text, excerpt, main picture, categories, tags, SEO title and description come from the file again, so a change you made to those in TeleportHQ is replaced. Its author, place in its category, related posts, status and dates change only when the file sets them. Its address never changes.
- The Help Center moves the same way. See How to move your help center from another website to TeleportHQ with ChatGPT or Claude.