Turn an RSS feed into a curated Notion knowledge base
When a new article hits your RSS feed, an agent pulls the full clean text, decides if it's on topic for your beat, and files a structured entry in Notion.
Build me an agent that turns an RSS feed into a curated knowledge base in Notion. The trigger should be a poll on an RSS feed for new_item events. I will provide the feed URL when I deploy.
On each new feed item, the agent should call Diffbot's Extract Article operation against the item URL to get clean body text, author, publish date, tags, sentiment, and images. Don't rely on the raw RSS body, it is usually truncated or full of HTML noise.
After extraction, the agent should make three judgement calls: (1) is this article on topic for the configured beat? If not, drop it and exit. (2) write a 4 to 6 sentence summary in a neutral, informative tone. (3) generate 3 to 5 topic tags. The beat description should be a configurable input that I can edit, for example 'enterprise AI infrastructure and developer tooling'.
For on-topic items, use Notion's Create a Page operation to add a new entry to a target Notion database (database ID configurable). Fill these properties: Title (article title), Source (feed name or publication), Author, Published Date, URL, Sentiment (from Diffbot), Tags (the agent-generated tags), and Summary (the 4 to 6 sentence summary). Then use Notion's Append Block Children operation on the new page to append the full extracted article body as page content, broken into paragraph blocks. Include the lead image at the top if Diffbot returned one.
Before creating a page, check the database for an existing entry with the same URL property and skip if it already exists, so we don't get duplicates from feed republishes.
Integrations: RSS (poll trigger, new_item), Diffbot (Extract Article), Notion (Create a Page, Append Block Children). Configurable inputs: feed URL, beat description, target Notion database ID.
Additional information
What does this prompt do?
- Watches any RSS or Atom feed and reacts the moment a new article shows up.
- Pulls the full clean article body, author, publish date, images, and tags so you don't end up with the truncated junk most feeds publish.
- Skips anything that isn't on topic for the beat you defined, so your knowledge base stays focused.
- Writes a 4 to 6 sentence summary, picks 3 to 5 topic tags, and files a tidy entry in your Notion database with the full article appended as page content.
What do I need to use this?
- A Notion workspace with a database set up for your knowledge base. The database should have properties for Title, Source, Author, Published Date, URL, Sentiment, Tags, and Summary.
- A Diffbot account for clean article extraction. A free trial works for low volumes.
- An RSS or Atom feed URL for the source you want to monitor. You can swap or add feeds at any time.
- A short description of the beat or topics you want to cover, so the agent knows what to keep and what to drop.
How can I customize it?
- Change the feed URL or add multiple feeds covering the same beat.
- Tighten or loosen the on-topic filter by editing the beat description, for example narrowing from 'AI' to 'AI infrastructure for enterprise'.
- Adjust the summary length and tone, or swap the tag style for fixed categories your team already uses.
- Point at a different Notion database or change which properties get filled in.
Frequently asked questions
Why use Diffbot instead of just reading the RSS body?
Can I monitor more than one feed?
How does the agent decide what is on topic?
What if my Notion database has different properties?
Will duplicate articles get filed twice?
Stop letting good articles rot in your unread feed.
Connect Diffbot and Notion once, and Geni files a clean, summarized entry every time your beat publishes something new.