A Guide to WebAuthn
webauthn.guide
webauthn.guide
As far as I can imagine, it boils down to relying on the possession of a certain piece of hardware instead of knowing a certain password. That is, I have to carry around some kind of USB device and plug it into every computer I want to use. The device holds all my secret keys and does all the magic behind the scenes. But if I lose my device, then I can't log in. And if someone steals my device, then they can log in.
Maybe you can make it more convenient by using my phone instead of a specialized USB device, since everybody is always carrying around their phone everywhere they go anyway, right? But I just want to know that I understand the entire landscape first, from the point of view of the user, at a broad level. What are the user's options? What exactly does the user do? Is this only for smart users? If I made a web app for the general public, and my only method of authentication was WebAuthn, would the average user easily figure out what to do, or would they give up and walk away?
Some have mentioned "software-based" tokens, but then how are those kept safe? With a master password? Then we're back to passwords.
Mainly in a corporate setting, a separate hardware key may provide a root of trust (and audit trail if the key is modified to be trackable) , with which you can then unlock your devices in a self-service manner.
You're right that software authnrs are a bad idea.
For services that don't want the security to be pierced by such unsafe fallbacks, initial key attestation can whitelist the acceptable authenticators.
One thing that is too infrequently highlighted is that FIDO2 is decentral authentication between you and the services, unlike "login with big-corp".
In this scenario where it's only a second factor, yes, it relies on you possessing a Security Key, and on the bad guys not possessing it. This matches the threat model, which is typically that bad guys are not your flatmate or your mother they live in another country and will never meet you.
WebAuthn (and U2F) were specifically designed to register multiple tokens, the server-side stuff described explains how a site insists "Look I need to register a new token, not these tokens which I already registered" while also being open to whichever token you have on login, "Any of these tokens can log in, I don't mind which". So I have a token in my pocket, but if I got mugged I have a backup token at home that can substitute. Well designed sites let you nickname them during registration so you can go "I lost green hulk token but I still have Yubikey" later.
Yes, if you use the flows where both factors are handled by the FIDO token then you're "back to passwords" but with a very important difference. Very, very important:
Right now your password for news.ycombinator.com is known by the Y Combinator servers. Now, I'm sure they take good care of it, using a modern salted and pessimised hash for the purpose, but this creates many extra opportunities for vulnerabilities and a bad guy only needs to find a crack in one place.
Whereas if I have a PIN for my FIDO token the PIN lives on the FIDO token. Not on every web site I use WebAuthn with, not in a database that some poor new sysadmin accidentally uploads to the Cloud, it's not sent over the wire to some Java backend sevice that might accidentally log it, or whatever. The device can insist upon rate-limiting attempts because it's a piece of hardware, like an iPhone.
Yes, that's a good point. No security is foolproof. But this is an easy way to curb 99%
> "back to passwords" but with a very important difference
Also a great point. It may not totally get rid of passwords, but it does stop leaks.
I love FIDO tokens, but I’m not confident in my ability to keep myself from getting locked out of a service I use infrequently, and I’m a fairly technical user.
This right here is what drives me crazy about most 2fa implementations (and now apparently webauthn). I currently store 1 backup token offsite, and carry one with me. The problem is - I have no good way to access the backup device (so I can add it to my account) without having both in my possession, which destroys its abilities as a backup token.
I would love to be able to somehow have 2 fido keys use the same key, but I'm not sure how that's possible without exposing the private key externally, destroying some of the device's security properties.
One common example is a sheet of recovery codes you receive at the enrolling stage that you are prompted to print and store somewhere safely.
It's a bit archaic in a way, but works wonderfully as an accessable and relatively secure backup strategy.
https://docs.crp.to/usersguide.html#secure-encrypted-backup-...
Allowing backups is as you mentioned a tradeoff of security vs usability. The greatest risk to your accounts is not always account compromise, it can be losing access to your own accounts. With OnlyKey you can choose to enable this feature or not. Backup requires user physical presence and restore requires the backup file and correct key/passphrase.
A Google search only turns up HN posts written by you.
Here's a anecdote about phone and required 2FA: I was in Laos needing to pay rent when my phone broke. I was locked out of TransferWise to get my paycheck because their 2FA is behind their proprietary app so I couldn't authenticate on my laptop. Luckily I had just enough cash in my bank account. Too bad the bank only had 2 2FA options; 1) behind their proprietary app which doesn't allow rooted devices (despite root for more privacy tools on Android) or 2) SMS in which I was in another country so I didn't have access to my home SIM.
Because I needed a new phone to fix this issue, I went to buy a new one online. I couldn't pay for it because I needed a working phone to confirm a payment with my bank.
All of this could be solved by allowing TOTP (which doesn't require a phone) or FIDO2 (which I have a key).
If you have a hardware token with NFC, you can now test it everywhere: Mac, Windows, Linux (), Android (all browser), iOS (Safari iOS 13.3+). This is a big achievement as less than 1y ago the story was sooo much different.
If you do NOT have a hardware token, here are some options:
- Windows with TPM or Hello
- Mac with touchID sensor, you have to use Chrome
- Android phone, Chrome mobile, should work with fingerprint and face recognition
- An Android phone itself can be used as a bluetooth hardware token, with Chrome on your laptop.
Hope this list helps, please add if I forgot anything.
Which half doesn't work?
If you're referring to the caBLE (cloud-assisted Bluetooth LE) transport: this is still being worked on (https://github.com/w3c/webauthn/pull/909 but the actual discussion in the FIDO Alliance is non-public) and at the time of writing only works on Google properties.
To add: https://github.com/herrjemand/awesome-webauthn also mentions implementations completely in software. Whether those are a good idea depends on your taste, but for playing around you don't need any kind of hardware.
Why not a single CBOR object?
The answer is as far as I know:
Backwards compatibility with bad decisions made in U2F
Also; relying on base64url is odd as there are no built in functions in JavaScript to encode and decode it.
I'm probably missing something, but aren't window.atob and window.btoa just that? It's not available in Node though.
Go to https://webauthn.io on your phone and it will ask for your pin or fingerprint and then use the titan chip for the handshake! If it doesn't work. Select "Platform authenticator" in the drop down
Will also work on iOS 13 beta.
If your computer has a TPM 2.0 chip then it will also work on Microsoft Edge in Windows 10 (and maybe also other browsers. If they implement webauthb Microsoft Hello API )
Website is implemented with the excellent https://github.com/duo-labs/webauthn library by the way
EDIT: Disregard the rest of what I wrote here previously, I didn't realize it was Duo Labs's own website.
But webauthn.io and webauthn.guide are both copyright of Duo Labs and so is the library so I guess they can do whatever they want ;) I assume the two websites and the library are by the same author
Also the library lives on the server side ; not client ; and given the license is not AGPL there is no need to state that they're using it either.
(Edit: original comment was accusing me of copyright infringement for some reason)
Yeah, I didn't realize that. Tried to edit it in, but too late.
> Also the library lives on the server side ; not client ; and given the license is not AGPL there is no need to state that they're using it either.
And I didn't notice that. I thought it was JavaScript, but it was Go all along. Doubly my bad.
> (Edit: original comment was accusing me of copyright infringement for some reason)
I was (wrongly) assuming webauthn.io was in violation of the license of the library because I thought it was a client-side library.
That is up to the implementor.
The implementor can ask the browser for certain security features of the authentication device. E.g. is the authentication device the same device as where the authentication flow is happening; is there a biometric check on the device or a pin on the device (i.e. the second factor is there but device-local) who is the manufacturer of the device (with consent of the user; given this is privacy-sensirive info) etc. And the implementor can then make a decision whether a device is 'strong' enough for single factor auth. All this information is cryptographically attested by the device.
You can ignore all that and only use it as a second factor always though. That's totally up to you and how you use the authentication primitives that webauthn provides.
> Currently, the WebAuthn second-factor use case (the FIDO U2F user experience) is the only log in flow that is supported. Security key-based biometrics or PIN (without the use of username and password) are not supported yet.
https://www.yubico.com/2019/12/native-support-for-webauthn-a...
I enter a username, press "Register" and Firefox gives me a prompt to "login [with a security key] and authorize", with only a "Cancel" button. And that's all that happens.
Wait, so I can only use this stuff with a hardware token ? Bummer !
security.webauth.webauthn_enable_softtoken=true
security.webauth.webauthn_enable_usbtoken=false
https://pypi.org/project/django-webauthin/
It's live on some of my side projects, if you want to try it out:
19 points?! I haven’t dug too deep into the details, but so far everything I’ve seen re: implementing WebAuthn seems so... over complicated? Compared to implementing, say, TOTP based 2FA, which is very straightforward.
The nice thing about software is that it can be copied, so the complicated part only needs to be done once by somebody suitably skilled in the art.
I assume that if you only connect one computer (authenticator) and lose the device, you're either SOL or the service has some workaround where you pre-register an email address or a separate "Login With" service.
Eg your password manager could implement this so all you'd have to do is click "log in" and you'd be in, without usernames or passwords to remember or steal.
The big win is that, with WebAuthn, you don't need to also hide your authentication from site operators, your OS, key loggers, phishers, etc etc.
I've written an implementation of WebAuthn for Python-based webapps: https://github.com/pyauth/pywarp that may be useful to implementers.
I've always thought myself very able to explain technical things quite well to non-technical people, but always struggled to explain the public/private key exchange in a simple way. This is really neat.
If you're interested in something similar that actually works and is very simple, take a look at https://github.com/btcontract/lnurl-rfc/blob/master/spec.md#...
It would have to be something like;
1) Login on primary device
2) Generate a one-time code or link
3) Enter one-time code or click link on 2nd device
4) Click an “Enroll” button on the 2nd device which would generate a second enrollment saved in the account.
You would have to do this for each and every service so that seems like a total non-starter.
Alternatively, a way to share the private keys securely between the devices. But I’m not sure if there are site-specific salts as part of the WebAuthN signing process which would require additional state be sync’d between the devices in order to authenticate to a new service on both devices?
This seems like a huge blocker for adoption. I currently run into this issue with FIDO for 2fa - I store my backup key offsite, which means enrolling the second device requires me to make a special trip to retrieve the device. My current approach is to retrieve the backup token every few months and add it to all of the new services I have enrolled in, but I don't have a good system for remembering all of these services, so I inevitably forget one (despite only using the key on a few services)...
Yes, you'd have to use some other way to authenticate as yourself on a second or subsequent platform device. If you have a separate FIDO token (like a Yubikey or any of their cheaper Security Key products or dozens of others) that would be the obvious first choice to enroll.
WebAuthn seems to be designed by big businesses to take control of the authentication ecosystem.
Can you expand on that?
The whole spec is designed with hardware tokens in mind, software tokens are not mentioned anywhere, and WebAuthn spec designers wave their hands when asked about software tokens.
I can't find anything like that, can you point us to the docs/examples?
OTOH, I see that someone right now posted a reply to https://github.com/w3c/webauthn/issues/1175#issuecomment-570... trying to downplay the issue of purely software tokens
Funnily, https://github.com/herrjemand/awesome-webauthn#software-auth... is not about software tokens but about webauthning with an Android phone or a Wear OS.
https://news.ycombinator.com/item?id=21963399
Which comment is trying to downplay the issue? What is the issue?
FIDO2 starts with the idea of safe defaults, where either client devices (Android, laptop TPM,...) store the keys safely, or dongle vendors (like us, SoloKeys). These have a business interest in doing their job properly.
But there's nothing preventing software implementations, it's an open standard in that respect (I do have other issues with it but your specific concern is unfounded imho).
This is a fundamental difference.
"But attestation is not spoofable. Therefore, if sites launch webauthn support and accept attestations from the current set of token vendors, future vendors may be locked out of the market: Their devices won’t work because their attestations aren’t trusted, and they won’t be able to get sites to update because they won’t have enough market presence to matter."
Similarly, if China mandates a government-approved (backdoored?) device when accessing Chinese websites, I don't suppose Chinese citizens are going to boycott those. Other countries might reasonably retaliate by forbidding sites to accept authentication by devices made in China.
Do you think that if Apple decreed that you can only securely log in to Apple websites with an iPhone (or some other device made by Apple) people would switch to Android? It's hard enough to get people to boycott Facebook despite repeated scandals that are much worse than having a whitelist of supported authentication devices.
It is rational to not give your adversaries the tools needed to oppress you.
If my bank wanted to mandate the use of Yubikeys, it would do that, WebAuthn or no. It would sell me the Yubikey, as it now sells me its RSA token. I don't think the protocol the token speaks matters.
I accept your point about the protocol being irrelevant, however. One could even say that, if services are going to require specific pieces of hardware, it makes sense that the requirements are communicated and enforced using an open standard protocol, as that does allow a certain degree of interoperability and Free Software implementation.
The counter-argument, though, is that currently there are barriers to sites implementing such policies (due to the cost of issuing these devices and linking them to a specific account/address). If we set the precedent that every site should be using this technology, and reduce the cost of doing so, we bring about a set of dynamics where individual sites can start to introduce incompatibilities, whether with good intentions, or for anti-competitive reasons, or by accident.
Maybe this won't lead to people needing to carry around five dongles and two phones with them, but it could easily lead to having a lot of power concentrated into the hands of a small number of entities, or even just a single one, like the bad old days of "This site works best in Internet Explorer". Moreover, this is not just a theoretical concern, as it has already started to happen, as the link above explained:
"FIDO does not dismiss these worries and their answer, for the moment, is the metadata service (MDS). Essentially this is a unified root store that all sites checking attestation are supposed to use and update from."
I sense a bit of 90s security thinking from your arguments though, where every end user and mid-level admin handles security decisions they're frankly not qualified for.
This is what I meant by "safe defaults". Yes some people use e.g. password managers, but no, most people don't. Yes, some people manage to use GPG to manage their ssh keys, but no most people, even qualified, don't/can't/won't.
"Bad defaults with patches hopefully making it safe" is just not the way we should be heading.
It's better for the society to let them make their share of mistakes. In the longer term, everyone will be safer and, incidentially, more intelligent.
OnlyKey is also open source here are some of the features: - On device PIN - FIDO2 (15 Resident keys) - TOTP (24 accounts) - Static passwords (24 passwords up to 56 char long) - OpenPGP
Not necessarily: the WebAuthn spec mentions two other types of authenticators in the introduction section (https://www.w3.org/TR/webauthn-1/#intro): "Broadly, compliant authenticators protect public key credentials, and interact with user agents to implement the Web Authentication API. Implementing compliant authenticators is possible in software executing (a) on a general-purpose computing device, (b) on an on-device Secure Execution Environment, Trusted Platform Module (TPM), or a Secure Element (SE), or (c) off device."
> There does not seem to be a way to have privately generated software keys
A specification is something different than an implementation. On https://github.com/herrjemand/awesome-webauthn you'll find (at the time of writing) two software implementations. https://krypt.co/ can be a third if you want to consider a U2F implementation as well.
Perhaps https://github.com/bodik/soft-webauthn is closer to what you're looking for.
The standard also define the concept of resident keys (RKs), that you can use for passwordless authentication (typically with a PIN on the token, to avoid theft). In this case of course each key consumes a bit of storage, so the number of sites is limited. For example, a Solo key can currently store 50 resident keys. To the best of my knowledge, apart from demo sites, the only "real" site supporting resident keys is microsoft/outlook.com, so this is currently not a real practical limitation.
I recently (last week) wrote a Django library for WebAuthn (https://pypi.org/project/django-webauthin/) and use it on a few of my sites (https://www.pastery.net, https://www.eternum.io and https://www.deadmansswitch.net, if you want to try it out).
https://www.agwa.name/blog/post/always_review_your_dependenc...
Claims there are it's written by people who don't understand Go that well.
I guess your public key could get associated with your identity. But the pub key is only sent when you register, I believe. So I can't see how this would be used for tracking.