Our crawler

TurmaGrowthBot

If this address is in your server log, this page says why we were there and how to tell us not to come back.

1. Who it is

TurmaGrowthBot is the crawler of Proscenium, a platform for dance schools. It reads the public pages of a dance school's website so that a person at Proscenium can build a private preview of that school on the platform and write to the school about it — once, signed by a person, with a way to say no in the first message. Nothing it reads is published, sold or used to train anything.

2. What it reads

At most 25 pages of one site: the home page, the pages its navigation and its site map point at — timetable, classes, prices, enrolment, contacts, team, rules — and a few common addresses it asks for directly.
The timetable and price documents those pages link to — a PDF, or a picture of the timetable — and the photographs a page places beside a teacher, a style or a room. Never a private area, a login, or anything behind one; and it never fills in or submits a form.
One step off the site, and only to these hosts, when a page links there: dl.dropboxusercontent.com, docs.google.com, drive.google.com, dropbox.com, forms.gle, linktr.ee. Never a second step, and never Instagram or Facebook.
One request every 1 second per host, and at most 4 MB per file. What it read is kept, the preview is built from that copy, and the site is read again only when a person at Proscenium asks for it.
It does not run your JavaScript, submit forms, follow links off your site beyond the hosts above, or read anything robots.txt tells it not to.

3. How to stop it

Two ways, and either one is enough.

Add these lines to your robots.txt. They are honoured on the next visit, before a single page is asked for.
```
User-agent: TurmaGrowthBot
Disallow: /
```
Write to ola@turma.studio with your site's address. We add it to a list the crawler checks before it reads anything, and we delete what was read from it within 72 hours.

4. What happens to what it read

The copy of a school's pages is kept with the proposal it was read for, and is deleted with it: a preview built from it is deleted after 30 days or on request, and a school that never answers is forgotten after 12 months. The privacy notice says the rest.