That would prolly work for simple sites, but you still need the dedicated scraping service with a browser to render sites that are more complex (i.e. SPAs)

Well, the conversion process into the JSON representation is going to take CPU, and then you have to store the result, in essence doubling your cache footprint.

Doing it on demand still utilizes their cached version, so it saves a trip to the origin, but doesn’t require doubling the cache size. They can still cache the results if the same site is scraped multiple times, but this saves having to cache things that are never going to be requested.

Cache footprint management is a huge factor in the cost and performance for a CDN, you want to get the most out of your storage and you want to serve as many pages from cache as possible.

I know in my experience working for a CDN, we were doing all sorts of things to try to maximize the hit rate for our cache.. in fact, one of the easiest and most effective techniques for increasing cache hit rate is to do the OPPOSITE of what you are suggesting; instead of pre-caching content, you do ‘second hit caching’, where you only store a copy in the cache if a piece of content is requested a second time. The idea is that a lot of content is requested only once by one user, and then never again, so it is a waste to store it in the cache. If you wait until it is requested a second time before you cache it, you avoid those single use pages going into your cache, and don’t hurt overall performance that much, because the content that is most useful to cache is requested a lot, and you only have to make one extra origin request.

Not the same thing, but they have something close (it's not on-by-default, yet) [1]:

> Cloudflare's network now supports real-time content conversion at the source, for enabled zones using content negotiation headers. Now when AI systems request pages from any website that uses Cloudflare and has Markdown for Agents enabled, they can express the preference for text/markdown in the request. Our network will automatically and efficiently convert the HTML to markdown, when possible, on the fly.

[1] https://blog.cloudflare.com/markdown-for-agents/

Offering wholesale cache dumps blows up every assumption about origin privacy and copyright. Suddenly you are one toggle away from someone else automatically harvesting and reselling your work with Cloudflare as the unwitting middle tier.

You could try to gate this behind access controls but at that point you have reinvented a clunky bespoke CDN API that no site owner asked for, plus a fresh legal mess. Static file caches work because they only ever respond to the original request, not because they claim to own or index your content.

It is a short path from "helpful pre-scraped JSON" to handing an entire site to an AI scraper-for-hire with zero friction. The incentives do not line up unless you think every domain on Cloudflare wants their content wholesale exported by default.

> I'm surprised that Cloudflare hasn't started hosting a pre-scraped version of websites that use Cloudflare's proxy

It's entirely possible that they're doing this under the hood for cases where they can clearly identify the content they have cached is public.

That was my first thought when I read the headline. It would make perfect sense, and would allow some websites to have best of both worlds: broadcasting content without being crushed by bots. (Not all sites want to broadcast, but many do).

But think about poor phishers and malware devs protected by Cloudflare.

In front: https://developers.cloudflare.com/browser-rendering/rest-api...

I think Common Crawl already offers this, although it's free: https://commoncrawl.org/

> Doing it on demand still utilizes their cached version, so it saves a trip to the origin, but doesn’t require doubling the cache size. They can still cache the results if the same site is scraped multiple times, but this saves having to cache things that are never going to be requested.

Isn't this solving a slightly, but very significantly different problem?

You could serve the very same data in two different ways: One to present to the users and one to hand over to scrapers. Of course, some sites would be too difficult or costly to transform into a common underlying cache format, but people who WANT their sides accessible to scrapers could easily help the process along a bit or serve their site in the necessary format in the first place.

But the key is:

A tool using a "pre-scraped" version of a site has very likely very different requirements of how a CDN caches this site. And this could be easily customizable by those using this endpoint.

Want a free version? Ok, give us the list of all the sites you want, then come back in 10min and grab everything in one go, the data will be kept ready for 60s. Got an API token? 10 free near-real-time request for you and they'll recharge at a rate of 2 per hour. Want to play nice? Ask the CDN to have the requested content ready in 3 hours. Got deep pockets? Pay for just as many real-real-time requests as you need.

What makes this so different is that unless customers are willing to hand over a lot of money, you dont need to cache anything to serve requests at all. Potentially not even later if you got enough capacity to serve the data for scheduled requests from the storage network directly.

You just generate an immediate promise response to the request telling them to come back later. And depending on what you put into that promise, you've got quite a lot of control over the schedule yourself.

- Got a "within 10min" request but your storage network has plenty if capacity in 30s? Just tell them to come back in 30s.

- A customer is pushing new data into your network around 10am and many bots are interested in getting their hands on it as soon as possible, making requests for 10am to 10:05? Just bundle their requests.

- Expected data still not around at 10:05? Unless the bots set an "immediate" flag (or whatever) indicating that they want whatever state the site is in right now, just reply with a second promise when they come back. And a third if necessary... and so on.

Interesting - its sounds like this could be combined with some creative cache parsing on their side to provide this feature to sites that want it.

Based on the post, it seems likely that they'd just delay per the robots.txt policy no matter what, and do a full browser render of the cached page to get the content. Probably overkill for lots and lots of sites. An HTML fetch + readability is really cheap.

It’s a bit more complicated than that. This is their product Browser Rendering, which runs a real browser that loads the page and executes JavaScript. It’s a bit more involved than a simple curl scraping.

No: https://developers.cloudflare.com/browser-rendering/rest-api...

How would they know the content hasn’t changed without hitting the website?

They always have been.

They also use their dominant position to apply political pressure when they don’t like how a country chooses to run things.

So yeah, we’ve created another mega corp monster that will hurt for years to come.

Well this scraper honours robots.txt so I'm sure most AI crawlers will find it useless.

I know in practice it no longer is the case, if it ever was.

But semantic HTML is exactly that explicit machine-readable entrypoint. I am firmly entrenched in the opinion that HTML, and the DOM is only for machines to read, it just happens to be also somewhat understandable to some humans. Take an average webpage, have a look at all characters(bytes) in there: often two third won't ever be shown to humans.

Point being: we don't need to invent something new. We just need to realize we already have it and use it correctly. Other than this requiring better understanding of web tech, it has no downsides. The low hanging fruit being the frameworks out there that should really do a better job of leveraging semantics in their output.

> It also raises interesting questions about whether sites will eventually provide different views for humans vs. automated agents in a more formalized way.

This question raises an interesting question about if this would exacerbate supply chain injection attacks. Show the innocuous page to the human, another to the bot.

I think there's some space being absolutely snuffed by the countless bots of everyone, ignoring everything, pulling from residential proxies, and this, supposedly slower, well behavior, smarter bot.

Like there's a difference between dozens of drunk teenagers thrashing the city streets in the illegal street race vs a taxi driver.

I expect that if we still used REST indexing would be even less wasteful.

I've found myself falling pretty hard on the side of making APIs work for humans and expecting LLM providers to optimize around that. I don't need an MCP for a CLI tool, for example, I just need a good man page or `--help` documentation.

Apart from the obvious problem: presenting something different to crawlers and humans.

I just do a query param to toggle to markdown/text if ?llm=true on a route. Easy pattern that's opt-in.

What % of sites have a content update volume that exceeds what you can get respecting crawl delay?

If your delay is 1s and you publish less than 60 updates a minute on average I can still get 100%. Most crawls are not that latency sensitive, certainly not the ai ones.

HFT bots, now that is an entirely different ballgame.

I think the simple explanation is that they weren't selling scraping countermeasures, they were selling web-based denial of service protection (which may be caused by scrapers).

No: https://developers.cloudflare.com/browser-rendering/rest-api...

Cloudflare has been trying to mediate publishers & AI companies. If publishers are behind Cloudflare and Cloudflare's bot detection stops scrapers at the request of publishers, the publishers can allow their data to be scraped (via this end point) for a price. It creates market scarcity. I don't believe the target audience is you and me. Unless you own a very popular blog that AI companies would pay you for.

Their free DNS is only a small piece of the pie.

The fact that 30%+ of the web relies on their caching services, routablility services and DDoS protection services is the main pull.

Their DNS is only really for data collection and to front as "good will"

Was it ever not one? They protect a lot of DDoS-for-hire sites from DDoS by their competitors. In return they increase the quantity of DDoS on the internet. They offer you a service for $150, then months later suddenly demand $150k in 24 hours or they shut down your business. If you use them as a DNS registrar they will hold your domain hostage.

no? it takes 10 seconds to check:

> The /crawl endpoint respects the directives of robots.txt files, including crawl-delay. All URLs that /crawl is directed not to crawl are listed in the response with "status": "disallowed".

You don't need any scraping countermeasures for crawlers like those.

The only ones benefitting from 'wastefull' crawling are the anti-bot solution vendors. Everyone else is incentivized to crawl as efficiently as possible.

Makes you think, right?

Isn't it already covered by sitemaps and sitemap index files, which are machine readable XML?

I have tried theirs... they are NOT proxies.. that means majority of the popular sites actually block scraping... even if they are protected by cloudflare itself.

The most egregious thing Perplexity did was to straight up ignore robots.txt. Cloudflare promise not to do that, so if we take their word for it, it's a quite different setup.

That said, I'm not fan of letting users forge whatever user agents they please. Instead, AIUI to opt-out of getting crawled I have to look for the existence of certain request headers[1].

[1]: https://developers.cloudflare.com/browser-rendering/referenc...

LLM generated comment

I’ve actually written a crawler like that before, and still ended up going with Firecrawl for a more recent project. There’s just so many headaches at scale: OOMs from heavy pages, proxies for sites that block cloud IPs, handling nested iframes, etc.

This presumably is going to be cheap and effective. Its much easier to wrap a prompt round this and know it works that mess around with crawling it all yourself.

You'll still be hand-rolling it if you want to disrespect crawling requirements though.

They say they obey robots.txt - isn’t that the easier way?

That'd be more like that draw an owl meme. Devil's in the details. Holy shit, there's so many details...

You put a governor on the domain, and you return from the cache instead.

> Browser Rendering is only available on the Workers Paid plan ($5/month). It is not part of the free tier.

The post says it's available for both free and paid plans. According to the pricing page of the Browser Rendering, the free plan will have 10 minutes/day browsing time.

[0] seems to suggest even paid plans are effectively limited to 500 web pages per day, right?

    Crawl jobs per day 5 per day
    Maximum pages per crawl 100 pages

[0] https://developers.cloudflare.com/browser-rendering/limits/#...

They say it doesn't: https://developers.cloudflare.com/browser-rendering/faq/#wil...

Further down they also mention that the requests come from CFs ASN and are branded with identifying headers, so third party filters could easily block them too if they're so inclined. Seems reasonable enough.

Yeah, that'd be huge, like 90% of my search engine results are just cloudflare bot checks if I don't filter it out.

So does that mean it can replace serpapi or similar?

That is funny because on this page there is a warning block with the following text:

   Refer to Will Browser Rendering bypass Cloudflare's Bot Protection? for instructions on creating a WAF skip rule.

And "Will Browser Rendering bypass Cloudflare's Bot Protection? " is a hash link to the FAQ page, that surprisingly doesn't anything available for this link entry.

Is it because it was removed (/hidden) or because it is not yet available until everyone forget the "we are no evil, we are here to protect the internet"?

They wouldn't, well there's Etag and alike but it still a round trip on level 7 to the origin. However the pattern generally is to say when the content is good to in the Response headers, and cache on that duration, for an example a bitcoin pricing aggregator might say good for 60 seconds (with disclaimers on page that this isn't market data), whilst My Little Town news might say that an article is good for an hour (to allow Updates) and the homepage is good for 5 minutes to allow breaking news article to not appear too far behind.

Keeping track of when content changes is literally the primary function of a CDN.

Caching headers?

(Which, on Akamai, are by default ignored!)

I don’t really get the usecase. Is your site static? Then you should just render it to html files and host the static files. And if it’s not static, how would a snapshot of the pages help if they change later? And also why not just add some caching to the site then?

They already do...

A lot of known crawlers will get a crawler-optimized version of the page

Do you think cloudflare is responsible for all of the network traffic routing in the entire world and can simply fix any problem even if it's on somebody else's network?

It is possible that Claro has a bad route that sends all traffic destined for Cloudflare through New York.

If this does bypass their own (and others') anti-AI crawl measures, it'd basically mean that the only people who can't crawl are those without money.

We're creating an internet that is becoming self-reinforcing for those who already have power and harder for anyone else. As crawling becomes difficult and expensive, only those with previously collected datasets get to play. I certainly understand individual sites wanting to limit access, but it seems unlikely that they're limiting access to the big players - and maybe even helping them since others won't be able to compete as well.

"I didn't want to scrape/crawl it myself because I felt like it would feel like yet another scraping effort for AI and strain resources of developers"

You feel better paying someone to do the same thimg?

This was always also sold as bot protection and anti-scraping / crawling features like https://www.cloudflare.com/lp/pg-ai-crawl-control/

That's not the perfect defense you think it is. Plenty of robots.txts[1] technically allow scraping their main content pages as long as your user-agent isn't explicitly disallowed, but in practice they're behind Cloudflare so they still throw up Cloudflare bot check if you actually attempt to crawl.

And forget about crawling. If you have a less reputable IP (basically every IP in third world countries are less reputable, for instance), you can be CAPTCHA'ed to no end by Cloudflare even as a human user, on the default setting, so plenty of site owners with more reputable home/office IPs don't even know what they subject a subset of their users to.

[1] E.g. https://www.wired.com/robots.txt to pick an example high up on HN front page.

Next step will be their default "free" anti-bot denying all but their own bot. They know full well nearly nobody changes the default.

> Most crawls are not that latency sensitive, certainly not the ai ones.

They certainly behave like they are. We constantly see crawlers trying to do cache busting, for pages that hasn't change in days, if not weeks. It's hard to tell where the bots are coming from theses days, as most have taken to just lie and say that they are Chrome.

I'd agree that the respecting robots.txt makes this a non-starter for the problematic scrapers. These are bots that that will hammer a site into the ground, they don't respect robots.txt, especially if it tells them to go away.

All of this would be much less of a problem if the authors of the scrapers actually knew how to code, understood how the Internet works and had just the slightest bit of respect for others, but they don't so now all scrapers are labeled as hostile, meaning that only the very largest companies, like Google, get special access.

Ask yourself, why would a scraper ddos? Why would a ddos-protection vendor ddos?

> The fact that 30%+ of the web relies on their caching services

30% of the web might use their caching services. 'Relies on' implies that it wouldn't work without them, which I doubt is the case.

It might be the case for the biggest 1% of that 30%. But not the whole lot.

I feel there is a conflict of interest here..

I'm split between: Yes! At last something to get CF protected sites! And: Uh! Now the internet is successfully centralized.

I yearn for the days when a single kb get was enough. Now it's endless wastage spawning entire browsers larger than operating systems with mitigations, hacks and proxies. Requesting access directly from webmasters is only met with silence. All of my once simple, hobbyist programs are now bloated beyond belief and less reliable than ever

Like they explain in the docs, their crawler will respect the robots.txt dissalowed user-agents, right after the section hat explains how to change your user-agent.

yeah, GP completely fails to realize that Cloudflare has always played both sides. that is their entire business model, and it was transparent from the beginning that they would absolutely do the same here.

Where can I learn more about the 150k in 24h?

So what’s the user agent for their bot? They don’t seem to specify the default in the docs and it looks like it’s user configurable. So yet another opt out bot which you need your web server to match on special behaviour to block

Ah the use-case is archive.org but fast. But it's okay. Before I die I will make the static copy of my site myself.

No. I do think that Cloudflare is a great company and got where it's at today because they care for this type of issue, and has a much better chance of contacting their peering traffic partner than me because they take care of ~20% of all internet traffic, while I take care of none.

Every once and a while we have had Bell Canada route a request that should be going about 6 blocks away across the continent and back.

They are not super helpful fixing it either.

Common Crawl has free egress

Do they? AFAIK Google forbids that, and they’ll occasionally test that you aren’t doing it.

most websites, particularly those behind cloudflare, are very restrictive even to crawlers that obey robots. Proof: a ton of my time over the last year, and my crawlers very carefully obey robots.

It's hard to see how this isn't extorting folks by offering a working solution that, oh, cloudflare doesn't block. As long as you pay Cloudflare.

Perhaps I'm overly cynical, but I'd be quite surprised if cloudflare subjected their own headless browsing to the same rules the rest of the internet gets.

I actually don't but it seems that cloudflare caches responses so if anything instead of straining the developer resources, it would strain more cloudflare resources and cloudflare could better handle that more efficiently with their own crawl product.

Also, I am genuinely open to feedback (Like a lot) so just let me know if you know of any other alternative too for the particular thing that I wish to create and I would love to have a discussion about that too! I genuinely wish that there can be other ways and part of the reason why I wrote that comment was wishing that someone who manages forums or knows people who do can comment back and we can have a discussion/something-meaningful!

I am also happy with you also suggesting me any good use cases of the domain in general if there can be made anything useful with it. In fact, I am happy with transferring this domain to you if this is something which is useful to ya or anyone here (Just donate some money preferably 50-100$ to any great charity in date after this comment is made and mail me details and I am absolutely willing to transfer the domain, or if you work in any charity currently and if it could help the charity in any meaningful manner!)

I had actually asked archive team if I could donate the domain to them if it would help archive.org in any meaningful way and they essentially politely declined.

I just bought this domain because someone on HN said mirror.org when they wanted to show someone else mirror and saw the price of the .org domain being so high (150k$ or similar)and I have habit of finding random nice TLD and I found mirror.forum so I bought it

And I was just thinking of hmm what can be a decent idea now that I have bought it and had thought of that. Obviously I have my flaws (many actually) but I genuinely don't wish any harm to anybody especially those people who are passionate about running independent forums in this centralized-web. I'd rather have this domain be expired if its activation meant harm to anybody.

looking forward to discussion with ya.

Because the scraper is either impatient, careless or indifferent; and if they scrape for training data they don't plan to come back. If they don't plan to come back they don't care if you tighten up crawling protections after they have moved on. In fact they are probably happy that they got their data and their competition won't

Ask yourself, why would everyone except you say that they do?

The number of git forges behind Anubis et al and the numerous public announcements should be enough.

Scrappers seem to be exceedingly careless in using public resources. The problem is often not even DDOS (as in overwhelming bandwidth usage) but rather DOS through excessive hits on expensive routes.

> We constantly see crawlers trying to do cache busting

Do you have a source for this? Not saying you're wrong, I'd just like to know more

> Ask yourself, why would a scraper ddos?

Don't need to ask anything i can tell you exactly - because they have no regard for anything but their own profit.

Let me give you an example of this mom and pop shop known as anthropic.

You see they have this thing called claudebot and at least initially it scraped iterating through IP's.

Now you have these things called shared hosting servers, typically running 1000-10000 domains of actual low volume websites on 1-50 or so IPs.

Guess what happens when it is your networks time to bend over? Whole hosting company infrastructure going down as each server has hundreds of claudebots crawling hundreds of vhosts at the same time.

This happened for months. Its the reason they are banned in WAFs by half the hosting industry.

>'Relies on' implies that it wouldn't work without them

Last time Cloudflare went down, their dashboard was also unavailable, so you couldn't turn off their proxy service anyway.

I imagine it's referencing this story:

https://robindev.substack.com/p/cloudflare-took-down-our-web...

HN Discussion:

https://news.ycombinator.com/item?id=40481808

Isn't this covered here? https://developers.cloudflare.com/browser-rendering/referenc...

>So yet another opt out bot which you need your web server to match on special behaviour to block

Given that malicious bots are allegedly spoofing real user agents, "another user agent you have to add to your list" seems like the least of your problems.

I haven't checked in a while but I know for a fact that Amazon does or did it

With google covering only 3% I wonder how much people still care and if they should. Funny: I own and know sites that are by far the best resource on the topic but shouldn't have so many links google says. It's like I ask you for a page about cuban chains then you say you don't have it because they had to many links. Or your greengrocer suddenly doesn't have apples because his supplier now offers more than 5 different kinds so he will never buy there again.

>most websites, particularly those behind cloudflare, are very restrictive even to crawlers that obey robots. Proof: a ton of my time over the last year, and my crawlers very carefully obey robots.

The docs are pretty equivocal though:

>If you use Cloudflare products that control or restrict bot traffic such as Bot Management, Web Application Firewall (WAF), or Turnstile, the same rules will apply to the Browser Rendering crawler.

It's not just robots.txt. Most (all?) restrictions that apply to outside bots apply to cloudflare's bot as well, at least that's what they're claiming. If they're being this explicit about it, I'm willing to give them the benefit of the doubt until there's evidence to the contrary, rather than being a cynic and assuming the worst.

This is used to scrape third-party sites not necessarily behind cloudflare so it has nothing to do with whether cloudflare caches it or not plus when using their browser rendering it doesn't even fetch cached responses anyways....

> they don't plan to come back

To me the current behavior of those scrapers tells me that "they don't plan", period.

Looks like they hired a bunch of excavators and are digging 2 meters deep on whole fields, looking for nuggets of gold, and pilling the dirt on a huge mountain.

Once they realize the field was bereft of any gold but full of silver? Or that the gold was actually 2.5 meters deep?

They have to go through everything again.

So how would you avoid this specific situation as a web-crawler that tries to be well behaved? You strictly adhere to robots.txt as specified by each domain. The problem is not with any of the sites but the density (1000-10000) by which the hoster packed them. If e.g. the crawler had a 1 sec between page governor even if robots.txt had no rate specified, which to be fair is very reasonable, this packing could still lead to high server load.

Not really, given that the work we do in that direction isn't exactly public. You can recreate the scenario though. Spin up a wiki of some sort, scrapers love wikis, ideally enable some form of caching, and just sit back and watch scrapers throw random shit in the URL parameters.

No, hence all their examples using User-Agent: *

Not 'allegedly' - it's just a fact. Even if you're not malicious however it's still sometimes necessary because the server may have different sites for different browsers and check user agents for the experience they deliver. So then even for legitimate purposes you need to at least use the prefix of the user agent that the server expects.

It is cloudflare who made the claim that they are well behaved unlike those other bots and that their behaviour can be controlled by robots.txt

If I need to treat cloudflare bots the same as malicious bots, that undermines their claim.

I didn't know that it doesn't fetch catched responses, my apologies. I had only read through it with a glance and it felt like something that cloudflare might've done. Is there any particular reason that they don't use the cached responses, feels like a missed opportunity but maybe I am missing something?

It's a browser rendering API which means people are paying a premium specifically to have a browser render a live website. If you want to get a cached response of a page and still possibly get blocked by cloudflare you could just make a node script with a simple fetch and save your money.