Saturday, November 26, 2011

Tool review: NetworkMiner Professional 1.2

I've been slow in undertaking this review as NetworkMiner's Erik Hjelmvik sent me NetworkMiner Professional 1.1 when it was released and 1.2 is now available.
Seeing Richard Bejtlich's discussion of Pro 1.2 has served to get me off the schnide and is helpful as I will point you to his post as an ideal primer while I go into to a bit deeper detail as to some of NetworkMiner's power as well as what distinguishes Professional from the free edition.
I covered NetworkMiner in toolsmith in August 2008 back when it was version 0.84. Erik has accomplished all of his goals for improvement as identified in the article including reporting, faster parsing of large PCAP files (.735 MB/s at the command-line),  more protocols implemented, and PIPI (Port Independent Protocol Identification). NetworkMiner Professional 1.2 incorporates all of the above.
To exemplify NetworkMiner Professional's PIPI capabilities, I changed my lab web server port to 6667, then set NetworkMiner to grab a live capture while browsing to the reconfigured server.
Note: you need to Run as Administrator to grab the interface on Windows 7.
Sure, it's more likely that someone would be more likely to hide evil traffic over port 80 but you get the point. As Richard said, "PIPI has many security implications for discovery and (preferably) denial of covert channels, back doors, and other policy-violating channels."
Note as seen in Figure 1 that NetworkMiner Professional clearly differentiates HTTP traffic regardless of the fact that it traversed port 6667.

Figure 1
I was a bit surprised to note that the Hosts view as seen in Figure 1 did not identify that any data was pushed as cleartext although it unequivocally identified the admin/password combination I sent in both the Cleartext view and the Credentials view.
I used an 18.8MB PCAP from the Xplico sample set as it includes a plethora of protocols and carve-able content with which to test NetworkMiner Professional.
Exporting results to CSV for reporting is as easy as File --> Export to CSV and selecting output of your choosing. As seen in Figure 2 I opted for Messages as NetworkMiner Professional cleanly carved out an MSN to Yahoo email session (HTTPS, anyone?).

Figure 2
Geo IP localization is a real standout too. You'll see it in play as you explore host details in Hosts view as seen in Figure 3.
Figure 3
You may find host coloring useful too should you wish to tag hosts for easy identification later as seen in Figure 4.

Figure 4
Finally, I am most excited about NetworkMinerCLI for command-line scripting support. 
I ran a PCAP taken from a VM infected with Trojan-Downloader.Win32.Banload.MC through NetworkMinerCLI and was amply rewarded for my efforts...right after I excluded the output directory from AV detection.
Figure 5 shows the command executed at the prompt coupled with the resulting assembled files and CSVs populated to the output directory as seen via Windows Explorer.

Figure 5
The assembled files included all the malicious binaries disguised as JPGs as downloaded from the evil server. File carving network forensic analysis juju with easy CLI scripting. Bonus!

In closing, NetworkMiner Professional 1.2 is a mature, highly useful tool and well worthy of consideration for purchase by investigators and analysts tasked with NFAT activity. 
I'm glad to provide further feedback via email and recommend you reach out to Erik as well via info [at] netresec.com if you have questions.






Wednesday, November 02, 2011

toolsmith: OWASP ZAP - Zed Attack Proxy



Prerequisites
Java Runtime Environment
ZAP runs on Linux, Mac OS X, and Windows

Happy Thanksgiving: "As we express our gratitude, we must never forget that the highest appreciation is not to utter words, but to live by them." -JFK

Introduction
November 2011’s toolsmith is the 61st in the series for the ISSA Journal, thus marking five years of extensive tools analysis for information security practitioners. Thank you for coming along for the ride.
Fresh on the heels of a successful presentation on OWASP Top 10 Tools and Tactics at an even more successful ISSA International in Baltimore I was motivated to give full coverage this month to the OWASP Zed Attack Proxy, better known as ZAP. I had presented ZAP as a tool of choice when assessing OWASP Top Ten A1 – Injection but, as so many of the tools discussed, ZAP delivers plenty of additional functionality worthy of in-depth discussion.
OWASP ZAP is a fork of the once favored Paros Proxy, which has not been updated since August 2006. As such, it should be noted with no small irony that we covered Paros in December 2006; this is an excellent opportunity to show you how far ZAP has come from the original project.
ZAP is the result of Simon Bennetts’ (Psiinon) hard work, though he’s got help from co-lead Axel Neumann (@a_c_neumann) and many contributors.
As an official OWASP project, ZAP enjoys extensive use and development support as an “easy to use integrated penetration testing tool for finding vulnerabilities in web applications.”
Simon offered a veritable plethora of feedback for this article, as provided throughout the rest of the introduction. He indicated that he originally released ZAP specifically for developers and functional testers; a group which he believes is poorly represented in the security tools market.
Ease of use was a prime concern, as was documentation and to his surprise it turned out that it was the security folk who took up ZAP the quickest, providing great feedback, reporting issues and asking for lots of enhancements. Simon still wants ZAP to be ideal for people new to web application security but it’s also going to be enhanced with more and more advanced features aimed at profession penetration testers.
Simon also wanted ZAP to be a community project; there are many open source security tools that are tightly controlled by one individual or company. While he doesn’t have a problem with that fact he does believe that the real strength of open source comes when anyone can contribute to a project and take it in directions its initial developers never envisaged.
Anyone and everyone is welcome to contribute to ZAP, and not necessarily coding only; they welcome help with testing, documentation, localization, issues identification and enhancement requests. Help spread the word as well via articles, reviews, videos, blogs, Twitter, etc.
ZAP is also one of the few open source security tools to be fully internationalized. It has been translated into 10 languages and download statistics indicate that approximately half of the ZAP users worldwide are likely to be non-native English speakers.
ZAP is intended to provide everything that you need to perform a penetration test on a web application.
If you are new to web application security then it might be the only security tool you need. However, if you're an experienced penetration tester be sure to include it as one of the many tools in your toolbox.
As a result, the development team is trying to make it as easy as possible to integrate ZAP with other tools. They provide a way to invoke other applications from within ZAP passing across the current context. In version 1.3 they introduced an API which allows the core ZAP functionality to be invoked by a REST API, and will be extended to cover even more of ZAP's features in future releases.
This is an ideal way for other applications to directly drive ZAP, and can be used when ZAP is running in 'headless' mode (i.e. without the UI).
They've also put together a POC showing how ZAP can be used by developers to include basic security tests in their continuous integration framework and be alerted to potential security vulnerabilities within hours of checking code.
Simon and team don’t believe in reinventing the wheel, which is why they always seek high quality open source components to reuse before implementing a new feature from scratch.
As such, the brute force/forced browsing support is provided via DirBuster and fuzzing makes use of the JBroFuzz libraries (both OWASP projects).
Amongst the more advanced features that users might not be aware of is that ZAP keeps track of all of the anti-CSRF tokens it finds. If fuzzing a form with an anti CSRF-token in it, ZAP can regenerate the token for each of the payloads you fuzz with. There’s also an experimental option that allows this to be turned on when using the active scanner as well. I can say that quality CSRF testing is not commonplace among ZAP’s web application testing contemporaries.
For ZAP version 1.4 the development team has decided to focus on:
·         Improving the active and passive scanners
·         Improving stability (especially for large sites)
·         Session token analysis
In July 2011 ZAP was evaluated and designated as a 'stable' OWASP project, the highest level currently available. Further, OWASP projects are now being restructured; ZAP has been designated as one of the small number of 'flagship' projects.
Rightfully so; thank you Simon.
Let’s run ZAP through its paces.

ZAP Installation and Configuration

ZAP is installation is very simple. Once unpacked on your preferred platform, invoke ZAP from the application icon or at the command prompt via the appropriate executable. A current Java Runtime Environment is a requirement as all the executables (EXE, BAT, SH) invoke java –jar zap.jar org.zaproxy.xap.ZAP.
Most importantly ZAP, runs as a proxy. Configure your preferred browser to proxy via localhost and the default port of 8080. I change the port to 8088 to avoid conflict with other proxies and services. You can change the port under Tools à Options à Local proxy if you run multiple proxies that you bounce between during assessments. I do and as such I use the Firefox add-on FoxyProxy to quickly dial in my proxy of choice.
You must also generate an SSL certificate in order to use and test SSL enabled sites. You will be prompted to do when running ZAP for the first time.

ZAP Use

In addition to the aforementioned Security Regression Tests for developers, the OWASP ZAP project offers ZAP Web Application Vulnerability Examples, or ZAP WAVE. Download it and drop zap-wave.war in the webapps directory of your favorite servlet engine. On Debian/Ubuntu systems sudo apt-get install tomcat6 will get you in business with said servlet engine quickly. In addition to a LAMP stack on an Ubuntu 11.10 VM I run Tomcat for just such occasions. OWASP WebGoat also runs as a standalone test bed or via a servlet engine.
Enable ZAP, with your browser configured to proxy through it, then navigate to the system (VM or real steel) hosting ZAP WAVE, usually on port 8080. As an example: http://192.168.140.137:8080/zapwave/.
ZAP WAVE includes “active” vulnerabilities such as cross-site scripting and SQL injection as well as “passive” vulnerabilities including three types of information leakage and two session vulnerabilities.
There are also pending false positives that are not yet ready for primetime.
The developers recommend that you explore the target app with ZAP enabled as a proxy, and touch as much of it as possible before spidering. Doing so helps ZAP find more vulns as you may cross paths with error messages, etc.
I typically visit the root of the application hierarchy for a web application I wish to assess, right-click on it, select Attack, then Spider site. This crawls the entire site hierarchy and populates the tree view under the Sites tab in ZAP’s left pane as seen in Figure 1.

Figure 1: ZAP spidering
Crawling/spidering can have unintended side-effects on an application, even adding or deleting records in a database, so be advised.
A good crawl ensures a better active scan, but before beginning a scan, set your Scan Policy via Analyze à Scan Policy as seen in Figure 2. You may wish to more narrowly scope your scan activity to just the likes of information gathering or SQL injection as seen in Figure 2.

Figure 2: ZAP scan policy
Spidering and scan policy configuration complete, right click the root, or a specific node you wish to assess as you can choose Attack à Active scan site or Attack à Active scan node.
You can also exclude a site from the scope in a similar fashion.
A full scan of the ZAP WAVE instance completed in very short order; results were immediate as seen in Figure 3.

Figure 3: ZAP scan results
ZAP includes the expected Encode/Decode/Hash functionality via Edit à Encode/Decode/Hash or Tools à Encode/Decode/Hash along with a manual editor for generating manual requests. I’ll often run ZAP for nothing more than encoding, decoding, and hashing; it’s a great utility.
The Port Scan feature is also useful. It will select the in-scope host by default; just click the Port Scan tab then the start button.
The Brute Force tab is a function of the above-mentioned DirBuster component and includes seven dictionary lists to choose from. I ran this against my full host VM rather just the servlet element and included the dictionary-list-1.0 dictionary for a simple, quick test.

Figure 4: ZAP DirBuster at work
One of my favorite ZAP features (there are many) is the Fuzzer. Per the Fuzzer component guidance:
·         Select a request in the Sites or History tab
·         Highlight the string you wish to fuzz in the Request tab
·         Right click in the Request tab and select 'Fuzz...'
·         Select the Fuzz Category and one or more Fuzzers
·         Press the Fuzz button
·         The results listed in the Fuzzer tab - select them to see the full requests and responses.
The fuzzer, like the scanner, includes functionality which causes ZAP to automatically regenerate the tokens when required
I ran Fuzzer against http://192.168.140.137:8080/zapwave/active/xss/xss-form-anti-csrf.jsp and fuzzed the anticsrf and name variables as it is a recent addition per the ZAP WAVE download site.
As seen in Figure 5, the fuzzer offers a wider array fuzzers within a given category.

FIGURE 5: ZAP fuzzer config
In the understanding that fuzzing is the art of submitting a great deal of invalid or unexpected data to a target, you can look for variations in results such as response code (200 OK) and response times. Where normal response times per request average between 2ms and 4ms for ZAP WAVE hosted on a local VM, one request in particular stood out at a 402ms response time. I checked for the string passed and cracked up.
%3CIMG+SRC%3D%60javascript%3Aalert%28%22RSnake+says%23%23%23+%27XSS%27%22%29%60%3E
Or, courtesy of the handy ZAP decoder:



Mr. Slowloris HTTP DoS himself causing grind even here. ;-)

In Conclusion

ZAP deserves its status as an OWASP flagship project. Whether you’re a seasoned veteran or new to the web application security game make the Zed Attack Proxy part of your arsenal. I’d go so far as to say, as 2011 is winding down, that ZAP feels like a likely front runner for 2011 Toolsmith Tool of the Year. But that is for you to decide, dear reader. Let me know if you agree.
Ping me via email if you have questions (russ at holisticinfosec dot org).
Cheers…until next month.

Acknowledgements

Simon Bennetts (Psiinon) for project feedback and details
Axel Neumann (@a_c_neumann) for draft review