Wow the people in this thread are a huge bummer. This is much cooler and I doubt this is a real safety issue. You can already sign up for a free cloudflare account and deploy it for free, on your own, on a free workers.dev domain. The friction removal here isn't going to meaningfully change the security / amount of malicious content.
Well according to the people in this thread it was previously impossible for bad actors to host a website, and CloudFlare has now given them this unique ability.
I thought it was a reference to the Mac OS X `~/Public/Drop Box` directory, which was a write-only place for people to send files to your user, which has been around since the first OS X beta came out in 2000.
Cloudflare is obviously more trustworthy/robust here, but if name of the url matters to you, my site non.io [1] allows for named uploads, ie https://html.non.io/solara [2]
Somewhat useful if you want a url that isn't a hash / is more self descriptive.
There must be some really good protection on this. If I enabled such a thing on any of my servers it would be full of warez, porn, malware, CSAM and who knows what else within minutes. Curious how they manage to keep it clean.
I've never used CF so I could be ignorant in this matter. I assumed perhaps incorrectly that people had to verify their email address and delegate their domain(s) to CF including setting the glue records in the TLD servers meaning there is likely a financial trail somewhere probably in the DNS registrars and perhaps a mail provider, whereas this is just drag-and-drop with no money trail.
I have no idea what guardrails they have in place in the background that blocks malware, CSAM, warez and such on their free accounts.
I have been hosting static websites with cloudflare for years and finding how to do it on the UI is getting harder as they add more things and reoranize.
Isn't there already thousands of ways for exfiltrating data that must be whitelisted by corporate firewalls? office365/gsuite, for instance. Not to mention the classics like dns.
Your code appears to have a bug where if the arrow keys trigger a change of direction twice in a single frame interval, it can mistakenly send the snake back on itself.
This is pretty cool, thanks for sharing.
It really enables less tech savvy users. It would really enable frontpage/dreamworks-like flows for some people
Wait, my first impression was that it points a local browser to your local browser. Now it looks like it uploads your folder to Cloudflare and temporarily serves it over the web. But is that different from what we used to do with FTP? Are there any databases or anything like basic PHP hosts supply? It's just static sites?
Is this a product or what? What's the purpose? Is there an API?
A minute ago I had an HTML doc I wanted to share with a PM. It was a Claude prepared demo of a hypothetical feature. Lots of screenshots.
I ended up just embedding them directly in the HTML as base64 and sending him a 15mb file, but hypothetically this would have been a nice solution instead.
Absolutely agree. There's an insane "feature" of Claude Design which means you can only share the link to the design with other users on your account?! You can export the design, though, but then you need somewhere to quickly drop a bunch of HTML + assets. This would be perfect for that.
> Shareable Deliverables → jlnk.us (default)
The jlnk MCP server is configured machine-wide for all team agents. It publishes disposable public links: create_link(content, ttl) returns an unguessable URL anyone can open without logging in; it self-destructs after its TTL (4h/24h/72h, default 4h, max 5 MB). Also list_my_links() and delete_link(id).
> When handing a human (Founder, CEO reviewer) something to look at — QC screenshots, prototypes, reports, before/after comparisons — default to a jlnk.us link instead of a repo file path or local path. Use 72h for Founder review, shorter when the review window is same-day.
> Content must be ONE self-contained HTML file: inline CSS/JS, embed screenshots as base64 data URIs ().
> Downscale images to stay under the 5 MB cap.
> Links are public to anyone holding the URL. NEVER publish secrets, API keys, credentials, or private client data.
> Links expire — they are a viewing convenience, not the system of record. Durable artifacts still go to the repo and issue attachments as usual.
There are also solutions for sharing your homelab with others (basically tunneling from your machine->server (internet accessible) <-> client. Though, if your machine would go to sleep that whole chain would fall apart. A few good automatic solutions out there that solve the problem (no "just replace dropbox with ftp" type of argument).
However, I see the appeal of this. Kind of surprised it hasn't happened yet to be honest.
Replit is used a lot in this context. Their agent is good, but their circumvent-policies-to-get-something-in-front-of-execs-quickly is an amazing and mis-priced feature.
You could just upload to a personal or other website? I sometimes do that. Is there any security or privacy (e.g. password protection) for this Cloudflare Drop site?
Hah! This is exactly how I’m serving the vestigial remnant of my blogging in the early 2000s from a ZIP-backed Cloudflare Worker today. Should I rebuild my site with Drop+Claim or is it fine as-is? I kind of feel like ‘if what I have works, don’t change it’ is the best path.
Dropped a folder with a small HTML project, and after 20 seconds got "Something went wrong. An unexpected error occurred. Please try again or contact support.".
Note how the error has zero information.
Looking in the network tab, a POST request to /upload returned 403 and an HTML page starting with "Sorry, you have been blocked", and to "email the site owner to let them know you were blocked".
I'm very tired of this adversarial approach to software coupled with vague errors.
EDIT: it was the file './git/hooks/fsmonitor-watchman.sample' created by default on git init. Maybe because it's Perl. Worse-than-useless "please try again" and "you've been blocked" for committing the sin of uploading a folder that's a git repository. Sigh...
I tried uploading a git repository that I have previously successfully published on Github pages. This is a "no build" website I have built with the help of Claude. It should just work but I keep getting an error. Who can I reach out to give them steps to reproduce? The website repository is public and I feel like anyone at Cloudflare who wants to reproduce my problem can quite literally clone my repo and upload it to cloudflare drop.
Please drop your cloudflare email address and I will reach out to you with my repository information.
Or you could do some of your own troubleshooting? Uploading a git repo is different than uploading a zipped/folder, especially if your index.htm/l isn't at the root.
It could be fun to use a temporary Mediafire/Rapidshare/Megaupload service. Especially if you need to transfer something between an Android and an iPhone.
It would be nice if we could see some information such as file size limitations, demos, link structure, management, etc. Am I expected to upload a random HTML file and see how it works?
Yeah I'm very lost on what this is supposed to do -- "Summon your site" is quite vague. "see it live", like a demo? or is this actually published somewhere? Is it forever?
Desktop mode doesn't show any more information either
Cloudflare is really good at launching features that facility low-friction deployment of malicious content (such as phishing) on the Internet, piggybacking on their hosting reputation and the fact that you can't easily block their ASN or domains.
I don't know your experience. Once I was toying around and doing a basic auth with registration and so. The weekend was over and couldn't get back to that couple of months. The worker was quarantined and marked as phishing automatically. So I believe they have something in place to prevent those you complain.
To be fair, CF mainly develops defensive cybersecurity products, the extent to which their tools might be used maliciously is pretty on par with other regular tools.
But, it just has bad optics and potential COI/Racketeering when CF is at both sides of the counter.
To be explicit, in case it isn't obvious,Cloudflare emerged as a DDoS protection company, detecting attacks from distributed sources is part of the raison d'etre, and domains and IP addresses are a key part of that infrastructure.
By subletting their own IP addresses for navigation with warp, and their own domains for hosting of webcontent with subdomain hosting, they are providing pooled anonimity for their customers, which is precisely what makes it very hard for defenders on the other side to implement foundational security measures like IP bans, or IP block bans, or domain bans, or Whois/RDAP domain analysis.
Somewhat useful if you want a url that isn't a hash / is more self descriptive.
[1] Launch discussion: https://news.ycombinator.com/item?id=36296695
[2] This was a demo of the output of a design tool I'm working on, only the home/accommodations/about pages work.
But that won't stop people doing bad stuff for an hour I guess. Vibe code up some on-demand thing that you ping...
I have no idea what guardrails they have in place in the background that blocks malware, CSAM, warez and such on their free accounts.
I have been hosting static websites with cloudflare for years and finding how to do it on the UI is getting harder as they add more things and reoranize.
I honestly miss those days of deployment simplicity.
Also it seems to me that this is a good way to exfiltrate data, rubber stamped by cloudflare themselves.
https://drop-e7e6d363-601.important-seat.workers.dev
Is this a product or what? What's the purpose? Is there an API?
I ended up just embedding them directly in the HTML as base64 and sending him a 15mb file, but hypothetically this would have been a nice solution instead.
Here's the instructions my agents have:
> Shareable Deliverables → jlnk.us (default) The jlnk MCP server is configured machine-wide for all team agents. It publishes disposable public links: create_link(content, ttl) returns an unguessable URL anyone can open without logging in; it self-destructs after its TTL (4h/24h/72h, default 4h, max 5 MB). Also list_my_links() and delete_link(id).
> When handing a human (Founder, CEO reviewer) something to look at — QC screenshots, prototypes, reports, before/after comparisons — default to a jlnk.us link instead of a repo file path or local path. Use 72h for Founder review, shorter when the review window is same-day.
> Content must be ONE self-contained HTML file: inline CSS/JS, embed screenshots as base64 data URIs ().
> Downscale images to stay under the 5 MB cap.
> Links are public to anyone holding the URL. NEVER publish secrets, API keys, credentials, or private client data.
> Links expire — they are a viewing convenience, not the system of record. Durable artifacts still go to the repo and issue attachments as usual.
However, I see the appeal of this. Kind of surprised it hasn't happened yet to be honest.
I have a few qualms with this app.
"Something went wrong An unexpected error occurred. Please try again or contact support."
"Please upload a screenshot of the error by dragging a zip of the png file."
Note how the error has zero information.
Looking in the network tab, a POST request to /upload returned 403 and an HTML page starting with "Sorry, you have been blocked", and to "email the site owner to let them know you were blocked".
I'm very tired of this adversarial approach to software coupled with vague errors.
EDIT: it was the file './git/hooks/fsmonitor-watchman.sample' created by default on git init. Maybe because it's Perl. Worse-than-useless "please try again" and "you've been blocked" for committing the sin of uploading a folder that's a git repository. Sigh...
I tried uploading a git repository that I have previously successfully published on Github pages. This is a "no build" website I have built with the help of Claude. It should just work but I keep getting an error. Who can I reach out to give them steps to reproduce? The website repository is public and I feel like anyone at Cloudflare who wants to reproduce my problem can quite literally clone my repo and upload it to cloudflare drop.
Please drop your cloudflare email address and I will reach out to you with my repository information.
(https://x.com/BraydenWilmoth/status/2074894829616509358)
https://news.ycombinator.com/item?id=48808481
https://github.com/SpeedcubeDE/speedcube.de-forum-archive is an example use case.
Sort of, but not quite, like cherry-picking files out of an archive blob in S3.
(I’ll see if Claude and I can come up with a WARC archive->zip file converter too)
Desktop mode doesn't show any more information either
And then sell its denizens malice protection services.
Kevlar:
https://developers.cloudflare.com/bots/ https://www.cloudflare.com/products/turnstile/
Guns:
https://support.cloudflarewarp.com/
To be fair, CF mainly develops defensive cybersecurity products, the extent to which their tools might be used maliciously is pretty on par with other regular tools.
But, it just has bad optics and potential COI/Racketeering when CF is at both sides of the counter.
To be explicit, in case it isn't obvious,Cloudflare emerged as a DDoS protection company, detecting attacks from distributed sources is part of the raison d'etre, and domains and IP addresses are a key part of that infrastructure.
By subletting their own IP addresses for navigation with warp, and their own domains for hosting of webcontent with subdomain hosting, they are providing pooled anonimity for their customers, which is precisely what makes it very hard for defenders on the other side to implement foundational security measures like IP bans, or IP block bans, or domain bans, or Whois/RDAP domain analysis.