The RewardVane collector

Effective October 16, 2026. If you found RewardVaneCollector in your server logs, this page is for you. It says who runs it, what it reads, how it behaves, and how to make it stop.

Who and why

RewardVane helps people use credit-card and travel rewards well. To keep our guides and our card library correct, our editors need to notice when an issuer, a loyalty programme or a regulator announces a change to a rule, a fee or an offer. The collector is the program that reads those public announcements so that a person does not have to check each site by hand.

It runs on RewardVane's own server. It never runs on a member's phone or browser, and it never signs in anywhere, with anyone's account.

Every request carries this identity:

RewardVaneCollector/1.0 (+https://rewardvane.com/bot)

What it reads

• Only sites we have admitted one by one. Each is entered in a ledger with the reason we may read it, the date we last checked its terms, and a person who is answerable for it. Nothing outside the ledger is read.
• Only public pages, only over https. No sign-in, no forms, no cookies, no paywalls, and no pages that need a script to appear.
• A feed or a sitemap first, when your site offers one. Otherwise only the pages we listed, never the whole site.

How it behaves

• It reads robots.txt first and follows the group for RewardVaneCollector. If the file cannot be read, it leaves your site alone for that visit. It honours Crawl-delay and the Content-Signal line: when you say ai-input=no, nothing from your site is given to an AI model.
• It is slow on purpose. At least two seconds between two requests to one site, and no source is visited more than once every fifteen minutes; most are read a few times a day at most.
• It asks only for what changed. It sends If-None-Match and If-Modified-Since, so an unchanged page costs you one short answer.
• It stops when you say no. An answer of 401 or 403 pauses the source until a person looks at it, and it does not try again under another name, address or browser identity. On a 429 or 503 it waits as long as you ask in Retry-After, or fifteen minutes when you do not say.
• It is bounded. A page over 1.5 MB, a feed over 2 MB, a reply slower than 15 seconds, or a fourth redirect ends the request.

What happens to what it reads

• We keep the words of a page for a limited time, at most two years and for most sources six months or less, and then wipe them.
• Models may help sort a page and propose a short note about it. Nothing is published automatically. A person at RewardVane reads each proposal against your page, writes it up in our own words, and links back to the source.
• We do not republish your text or your pictures, and nothing from your site is used to train a model.

Ask us to slow down or stop

Any of these works, and you do not need to give a reason:

• Change your robots.txt. Add a group for RewardVaneCollector with a Disallow or a Crawl-delay. We read the file again within the hour.
• Write to service@rewardvane.com with your site's name. A person reads it. We take your site off the ledger, wipe the pages we kept from it straight away rather than at the end of their term, look again at anything we wrote from them, and tell you when that is done.
• Block it. The identity above does not change. A block is treated as a no, and the collector stops.

If you think it is behaving badly (too fast, or reading something it should not), write to the same address and include the time and the path from your log. We would rather hear about it than have you guess. If your concern is something we have published rather than something we read, use the copyright notice form.