Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in
this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten their
password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind
of thing.
That makes be nervous.
Thanks for any feedback.
In general, there are four things to be aware of:
Credentials encoded in the URL are very easy to accidentally leak
in any outgoing Referer headers, in screenshots, etc. This can be
mitigated, but it's also easy to mess things up.
Credentials encoded in the URL will by default end up in the logs
of the HTTP server - although again, this can be fixed. They may also
get recorded by various proxies, AV tools, browser toolbars, etc -
which is a bit harder to contain.
The user will be dependent on having access to bookmarks or some
other way to retrieve that URL to be able to log in from a particular
computer, since it's impossible to remember; and if his bookmarks are
deleted or corrupted, you will need to deal with account recovery. If
you are willing to put up with this, you may also want to consider
client SSL certificates or so as an alternative form of
authentication.
Since all the information needed to access the account is stored on
the computer, any system compromise can be used to recover the access
token immediately, rather than waiting for the user to interact with
the targeted site. This may or may not matter.
/mz
Glen,
Check out this paper that advocates the use of such identifiers instead of
session cookies and other ambient credentials:
Web-key: Mashing with permission
http://w2spconf.com/2008/papers/s4p2.pdf
It addresses the 4 things that Michael listed and
it's generally a very pleasant read
Cheers
Nick
On Mon, Jan 28, 2013 at 6:55 PM, Michal Zalewski lcamtuf@coredump.cxwrote:
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
In general, there are four things to be aware of:
Credentials encoded in the URL are very easy to accidentally leak
in any outgoing Referer headers, in screenshots, etc. This can be
mitigated, but it's also easy to mess things up.
Credentials encoded in the URL will by default end up in the logs
of the HTTP server - although again, this can be fixed. They may also
get recorded by various proxies, AV tools, browser toolbars, etc -
which is a bit harder to contain.
The user will be dependent on having access to bookmarks or some
other way to retrieve that URL to be able to log in from a particular
computer, since it's impossible to remember; and if his bookmarks are
deleted or corrupted, you will need to deal with account recovery. If
you are willing to put up with this, you may also want to consider
client SSL certificates or so as an alternative form of
authentication.
Since all the information needed to access the account is stored on
the computer, any system compromise can be used to recover the access
token immediately, rather than waiting for the user to interact with
the targeted site. This may or may not matter.
/mz
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
Glenn, it's considered best practice not to send auth tokens in URL due to risks of upstream logging , also proxies and firewalls may log it and then making it permanent would only increase the risk.
Thanks
Subin
Sent from my iPhone
On Jan 28, 2013, at 4:13 AM, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
Well said but kick it up a notch. I actually recommend keeping all
sensitive data out of URLs for just this reason. Passwords, PII,
Authentication tokens - anything you do not want leaking should stay out
of URL's (HTTP/S GETS).
The more positive rule is to only submit sensitive data over HTTPS POST.
There are of course additional sub rules in the age of AJAX and JSON,
but this is a good start.
Cool?
Jim Manico
@Manicode
Glenn, it's considered best practice not to send auth tokens in URL due to risks of upstream logging , also proxies and firewalls may log it and then making it permanent would only increase the risk.
Thanks
Subin
Sent from my iPhone
On Jan 28, 2013, at 4:13 AM, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
Thanks for the good ideas everyone. I have a few things to think about.
When accessing through https what will upstream proxies log ? Just
the encrypted url right ?
On 28 January 2013 09:13, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in
this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten
their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind
of thing.
That makes be nervous.
Thanks for any feedback.
Depends on the proxy and its functions. Many proxies now monitor SSL
traffic by acting as a go between.
On Jan 28, 2013 4:16 PM, "Glenn Pierce" glennpierce@gmail.com wrote:
Thanks for the good ideas everyone. I have a few things to think about.
When accessing through https what will upstream proxies log ? Just
the encrypted url right ?
On 28 January 2013 09:13, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login
in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten
their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this
kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
yep, after the domain name its all encrypted via http(s). but it would still be in the clear in the web server logs.
-phil
On Jan 28, 2013, at 3:07 PM, Glenn Pierce glennpierce@gmail.com wrote:
Thanks for the good ideas everyone. I have a few things to think about.
When accessing through https what will upstream proxies log ? Just the encrypted url right ?
On 28 January 2013 09:13, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
Upstream proxies only have visibility over the network information
(hostname, IP and port, etc.) for HTTPS connections. That is, if they're
not acting as MITM proxies on purpose or not.
Regards,
Ray
On Tue, Jan 29, 2013 at 5:07 AM, Glenn Pierce glennpierce@gmail.com wrote:
Thanks for the good ideas everyone. I have a few things to think about.
When accessing through https what will upstream proxies log ? Just
the encrypted url right ?
On 28 January 2013 09:13, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login
in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten
their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this
kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
People often do confuse whether URI is encrypted when HTTPS is used. It is indeed and as phil mentioned, only domain name and a bit other network information is logged on transparent proxies. T
Though password less login is great idea, it has some practical disadvantages like remembering very long URL with token, expiry of those tokens. It might work in some small set-up of enterprise level web app but difficult to scale in multi-user applications.
-Rohit
From: BSDwiz bsdwiz@gmail.com
To: Glenn Pierce glennpierce@gmail.com
Cc: "websecurity@lists.webappsec.org" websecurity@lists.webappsec.org
Sent: Monday, January 28, 2013 4:23 PM
Subject: Re: [WEB SECURITY] Password-less login ?
yep, after the domain name its all encrypted via http(s). but it would still be in the clear in the web server logs.
-phil
On Jan 28, 2013, at 3:07 PM, Glenn Pierce glennpierce@gmail.com wrote:
Thanks for the good ideas everyone. I have a few things to think about.
When accessing through https what will upstream proxies log ? Just the encrypted url right ?
On 28 January 2013 09:13, Glenn Pierce glennpierce@gmail.com wrote:
Hi I like to have opinions on the security of logging into a website
with just a uid
ie
https://someserver.com/login/a4337bc45a8fc544c03f52dc550cd6e1e87021bc896588bd79e901e2
I have inherited a system that provides a login for tablets which login in this manner.
(It needs an automated login for the tablets)
Obviously the url in required to be encrypted by always requiring https.
We often provide one time tokens like this when someone has forgotten their password.
But why not allow this to be a permanent login ?
Why is requiring a uid like above worst than a username,password ?
I believe I am missing something stupid as you would see more of this kind of thing.
That makes be nervous.
Thanks for any feedback.
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org
The Web Security Mailing List
WebSecurity RSS Feed
http://www.webappsec.org/rss/websecurity.rss
Join WASC on LinkedIn http://www.linkedin.com/e/gis/83336/4B20E4374DBA
WASC on Twitter
http://twitter.com/wascupdates
websecurity@lists.webappsec.org
http://lists.webappsec.org/mailman/listinfo/websecurity_lists.webappsec.org