MediaWiki before 1.19.6 and 1.20.x before 1.20.5 does not allow extensions to prevent password changes without using both Special:PasswordReset and Special:ChangePassword, which allows remote attackers to bypass the intended restrictions of an extension that only implements one of these blocks.

MediaWiki before 1.19.6 and 1.20.x before 1.20.5 allows remote attackers to conduct cross-site scripting (XSS) attacks, as demonstrated by a CDATA section containing valid UTF-7 encoded sequences in a SVG file, which is then incorrectly interpreted as UTF-8 by Chrome and Firefox.

Cross-site scripting (XSS) vulnerability in the outputPage function in includes/SkinTemplate.php in MediaWiki before 1.17.5, 1.18.x before 1.18.4, and 1.19.x before 1.19.1 allows remote attackers to inject arbitrary web script or HTML via the uselang parameter to index.php/Main_page.

MediaWiki before 1.17.1 does not check for read permission before handling action=ajax requests, which allows remote attackers to obtain sensitive information by (1) leveraging the SpecialUpload::ajaxGetExistsWarning function, or by (2) leveraging an extension, as demonstrated by the CategoryTree, ExtTab, and InlineEditor extensions.

MediaWiki before 1.17.1 allows remote attackers to obtain the page titles of all restricted pages via a series of requests involving the (1) curid or (2) oldid parameter.

includes/User.php in MediaWiki before 1.16.5, when wgBlockDisablesLogin is enabled, does not clear certain cached data after verification of an auth token fails, which allows remote attackers to bypass authentication by creating crafted wikiUserID and wikiUserName cookies, or by leveraging an unattended workstation.

Cross-site scripting (XSS) vulnerability in MediaWiki before 1.16.5, when Internet Explorer 6 or earlier is used, allows remote attackers to inject arbitrary web script or HTML via an uploaded file accessed with a dangerous extension such as .shtml at the end of the query string, in conjunction with a modified URI path that has a %2E sequence in place of the . (dot) character. NOTE: this vulnerability exists because of an incomplete fix for CVE-2011-1578 and CVE-2011-1587.

Cross-site scripting (XSS) vulnerability in MediaWiki before 1.16.4, when Internet Explorer 6 or earlier is used, allows remote attackers to inject arbitrary web script or HTML via an uploaded file accessed with a dangerous extension such as .html located before a ? (question mark) in a query string, in conjunction with a modified URI path that has a %2E sequence in place of the . (dot) character. NOTE: this vulnerability exists because of an incomplete fix for CVE-2011-1578.

The transwiki import functionality in MediaWiki before 1.16.3 does not properly check privileges, which allows remote authenticated users to perform imports from any wgImportSources wiki via a crafted POST request.

The checkCss function in includes/Sanitizer.php in the wikitext parser in MediaWiki before 1.16.3 does not properly validate Cascading Style Sheets (CSS) token sequences, which allows remote attackers to conduct cross-site scripting (XSS) attacks or obtain sensitive information by using the \2f\2a and \2a\2f hex strings to surround CSS comments.

Cross-site scripting (XSS) vulnerability in MediaWiki before 1.16.3, when Internet Explorer 6 or earlier is used, allows remote attackers to inject arbitrary web script or HTML via an uploaded file accessed with a dangerous extension such as .html at the end of the query string, in conjunction with a modified URI path that has a %2E sequence in place of the . (dot) character.

Multiple directory traversal vulnerabilities in (1) languages/Language.php and (2) includes/StubObject.php in MediaWiki 1.8.0 and other versions before 1.16.2, when running on Windows and possibly Novell Netware, allow remote attackers to include and execute arbitrary local PHP files via vectors related to a crafted language file and the Language::factory function.

Cross-site scripting (XSS) vulnerability in MediaWiki before 1.16.2 allows remote attackers to inject arbitrary web script or HTML via crafted Cascading Style Sheets (CSS) comments, aka "CSS injection vulnerability."