we are trying to build an application which is accessable via onetime passwords without a "user" having the need to register.
We did came accross the term magic-links sent via email, but there is only some old experimental keycloack extention for this.
Is there any way to build some auth flow like the following with keycloack?
User A is a fully registered User creating some document. This document needs some interaction with a Third Party Person (TPP) not registered.
Now User A sends an E-Mail invite to the TPP with a link to the document. When the TPP opens the link our application should ask for the email address and send a magic link or code to this email. Whith that email or magic code the user gets access to the document for the time it takes to complete the approval process. After the work of the TTP is done, the access should expire (or expire automatically after X days not used).
It does sound quite similar to what is possible with SaaS offerings like https://magic.link/ or https://www.arengu.com/ but we are using keycloack and would like to integrate it into it as well.
Does anyone have an idea how to achieve this with keycloak?
I know this is outdated, but perhaps someone else coming along may find this useful. There is a newer implementation of this feature provided here: https://github.com/p2-inc/keycloak-magic-link
We have found it usable and useful for our needs and works well in the latest version of Keycloak (18.x).
I'm trying to develop a REST API web service. I have a question about how to handle user activation email. Currently, the API service handles email sending.
Here is the flow I have at the moment:
User registers via the client application
Client application POSTs to API service
API service validates and adds the user to the database
API service sends the User an activation link
User clicks on the activation link, which will take them to the client application activation page
Client application activation page POSTs to API service
Done
Here is where I currently see the issue:
Because the API service is currently sending the email, the client application does not have control over the look and feel of the email. And there may be URLs in the email that should point to the client application.
Another option is instead of the API service sending the activation email, it will return the activation key to the client application. The client application will then be able to send the activation email to the user.
Two issues I see with this strategy:
Security, as the activation key is now exposed to the client application.
Not DRY, as each client could be responsible for email sending.
What do you think is best way to handle this?
I would like to allow the client application to customize their email, as well as include client-specific URLs (activation page).
TL;DR
Create a small service for developers to create templates, let them declare which template they want to use when POSTing to your activation API
Summary of the problem:
e-mail needs to look different for every client app
sending mail should be implemented once
solution should be secure
There is no need for the e-mail to look different every time. So there's no need to send the e-mail format with the POST request.
Instead one of the following can be done:
1 Create a separate API endpoint to define templates and let the client app choose one of them when POSTing the request for activation.
This is not exactly secure, at least poses a challenge to make it safe if you want to accept HTML from the client apps.
Recommended solution:
2 Create a tool for developers (in the same website where they get their API key) that accepts templates and aids creating them. Client app can choose one of them when POSTing the request for activation. Fragment of the request body being something like:
...
"template": "foobar-app",
"fields": {
"title": "Welcome to foobar app",
"username": "jim78"
}
...
No HTML in the fields allowed.
This lets you have pre-defined templates prepared by the developer that can be used by your e-mail sending service and no bug in client app can cause the e-mail to become unsafe. Also, you get a place where the templates can be worked on and tested. (the developer can send them to himself to debug - making e-mail templates is horrible, belive me)
You'll be able to support your developers/clients better in the future and prepare a set of working templates tested in multiple mail clients.
A point about security and trust. Typically you send an activation email that contains a url link that has the activation code. The purpose of the email is to validate that the email is valid and that the user has access to that email. The only way the user could have received the verification link is through the email.
If you pass back the activation link to the client then anyone who has access to your API has access to the activation code. If they have access to the link they can bypass the verification process. This is really easy if you have a web app, as they just need to drop into the browser developer mode to see the link. If you have a fat client then they could snoop the network if you are not using encryption like https. They could also, if they were dedicated, decompile your binary (this is why you d not store keys in your binaries).
A backend should never trust a client to implement a security procedure because it never knows when it has been compromised. The safe and correct way is to do the activation email on the server side.
Another way to look at this, is that it is similar to the client saying "yes the user is authenticated so give me all the data"
As for the templates there are plenty of good answers above. I would suggest having a catalog of templates and a list of arguments that can be replaced.
So the way I achieved this in my opinion is quite a nice way. So I took the methodology of how JSON Web tokens work and applied it to my activation links. I'll explain how it works:
I have 2 web servers, one which handles the REST API, and one which handles the spa.
So the user registers, and the request is sent to the API. The response is then returned to the SPA at which point if successful sends a request to the SPA Backend which signs a token with the user's credentials, the purpose of the token (which is this case is to verify the email address) and it's expiry date.
This token is sent to the user's email address, however on the REST server there is a receiving route that will decode the token and if valid, verifies the email address.
This does mean that technically only 1st party clients can authenticate the email address as they are the only ones that can know your cipher secret. If your secret was freely handed out, then the problem would occur that anyone could verify their email address.
I hope this helps!
EDIT: another way would be to pass a template built in handlebars or something that swaps out variables for actual values. Then have the REST api render it, and email it. (This is probably the best way imo haha)
Your API could have an IEmailBodyFormatter object that is passed as a parameter to your API call....
I'd extend step 2 with additional post-data sent to the server:
"mail":{
"placeholder":"someStringChoosenByClientWhichWillBeReplaceByActivationCode",
"subject":"Hey there, please activate",
"ishtml":false,
"body":"SSdtIHRyeWluZyB0byBkZXZlbG9wIGEgUkVTVCBBUEkgd2ViIHNlcnZpY2UuIEkgaGF2ZSBhIHF1ZXN0aW9uIGFib3V0IGhvdyB0byBoYW5kbGUgdXNlciBhY3RpdmF0aW9uIGVtYWlsLiBDdXJyZW50bHksIHRoZSBBUEkgc2VydmljZSBoYW5kbGVzIGVtYWlsIHNlbmRpbmcu"
"attachments":[
{
"content-type":"image/png",
"filename":"inline_logo.png",
"content":"base64_data_of_image"
}
]
}
This would allow the client full control over sent message, but the activation procedure (mail generation & delivery) is still handled by the service.
Everything except the activation key can be generated for every user by the client (e.g. using "Hello XYZ" as Subject).
I'm not sure whether it's an good idea to allow html-Mails ("ishtml":false,), this depends on your application and the amount of time you want to spent implementing this.
Allow the client to manage their own email template(s). When they post a new user registration, allow them to specify which template to use. Then your application is sending the email message, but clients can control what it looks like.
POST /email-templates
{
"subject": "Complete Your Registration",
"body": "<html>Follow this link to complete your registration: {activationLink}. It is valid for 45 minutes.</html>"
}
POST /registration-requests
{
"name": "John Q. Public",
"emailTemplate": "/email-templates/45"
}
I think the proper way is to expose the activation key for the client to do whatever it wants with.
You could also add another endpoint to send the activation key for the user.
Returns user. (with the url like User/{userid} and other resources url like User/{userid}/ActivationKey)
User (POST)
This can returns the current user and other resources like Email, Activate, etc.
For info about the key (like dates, expiration, etc)
User/{userid}/ActivationKey
from there you can extend it as long as you want with :
Preview activation email:
User/{userid}/ActivationKey/Email (GET)
Update activation email with template, smtp server, etc of the email. :
User/{userid}/ActivationKey/Email (PUT)
Create (and send) activation email, possible with date to send or other send options (text-html versions, etc) :
User/{userid}/ActivationKey/Email (POST)
You could possibly list all email sent and preview them in another endpoint if necessary.
User/{userid}/Emails (GET)
User/{userid}/Emails/{emailid} (GET)
I join nauktur on the idea of letting the client send you a template of his email. (And +1 for talking about a way to test, because I agree on the awfulness of mail "development").
But why so complicated ? Client apps mean developers, so why not let them give them your default template (with HTML), let them play around if they want to, and send you back the version they prefer ?
It's not a lot of work for you (just a new field in the client table and a new route), and it gives them a lot of options.
Here is a basic example where we'll be exposing some parameters so that they can play around with the HTML without even having to know them :
app.name
app.description
activation_code
user.* registering info
Basic template
{
title: "Your activation code for %{app.name}",
body: "<p>Hi, you've been registered on %{app.name}.
<p>%{app.description}</p>
<p>Follow this link to confirm your inscription."
}
Register new template
Then the client says : "I prefer to have a more simple mail, but I want his name in it !".
[PUT] /api/email/templates/client_id
{
title: "Your activation code",
body: "<p>Hi %{user.fullname}, Follow this link to confirm your inscription."
}
And here you go. Let them play with HTML, it allows way more personalization.
There's no harm in it except for their image on their clients if they mess up, but they're their clients.
Security issues
It was pointed out that attackers could get access to the token of the client app could inject malicious content in the template. First of all, the risk is already so high if the token leaks, that this is the last of your concerns. Still, if you're scared of this, disallowing img tags and making the content of a tags match the href attribute should solve your issue.
I am building an API following RESTful principles as much as possible. The request in discussion is to allow a user to check his/her credits available in a system. At the point of request, the system verifies the user by comparing the provided username and password already in the system. Please note that changing the authentication method (to OAuth or the like) is not an option at the moment.
As this is a "Read" request, GET method is used. So, I would have the following:
GET http://mydomain.com/credit?username=XYZ&password=123
By following the RESTful principle and using the verb properly I fear that the username and password is easily readable / accessible. In a non-API scenario I would have just used a normal form POST with SSL...
Am I wrong to assume the risk mentioned above?
You are quite correct that exposing the username and password in plain text in the query string is a bad idea. Like worse than the last three Star Wars movies.
You should be fine though if the same request is made over SSL (assuming a trusted certificate).
REST also has a host of other mechanisms for security like the DOSETA specifications for digital signatures, JSON Web Signature and Encryption, and so on. But you seemed to hint those kinds of things aren't an option.
The Http Authentication header is designed to store information such as username and password. You should use that.
I'm trying to get access to the Google's todo feed with this url:
https://www.google.com/calendar/tdl?secid=<SECID>&tdl={%22action_list%22%3A[{%22action_type%22%3A%22get_all%22%2C%22action_id%22%3A%221%22%2C%22list_id%22%3A%2215052708471047222911%3A0%3A0%22%2C%22get_deleted%22%3Afalse}]%2C%22client_version%22%3A-1}
If I open this in my browser with a correct secid, it shows me right what I want.
Now, the question is: how do I get secid programmatically (specifically, in a java program)? I have access to the authToken (from CalendarService), but I have no clue how to use it to authorize my access to the URL above.
I tried to use the url http://google.com/accounts/ServiceLogin, but I didn't find any examples.
Any help, please?
From what I read secid is a session ID obtained from browser's cookies. Whereas your case uses Java which implies a server app. If that is the case, you want to drop the idea of using secid entirely.
Instead, you want to check out Google's OAuth2 documentation. If you are using Java, most likely you would be interested in the web-server OAuth flow. Pay special attention to the sequence diagrams.
The key steps include:
1) Obtain an authorization code from Google OAuth with the user's consent. For that, you redirect the user to Google with the appropriate scope. Check the list of calendar scopes for your case. Once the user consents, Google redirects back to you with an authorization code.
2) Call Google OAuth with the authorization code and your app's credentials to exchange for an access token.
3) Call Google's Calendar API using the access token.
And if you use Google's Java client as suggested by #ChaosPredictor, chances are some of the steps are already wrapped into the Java client (and your code will be much simpler).
The tool I'm developing would do a thing similar to when you join Facebook and they ask your GMail username and password and with it they can grab all your contacts.
But my tool doesn't use a browser, therefore it is difficult to get the proper headers. I assume that the Google application (Orkut) checks only my Cookies and Identity(ip, machine name, user-agent) as authentication method. In this case all I need is to get the proper cookies.
I tried to call https://www.google.com/accounts/ServiceLoginAuth?service=orkut with the parameters Passwd=realPasswordHere&Email=mymail#gmail.com . But the response was a similar HTML with Set-Cookie: GALX=A9iBuq7y5xU;Path=/accounts;Secure
None of these cookies are the real thing. Have you tried it yourself? Do you know how to do it? Have you ever seen an open source project that does it?
Consider using OAuth -- the URL I just gave details how to use OAuth to authorize the OpenSocial REST interface on Orkut, and this one has detailed specs on the RPC interface to OpenSocial, if that's what you prefer (the authorization part is basically the same, anyway).
If you are just looking for the contacts check out the provided API:
http://code.google.com/apis/contacts/
other google APIs
http://code.google.com/apis/gdata/
If you need source code to view how to do this you can check out the following project.
http://sourceforge.net/projects/gccontactman/
hope that helps, and good luck!
Perhaps you could check the source code of a Google tool that does this:
http://mail.google.com/mail/help/email_uploader.html