- Fixed a crash that could be triggered during 404 fingerprint
failures
- Signature IDs for detected issues are now stored in the report
JSON files.
- Added mod_status, mod_info, MySQL dump, phpMyAdmin SQL dump and
robots.txt signatures.
- Improved the Flash and Silverlight crossdomain policy signatures to
only warn about them when they use wildcards.
- Added Host header XSS testing.
- Added HTML encoding XSS tests to detect scenarios where our
injection string ends up in an attributes that execute HTML encoded
Javascript. For example: onclick.
- Bruteforcing is now disabled for URLs that gave a directory listing.
- Added subject alternate name checking for SSL certificates (cheers
to Matt Caroll for his feedback)
- Added signature matching (see doc/signatures.txt) which means a lot
of the content based issues are no longer hardcoded.
- Added active XSSI test. The passive XSSI stays (for now) but this
active check is more acurate and will remove issues detected by the
passive one if they cannot be confirmed. This reduces false
positives
- Added HTML tag XSS test which triggers when our payload is used
as a tag attribute value but without quotes (courtesy of wavsep).
- Added javascript: scheme XSS testing (courtesy of wavsep).
- Added form based authentication. During these authenticated
scans, skipfish will check if the session has ended and re-authenticates
if necessary.
- Fixed a bug where in slow scans the console output could mess up
due to the high(er) refresh rate.
- Fixed a bug where a missed response during the injection tests could
result in a crash. (courtesy of Sebastian Roschke)
- Restructure the source package a bit by adding a src/, doc/ and
tools/ directory.
- A bugfix to fprint_response() will help reduce false positives that
could occur for differential tests (i.e. the query and shell
injection tests)
- We now suppress implicit cache warnings when dealing with 302, 303
and 307 redirects.
- Added --no-checks which allows a scan to be run without any
injection tests. This still allows bruteforcing and combines well with the
new ability to load URLs from previous scan results.
- We can now parse the pivots.txt, which can be found in the output
directory of older scans. All URLs will be loaded which seriously
speeds up recurring scans.
- Directory bruteforcing now includes a content negotiation trick
where a using a fake mime in the Accept: header will cause some servers to
propose us files via a 406 response.
- A horrible bug fix which caused instable pages not be marked
as such. The result: false positives.
- Crawler update which gives more control over the injection test
scheduling. This comes with the --checks and --checks-toggle
flags to display and enable/disable checks.
- Pages where the response varies are no longer completely
discarded. Instead now we only disable tests that require stability
which increases scan coverage.
- Split the traversal and disclosure test to increase coverage:
traversal checks require stable pages, the disclosure checks can be
performed on all.
- Updated dictionaries and converted them to use the dictionary
optimisations we introduced in 2.03b
- Fixed offline report viewing (thanks to Sebastian Roschke)
- Added NULL byte file disclosure tests
- Added JSP inclusion error check to analyse.c
- Added XSS injection tests for cookies
- Directory listings are now reported as individual (info-type) issues
- Added warning in case the negotiated SSL cipher turns out to be a
weak one (leaving the cipher enumeration to network scanners)
- Added experimental -v flag which can be used to enable (limited)
runtime reporting. This output is written to stderr and should be
redirected to a file, unless you use the -u flag.
- The man page has been rewritten and now includes detailed
descriptions
and examples.
- A whole bunch of small bug fixes
- Fixed a NULL pointer crash when adding "callback" tests to JavaScript
URLs that have a parameter with no value.
- Bug fix in the redirect callback which expected 2 responses but
since 2.04b actually should process 4.
- Option -V eliminated in favor of -W / -S.
- Option -l added to limit the maximum requests per second
(contributed by Sebastian Roschke)
- Option -k added to limit the maximum duration of a scan (contributed
by Sebastian Roschke)
- Support for #ro, -W-; related documentation changes.
- HTTPS -> HTTP form detection.
- Added more diverse traversal and file disclosure tests (including
file:// scheme tests)
- Improved injection detection in <script> sections, where a ' or "
is all we need to inject js code.
- Added check to see if our injection strings end up server
Set-Cookie,
Set-Cookie2 and Content-Type reponse headers
- URLs that give us a Javascript response are now tested with a
"callback=" parameter to find JSONP issues.
- Fixed "response varies" bug in 404 detection where a stable page
would be marked unstable.
- Bugfix to es / eg handling in dictionaries.
- Added the "complete-fast.wl" wordlist which is an es / eg optimized
version of "complete.wl" (resulting in 20-30% fewer requests).
- Fixed a minor glitch in form parsing in analysis.c, courtesy of Niloufar Pahlevan Sadegh.
- Two database.c bugfixes to wordlist handler, courtesy of Shaojie Wang.
- Substantial improvement to SQL injection checks.
- Improvements to directory traversal checks (courtesy of Niels Heinen).
- Fix to numerical brute-force logic.
- Major improvement to directory brute force: much better duplicate elimination in some webserver configurations.
- Added a check for attacker-controlled prefixes on inline responses. This currently leads to UTF-7 BOM XSS, Flash, Java attacks (thanks to Niels Heinen).
- Minor bug fix to path parsing to avoid problems with /.$foo/,
- Improved PHP error detection (courtesy of Niels Heinen),
- Improved dictionary logic (courtesy of Niels Heinen) and new documentation of the same,
- Improved support for file.ext keywords in the dictionary,
- Fixed missing content_checks() in unknown_check_callback()(courtesy of Niels Heinen),
- Improved an oversight in dictionary case sensitivity,
- Improved pivots.txt data,
- Support for supplementary read-only dictionaries (-W +dict),
- Change to directory detection to work around a certain sneaky server behavior.
- TODO: Revise dictionaries!!!
- Proxy support! Currently only works for HTTP, put behind #ifdef PROXY_SUPPORT.
- Change to prefix() and change_prefix() macros to limit the risk of bugs.
signatures in report
- Child signatures now exposed in the report,
- Improvements to duplicate node detection,
- sfscandiff tool added to compare reports.
- Path-based injection attacks now also carried out on file / pathinfo nodes.
- Minor bugfix to try_list logic.
- Slight tweak to form parsing to properly handle specified but empty action=
strings.