Difference between revisions of "REST Security Cheat Sheet"

From OWASP
Jump to: navigation, search
(Adding a definition for REST)
(Protect Session State)
(21 intermediate revisions by 3 users not shown)
Line 1: Line 1:
 +
= DRAFT CHEAT SHEET - WORK IN PROGRESS =
 +
= Introduction =
 +
 
[http://en.wikipedia.org/wiki/Representational_state_transfer REST] or REpresentational State Transfer is a means of expressing specific entities in a system by URL path elements, REST is not an architecture but it is an architectural style to build services on top of the Web. REST allows interaction with a web-based system via simplified URL's rather than complex request body or <tt>POST</tt> parameters to request specific items from the system. This document serves as a guide (although not exhaustive) of best practices to help REST-based services.
 
[http://en.wikipedia.org/wiki/Representational_state_transfer REST] or REpresentational State Transfer is a means of expressing specific entities in a system by URL path elements, REST is not an architecture but it is an architectural style to build services on top of the Web. REST allows interaction with a web-based system via simplified URL's rather than complex request body or <tt>POST</tt> parameters to request specific items from the system. This document serves as a guide (although not exhaustive) of best practices to help REST-based services.
  
== Check Authorization for User-Specific Entities ==
+
= Authentication and Session Management =
While REST is useful for targeting specific entities in the system, the URL itself should not be the only authorizing token for sensitive entities (such as account transactions, personally-identifying information, etc.). Proper authentication should take place, and the token for authorization should be sent as a cookie. Furthermore, no Personally-Identifiable Information (such as bank-account number, credit card number, etc.) should be used as a parameter to request an entity. See the [DOR (Direct Object Reference) Prevention Cheat Sheet] for strategies for preventing Direct Object Reference weaknesses.
+
  
== Whitelist-Only Response Types ==
+
RESTful web services should use session based authentication, either by establishing a session token via a POST, or using an API key as a POST body argument or as a cookie. Usernames and passwords, session tokens and API keys should not appear in the URL, as this can be captured in web server logs and makes them intrinsically valuable.  
It is common for REST services to allow multiple response types (e.g. <tt>application/xml</tt> or <tt>application/json</tt>, and the client specifies the preferred order of response types by the <tt>Accept</tt> header in the request. '''Do NOT''' simply copy the <tt>Accept</tt> header to the <tt>Content-type</tt> header of the response. Reject the request (ideally with a <tt>406 Not Acceptable</tt> response) if the <tt>Accept</tt> header does not specifically contain one of the allowable types.
+
  
Because there are many MIME types for the typical response types, it's important to document for clients specifically which MIME types should be used.
+
OK:
 +
 
 +
* [https://example.com/resourceCollection/123/action https://example.com/resourceCollection/<id>/action]
 +
* https://twitter.com/vanderaj/lists
 +
 
 +
NOT OK:
 +
 
 +
* [https://example.com/controller/123/action?apiKey=a53f435643de32 https://example.com/controller/<id>/action?apiKey=a53f435643de32] (API Key in URL)
 +
* [http://example.com/controller/123/action?apiKey=a53f435643de32 http://example.com/controller/<id>/action?apiKey=a53f435643de32] (transaction not protected by TLS and API Key in URL)
 +
 
 +
== Protect Session State ==
 +
 
 +
Many web services are written to be as stateless as possible. This usually ends up with a state blob being sent as part of the transaction.
 +
 
 +
* Consider just using the session token or API key to maintain client state in a server side cache. This is directly equivalent to how normal web apps do it, and there's a reason why this is moderately safe.
 +
* Anti-replay. Attackers will cut and paste a blob and become someone else.  Consider using a time limited encryption key, keyed against the session token or API key, date and time and incoming IP address.
 +
* Don't make it easy to decrypt and change the internal state to be much better than it should be.
 +
 
 +
In short, even if you have a brochureware web site, don't put in https://example.com/users/2313/edit?isAdmin=false&debug=false&allowCSRPanel=false as you will quickly end up with a lot of admins, and help desk helpers, and "developers".
 +
 
 +
= Authorization =
 +
 
 +
== Anti-farming ==
 +
 
 +
Many RESTful web services are put up, and then farmed, such as a price matching website or aggregation service. There's no technical method of preventing this use, so strongly consider means to encourage it as a business model by making high velocity farming is possible for a fee, or contractually limiting service using terms and conditions. CAPTCHAs and similar methods can help reduce simpler adversaries, but not well funded or technically competent adversaries. Using mutually assured client side TLS certificates may be a method of limiting access to trusted organizations, but this is by no means certain, particularly if certificates are posted deliberately or by accident to the Internet.
 +
 
 +
== Protect HTTP methods ==
 +
 
 +
RESTful API often use GET (read), POST (create), PUT (replace/update) and DELETE (to delete a record). Not all of these are valid choices for every single resource collection, user, or action. Make sure the  incoming HTTP method is valid for the session token/API key and associated resource collection, action, and record. For example, if you have an RESTful API for a library, it's not okay to allow anonymous users to DELETE book catalog entries, but it's fine for them to GET a book catalog entry, whereas for the librarian, both of these are valid uses.  
  
 
== Whitelist Allowable Methods ==
 
== Whitelist Allowable Methods ==
 +
 
It is common with RESTful services to allow multiple methods for a given URL for different operations on that entity. For example, a <tt>GET</tt> request might read the entity while <tt>POST</tt> would update an existing entity, <tt>PUT</tt> would create a new entity, and <tt>DELETE</tt> would delete an existing entity. It is important for the service to properly restrict the allowable verbs such that only the allowed verbs will work, all others return a proper response code (for example, a <tt>403 Forbidden</tt>).
 
It is common with RESTful services to allow multiple methods for a given URL for different operations on that entity. For example, a <tt>GET</tt> request might read the entity while <tt>POST</tt> would update an existing entity, <tt>PUT</tt> would create a new entity, and <tt>DELETE</tt> would delete an existing entity. It is important for the service to properly restrict the allowable verbs such that only the allowed verbs will work, all others return a proper response code (for example, a <tt>403 Forbidden</tt>).
  
 
In Java EE in particular, this can be difficult to implement properly. See [https://www.aspectsecurity.com/wp-content/plugins/download-monitor/download.php?id=18 Bypassing Web Authentication and Authorization with HTTP Verb Tampering] for an explanation of this common misconfiguration.
 
In Java EE in particular, this can be difficult to implement properly. See [https://www.aspectsecurity.com/wp-content/plugins/download-monitor/download.php?id=18 Bypassing Web Authentication and Authorization with HTTP Verb Tampering] for an explanation of this common misconfiguration.
  
== Non-REST Authentication ==
+
== Protect privileged actions and sensitive resource collections ==
Authentication for a REST-based service should not take place by passing the user ID and password in path elements of the URL. URL's are commonly cached by browsers, proxies, etc. so sensitive information should not be included directly in the URL.
+
 
 +
Not every user has a right to every web service. This is vital, as you don't want administrative web services to be misused:
 +
 
 +
* https://example.com/admin/exportAllData
 +
 
 +
The session token or API key should be sent along as a cookie or body parameter to ensure that privileged collections or actions are properly protected from unauthorized use.
 +
 
 +
== Protect against cross-site request forgery ==
 +
 
 +
For resources exposed by RESTful web services, it's important to make sure any PUT, POST and DELETE request is protected from Cross Site Request Forgery. Typically one would use a token based approach. See [[Cross-Site Request Forgery (CSRF) Prevention Cheat Sheet]] for more information on how to implement CSRF-protection.
 +
 
 +
CSRF is easily achieved even using random tokens if any XSS exists within your application, so please make sure you understand [[XSS (Cross Site Scripting) Prevention Cheat Sheet|how to prevent XSS]].
 +
 
 +
== Direct object references ==
 +
 
 +
It may seem obvious, but if you had a bank account REST web service, you have to make sure there is adequate checking of primary and foreign keys:
 +
 
 +
* https://example.com/account/325365436/transfer?amount=$100.00&toAccount=473846376
 +
 
 +
In this case, it would be possible to transfer money from any account to any other account, which is clearly insane. Not even a random token makes this safe.
 +
 
 +
* https://example.com/invoice/2362365
 +
 
 +
In this case, it would be possible to get a copy of all invoices.
 +
 
 +
Please make sure you understand how to protect against [[Top_10_2010-A4-Insecure_Direct_Object_References|direct object references]] in the OWASP Top 10 2010.
 +
 
 +
= Input Validation =
 +
 
 +
== Input validation 101 ==
 +
 
 +
Everything you know about input validation applies to RESTful web services, but add 10% because automated tools can easily fuzz your interfaces for hours on end at high velocity. So:
 +
 
 +
* Assist the user > Reject input > Sanitize (filtering) > No input validation
 +
 
 +
Assisting the user makes the most sense, as the most common scenario is "problem exists between keyboard and computer" (PEBKAC). Help the user input high quality data into your web services, such as ensuring a Zip code makes sense for the supplied address, or the date makes sense. If not, reject that input. If they continue on, or it's a text field or some other difficult to validate field, input sanitization is a losing proposition but still better than XSS or SQL injection. If you're already reduced to  sanitization or no input validation, make sure output encoding is very strong for your application.
 +
 
 +
Log input validation failures, particularly if you assume that client side code you wrote is going to call your web services. The reality is that anyone can call your web services, so assume that someone who is performing hundreds of failed input validations per second is up to no good. Also consider rate limiting the API to a certain number of requests per hour or day to prevent abuse.
 +
 
 +
== Strong typing ==
 +
 
 +
It's difficult to perform most attacks if the only allowed values are true or false, or a number, or one of a small number of acceptable values. Strongly type incoming data as quickly as possible.
 +
 
 +
== Validate Incoming Content-Types ==
 +
 
 +
When POSTing or PUTing new data, the client will specify the a Content-Type (e.g. <tt>application/xml</tt> or <tt>application/json</tt>) of the incoming data. The client should never assume the Content-Type, but always check that the Content-Type header and the content is of the same type. A lack of Content-Type header or an unexpected Content-Type header, should result in the server rejecting the Content with a <tt>406 Not Acceptable</tt> response.
 +
 
 +
== Validate Response Types ==
 +
 
 +
It is common for REST services to allow multiple response types (e.g. <tt>application/xml</tt> or <tt>application/json</tt>, and the client specifies the preferred order of response types by the <tt>Accept</tt> header in the request. '''Do NOT''' simply copy the <tt>Accept</tt> header to the <tt>Content-type</tt> header of the response. Reject the request (ideally with a <tt>406 Not Acceptable</tt> response) if the <tt>Accept</tt> header does not specifically contain one of the allowable types.
 +
 
 +
Because there are many MIME types for the typical response types, it's important to document for clients specifically which MIME types should be used.
 +
 
 +
== XML Input Validation ==
 +
 
 +
XML-based services must ensure that they are protected against common XML based attacks by using secure XML-parsing. This typically means XML External Entity attacks, XML-signature wrapping etc. See  [http://ws-attacks.org http://ws-attacks.org] for examples of such attacks.
 +
 
 +
= Output Encoding =
 +
 
 +
== Send security headers ==
 +
 
 +
To make sure the content of a given resources is interpreted correctly by the browser, the server should always send the Content-Type header with the correct Content-Type. The server should also send an <tt>X-Content-Type-Options: nosniff</tt> to make sure the browser does not try to detect a different Content-Type than what is actually sent (can lead to XSS).
 +
 
 +
Additionally the client should send an <tt>X-Frame-Options: deny</tt> to protect against drag'n drop clickjacking attacks in older browsers.
 +
 
 +
== JSON encoding ==
 +
 
 +
A key concern with JSON encoders is to prevent arbitrary JavaScript remote code execution within the browser, or if you're using node.js, on the server. It's vital that you use the JSON library of your choice to encode user supplied data properly to prevent the execution of user supplied input on the browser.
 +
 
 +
When sending values to the browser, strongly consider using .value rather than .innerHTML updates, as this protects against simple DOM XSS attacks.
 +
 
 +
== XML encoding ==
 +
 
 +
XML should never be hand concatenated, it should always be constructed using an XML parser. This ensures that the XML content sent to the browser is parseable, and should not contain XML injection. For more information, please see the [[Web Service Security Cheat Sheet]].
 +
 
 +
= Cryptography =
 +
 
 +
== Data in transit ==
 +
 
 +
Unless completely read only public information, the use of TLS should be mandated, particularly where credentials, updates, deletions and any value transactions are performed. The overhead of TLS is negligible on modern hardware, with a minor latency increase that is more than compensated by safety for the end user.
 +
 
 +
Consider the use of mutually authenticated client side certificates to provide additional protection for highly privileged web services.
 +
 
 +
== Data in storage ==
 +
 
 +
Leading practices are recommended as per any web application when it comes to correctly handling stored sensitive or regulated data. For more information, please see [[Top 10 2010-A7|OWASP Top 10 2010 - A7 Insecure Cryptographic Storage] for more information.
 +
 
 +
= Related Articles =
 +
 
 +
{{Cheatsheet_Navigation}}
 +
 
 +
= Authors and Primary Editors  =
 +
 
 +
Erlend Oftedal - erlend.oftedal@owasp.org<br/>
 +
Andrew van der Stock - vanderaj@owasp.org<br/>
  
[[Category:How_To]] [[Category:Cheatsheets]]
+
[[Category:Cheatsheets]]

Revision as of 16:36, 26 November 2012

Contents

DRAFT CHEAT SHEET - WORK IN PROGRESS

Introduction

REST or REpresentational State Transfer is a means of expressing specific entities in a system by URL path elements, REST is not an architecture but it is an architectural style to build services on top of the Web. REST allows interaction with a web-based system via simplified URL's rather than complex request body or POST parameters to request specific items from the system. This document serves as a guide (although not exhaustive) of best practices to help REST-based services.

Authentication and Session Management

RESTful web services should use session based authentication, either by establishing a session token via a POST, or using an API key as a POST body argument or as a cookie. Usernames and passwords, session tokens and API keys should not appear in the URL, as this can be captured in web server logs and makes them intrinsically valuable.

OK:

NOT OK:

Protect Session State

Many web services are written to be as stateless as possible. This usually ends up with a state blob being sent as part of the transaction.

  • Consider just using the session token or API key to maintain client state in a server side cache. This is directly equivalent to how normal web apps do it, and there's a reason why this is moderately safe.
  • Anti-replay. Attackers will cut and paste a blob and become someone else. Consider using a time limited encryption key, keyed against the session token or API key, date and time and incoming IP address.
  • Don't make it easy to decrypt and change the internal state to be much better than it should be.

In short, even if you have a brochureware web site, don't put in https://example.com/users/2313/edit?isAdmin=false&debug=false&allowCSRPanel=false as you will quickly end up with a lot of admins, and help desk helpers, and "developers".

Authorization

Anti-farming

Many RESTful web services are put up, and then farmed, such as a price matching website or aggregation service. There's no technical method of preventing this use, so strongly consider means to encourage it as a business model by making high velocity farming is possible for a fee, or contractually limiting service using terms and conditions. CAPTCHAs and similar methods can help reduce simpler adversaries, but not well funded or technically competent adversaries. Using mutually assured client side TLS certificates may be a method of limiting access to trusted organizations, but this is by no means certain, particularly if certificates are posted deliberately or by accident to the Internet.

Protect HTTP methods

RESTful API often use GET (read), POST (create), PUT (replace/update) and DELETE (to delete a record). Not all of these are valid choices for every single resource collection, user, or action. Make sure the incoming HTTP method is valid for the session token/API key and associated resource collection, action, and record. For example, if you have an RESTful API for a library, it's not okay to allow anonymous users to DELETE book catalog entries, but it's fine for them to GET a book catalog entry, whereas for the librarian, both of these are valid uses.

Whitelist Allowable Methods

It is common with RESTful services to allow multiple methods for a given URL for different operations on that entity. For example, a GET request might read the entity while POST would update an existing entity, PUT would create a new entity, and DELETE would delete an existing entity. It is important for the service to properly restrict the allowable verbs such that only the allowed verbs will work, all others return a proper response code (for example, a 403 Forbidden).

In Java EE in particular, this can be difficult to implement properly. See Bypassing Web Authentication and Authorization with HTTP Verb Tampering for an explanation of this common misconfiguration.

Protect privileged actions and sensitive resource collections

Not every user has a right to every web service. This is vital, as you don't want administrative web services to be misused:

The session token or API key should be sent along as a cookie or body parameter to ensure that privileged collections or actions are properly protected from unauthorized use.

Protect against cross-site request forgery

For resources exposed by RESTful web services, it's important to make sure any PUT, POST and DELETE request is protected from Cross Site Request Forgery. Typically one would use a token based approach. See Cross-Site Request Forgery (CSRF) Prevention Cheat Sheet for more information on how to implement CSRF-protection.

CSRF is easily achieved even using random tokens if any XSS exists within your application, so please make sure you understand how to prevent XSS.

Direct object references

It may seem obvious, but if you had a bank account REST web service, you have to make sure there is adequate checking of primary and foreign keys:

In this case, it would be possible to transfer money from any account to any other account, which is clearly insane. Not even a random token makes this safe.

In this case, it would be possible to get a copy of all invoices.

Please make sure you understand how to protect against direct object references in the OWASP Top 10 2010.

Input Validation

Input validation 101

Everything you know about input validation applies to RESTful web services, but add 10% because automated tools can easily fuzz your interfaces for hours on end at high velocity. So:

  • Assist the user > Reject input > Sanitize (filtering) > No input validation

Assisting the user makes the most sense, as the most common scenario is "problem exists between keyboard and computer" (PEBKAC). Help the user input high quality data into your web services, such as ensuring a Zip code makes sense for the supplied address, or the date makes sense. If not, reject that input. If they continue on, or it's a text field or some other difficult to validate field, input sanitization is a losing proposition but still better than XSS or SQL injection. If you're already reduced to sanitization or no input validation, make sure output encoding is very strong for your application.

Log input validation failures, particularly if you assume that client side code you wrote is going to call your web services. The reality is that anyone can call your web services, so assume that someone who is performing hundreds of failed input validations per second is up to no good. Also consider rate limiting the API to a certain number of requests per hour or day to prevent abuse.

Strong typing

It's difficult to perform most attacks if the only allowed values are true or false, or a number, or one of a small number of acceptable values. Strongly type incoming data as quickly as possible.

Validate Incoming Content-Types

When POSTing or PUTing new data, the client will specify the a Content-Type (e.g. application/xml or application/json) of the incoming data. The client should never assume the Content-Type, but always check that the Content-Type header and the content is of the same type. A lack of Content-Type header or an unexpected Content-Type header, should result in the server rejecting the Content with a 406 Not Acceptable response.

Validate Response Types

It is common for REST services to allow multiple response types (e.g. application/xml or application/json, and the client specifies the preferred order of response types by the Accept header in the request. Do NOT simply copy the Accept header to the Content-type header of the response. Reject the request (ideally with a 406 Not Acceptable response) if the Accept header does not specifically contain one of the allowable types.

Because there are many MIME types for the typical response types, it's important to document for clients specifically which MIME types should be used.

XML Input Validation

XML-based services must ensure that they are protected against common XML based attacks by using secure XML-parsing. This typically means XML External Entity attacks, XML-signature wrapping etc. See http://ws-attacks.org for examples of such attacks.

Output Encoding

Send security headers

To make sure the content of a given resources is interpreted correctly by the browser, the server should always send the Content-Type header with the correct Content-Type. The server should also send an X-Content-Type-Options: nosniff to make sure the browser does not try to detect a different Content-Type than what is actually sent (can lead to XSS).

Additionally the client should send an X-Frame-Options: deny to protect against drag'n drop clickjacking attacks in older browsers.

JSON encoding

A key concern with JSON encoders is to prevent arbitrary JavaScript remote code execution within the browser, or if you're using node.js, on the server. It's vital that you use the JSON library of your choice to encode user supplied data properly to prevent the execution of user supplied input on the browser.

When sending values to the browser, strongly consider using .value rather than .innerHTML updates, as this protects against simple DOM XSS attacks.

XML encoding

XML should never be hand concatenated, it should always be constructed using an XML parser. This ensures that the XML content sent to the browser is parseable, and should not contain XML injection. For more information, please see the Web Service Security Cheat Sheet.

Cryptography

Data in transit

Unless completely read only public information, the use of TLS should be mandated, particularly where credentials, updates, deletions and any value transactions are performed. The overhead of TLS is negligible on modern hardware, with a minor latency increase that is more than compensated by safety for the end user.

Consider the use of mutually authenticated client side certificates to provide additional protection for highly privileged web services.

Data in storage

Leading practices are recommended as per any web application when it comes to correctly handling stored sensitive or regulated data. For more information, please see [[Top 10 2010-A7|OWASP Top 10 2010 - A7 Insecure Cryptographic Storage] for more information.

Related Articles

OWASP Cheat Sheets Project Homepage

Developer Cheat Sheets (Builder)

Assessment Cheat Sheets (Breaker)

Mobile Cheat Sheets

OpSec Cheat Sheets (Defender)

Draft Cheat Sheets

Authors and Primary Editors

Erlend Oftedal - erlend.oftedal@owasp.org
Andrew van der Stock - vanderaj@owasp.org