The Tripword bot · Tripword

Tripword watches RSS and Atom feeds and alerts people when a new post matches their keywords. If you found this page from your server logs, this is what made the request and how to control it.

How to recognize it

Every request sends this User-Agent:

Tripword/1.0 (+https://tripword.app/bot)

Requests come from Cloudflare Workers, so they arrive from Cloudflare's network with no fixed IP addresses. Cloudflare adds a CF-Worker header to each one. We never send a browser User-Agent.

What it requests

Feed URLs only: the RSS or Atom file a person chose to watch, or one listed in our source catalog.
When someone adds a site by its address, we fetch that one page and try a handful of common feed paths to find the feed. That happens once.
It does not crawl. It does not follow links or fetch article pages.
The Tripword home page shows a few recent posts from a short list of sources, with the title, a link back, and the image the publisher attached to that post. Each one is credited to its source.

How often

A watched feed is checked at most once a minute. Many are set slower.
A catalog feed is checked about every 30 minutes.
Requests are conditional. We send If-None-Match and If-Modified-Since, so an unchanged feed can answer 304 with no body.

How it backs off

A 429 or 503 with a Retry-After header is honored, up to 24 hours.
A 429 without that header pauses the feed for 15 minutes.
A feed that fails three checks in a row drops to one request every 15 minutes. After ten, one an hour.

What it keeps

Each item's ID, title, link, and timestamp, so the same post never alerts twice. For posts with no title, the first 300 characters stand in for it.
Catalog entries are deleted after 30 days.
We do not store or republish article bodies. An alert carries the title, a short excerpt from the feed, and a link to your page.

robots.txt

Tripword requests feed files that sites publish for readers like it, so it does not read robots.txt today. If you would rather it stayed away, use either option below.

How to stop it

Block the User-Agent. Matching on Tripword is enough. A 403 works, and the bot slows down on its own as described above.
Email bot@tripword.app with your domain. We will remove your feeds from the catalog and stop requesting them.

If the bot is misbehaving on your site, the same address reaches the person who can fix it.