BulbeProblemEngine

Last updated: October 2026

BulbeProblemEngine is the name Bulbe's server reads public posts and web pages under, for Bulbe, the app operated by Trade Vessel P.C. of Santorini, Greece. Bulbe groups public complaints into problems and shows the quotes behind each one. Every request the server makes for this carries the user agent below.

1. Its user agent

BulbeProblemEngine/1.0 (+https://bulbe.app/bot; research on public complaints)

A robots.txt names it by its first word, BulbeProblemEngine, in any mix of capitals.

2. What it reads

Public posts, through the public interfaces of the sites they are on, such as Reddit, Hacker News, Stack Exchange, GitHub, app store reviews and the US FDA's reports about medical devices, each under that site's own terms for its interface
A short list of public product forums: only each forum's list of feature requests and the requests in it, as the forum software publishes them, no faster than the forum's robots.txt asks. Each forum's robots.txt and terms were read before it was added
Web pages a web search named as the source of a quote: the server reads the page as text, once, to check that the quote is on it word for word, and drops the quote when it is not

From a web page it keeps only the quotes it found there word for word, each with the page's address, and none of the page's other text. What Bulbe keeps from public posts, for how long, and how to have a quote removed is in the privacy policy at bulbe.app/privacy.

3. How it follows a site's word

Before it reads a page it reads the site's robots.txt and its /.well-known/tdmrep.json, and keeps what they say for up to a day
It obeys the robots.txt group that names BulbeProblemEngine, and otherwise the group for every crawler (User-agent: *)
A robots.txt that answers with a server error, with 429 or not at all closes the whole site to it, and it does not ask again for 10 minutes. A robots.txt that is not there leaves the site open
It never keeps a page that reserves itself from text and data mining: tdm-reservation: 1 in the page's headers, in a meta tag or in tdmrep.json, or noai in an X-Robots-Tag header or a robots meta tag, said to every reader or to BulbeProblemEngine
An answer of 401, 403 or 429 is a refusal, and it is never tried again under another name or a browser's
It follows at most 3 redirects and checks each new address the same way. It runs no scripts, signs in nowhere and reads at most 2 MB of a page

4. How to keep it out

Add these two lines to your site's robots.txt, and it reads none of your web pages:

User-agent: BulbeProblemEngine
Disallow: /

To keep it from part of the site, disallow those paths instead. To reserve pages from text and data mining by every reader, a TDMRep reservation or noai does it, and BulbeProblemEngine honors both.

The web search that names those pages is run by OpenAI, not by BulbeProblemEngine, so these lines do not reach it. A quote from a page closed to BulbeProblemEngine is never kept, because the page cannot be read to check it, but the search's link to the page, with its title, can still be listed among the pages a search consulted.

The forums on its list are read through the interface their software publishes. If you run one of them and want it left out, or want a post or a quote removed, write to support@bulbe.app with the link.

5. Contact

Operator: Trade Vessel P.C.
Based in: Santorini, Greece
Email: support@bulbe.app
Website: bulbe.app