Automation
Digest a site daily
Pull today's items from a source and lay them out as something readable.
Fill it in
The site or feed to read.
How many to include.
What you care about, so the rest is dropped.
A line on what it says, not just the headline.
Your prompt
Build a digest of the top 20 items from [your source]. Only include: [your only include] Use the site's API if it has one. HTML scraping breaks the first time somebody changes a class name, and a digest that silently returns nothing is worse than one that never existed. If there is no API, parse defensively and say loudly when the shape stops matching. Summarise each item in one line that says what it ACTUALLY claims, not what the title implies. Headlines are written to be clicked. Lay it out as a readable page: source, points or signal where available, and a direct link. Group by theme if a theme is obvious, chronological otherwise.
Use Digest a site dailyOpens with everything above already filled in.
Why this works
Most digests break silently. They scrape HTML, someone renames a class, and the digest returns nothing while still arriving on time. This uses the API where one exists and says loudly when the shape stops matching, and it summarises what an item actually claims rather than what the headline implies.
Questions
- Can I run it daily?
- Yes, pair it with a schedule and it lands on its own.