Build a Podcast Episode Image Extractor with GPT

1. Get the RSS of this podcast

Thanks to

First, find the Apple Podcast URL of the podcast
https://podcasts.apple.com/cn/podcast/独树不成林/id1711052890
Pay attention to the id part

Then append the id to the end of this URL
https://itunes.apple.com/lookup?id=
To get
https://itunes.apple.com/lookup?id=1711052890
Open it in a browser to get a JSON file.
Open it with a text tool (or use a JSON tool such as https://www.json.cn/)
Find feedurl
https://feed.xyzfm.space/y9qnpfdrctnx
This is the RSS.

2. Find the cover images

Open the RSS link in a browser, and look for sections like the one below — those are the cover images
<itunes:image href="https://image.xyzcdn.net/Fgd_z5yexkQF_GB0LF4Xncqqf8CU.png"/>

3. Develop with GPT

Implement an HTML+JS-based tool
The page includes the following sections
1) A text box to input the podcast's RSS URL
2) A text box that fetches the RSS content from 1), and also allows manual input of RSS content
3) A text box that analyzes the content of 2), lists the URLs of all cover images. It should allow manual editing to add or remove entries
4) Display the images corresponding to the URLs in 3)

Example:
https://feed.xyzfm.space/y9qnpfdrctnx is a podcast's RSS URL, which will be entered into 1)
1) fetches the RSS data in XML format, which is then filled into 2)
2) contains elements like <itunes:image href="https://image.xyzcdn.net/Fgd_z5yexkQF_GB0LF4Xncqqf8CU.png"/>, and https://image.xyzcdn.net/Fgd_z5yexkQF_GB0LF4Xncqqf8CU.png should be filled into 3)
4) displays the images corresponding to the URLs in 3)
Send these requirements to GPT.
The result I got is quite satisfying.

Upload to Github


You can also access it via Github Page

4. Pick the images I find interesting and turn them into Telegram stickers

You need to resize the images. I used M$ PowerToys

You need to visit Telegram's official sticker bot

Result:

========

Postscript

For this development (2025-11-20), the GPT I used was



Comments