https://youtu.be/mHXH56uYedw
Blue teamer, IT security engineer, ex-pentester, work-life balancing wannabe stoic parent, human being aiming for happiness in life.
https://medium.com/@woff
https://www.linkedin.com/in/zagonmihaly
Blue teamer, IT security engineer, ex-pentester, work-life balancing wannabe stoic parent, human being aiming for happiness in life.
https://medium.com/@woff
https://www.linkedin.com/in/zagonmihaly
With this week’s #InvestigationPath, we’re trying to determine if someone used a legitimate tool for malicious purposes. There are a few things to consider here, but I want to focus on the idea of relationships.
Everything we care about in investigations centers on relationships — when they start, end, and change. A host connects to another host, a user logs into a system, a system executes a file… all relationships.
In this case, even with a single event we have several relationships we want to understand better, and each represents its own investigation path. For example, many mentioned the relationship between bitsadmin.exe and the parent process that launched it.
We care about process hierarchies because most processes are typically launched from a predictable set of parents. When a process has an unexpected parent, that becomes noteworthy. We expect svchost.exe to launch bitsadmin.exe, but maybe not powershell or an Office application.
Of course, sometimes this sort of analysis reveals new relationships to explore. If bitsadmin.exe was run from Microsoft Word, we know that a user opened a file in Word. That extends the path back in time. It also suggests the existence of other, yet unknown relationships. Where did the file come from?
There are more relationships and characteristics to examine here. What about the relationship between the system that executed BITS and the remote host it connected to? What was transferred? Is it normal for a user in this role to launch bitsadmin? Lots of good ideas in the replies.
An event is a start, end, or change in a relationship. A relationship always has at least two entities. The entities have characteristics and the relationship itself has characteristics. I describe this way of thinking as part of my TERC model.
As you break down investigations, think about the relationships you already know about and the ones that are suggested by the evidence you have available. Examine the characteristics of those entities and the relationships. That's how you make sense and move forward.
My response of the week goes to @bherund on Twitter who put in some detailed and diverse thoughts and even included a few lab screenshots with examples (see the whole thread). https://twitter.com/bherund/status/1651674968646733824
Speaking of relationships, what are the most common ones you spend time investigating? That’s something to think about… 🚀 #DFIR #SOCAnalyst
“@chrissanders88 Adversaries may abuse BITS Jobs to download remote arbitrary files. I'd start off with reviewing "Process Create" events - those with "Process Name"=bitsadmin.exe I'd pay close attention to the "Parent Process" & also the Cmdline as that could show file name. 1/n”
Having a paying Canary customer in Antarctica allowed us to say that we had customers on all 7 continents. (We loved it & used to lord this over Duosec when fighting them for "most loved company in infosec").
We no longer have just one customer there...
We now have two ✊💚🐧
☔ Tabletops are amazing tools for a modern security org.
Check out @4n6lady and the AWS CIRT conducting a tabletop exercise for compromised IAM access keys in "The Safe Room" series
Very cool 👏👏👏
---
RT @4n6lady
Our recent Twitch stream featured AWS CIRT conducting a tabletop exercise for compromised IAM access keys, with audience participation in a choose-your-own-adventure style game. 🥳
Check out the replay now:
https://www.twitc…
https://twitter.com/4n6lady/status/1636816989376380928
Another #interview dropped! With @maxpl0it, a Senior Vulnerability Researcher at Interrupt Labs.
"So my first actual CVE was probably in 2017/2018...It was a format string bug in a restricted CLI of a router."
https://medium.com/@xnomas/maxpl0it-interview-with-a-security-researcher-fe75969010e7
We talk about #VulnerabilityResearch, max's journey and some good advice on #MentalHealth.
Liked the #article? Share it, it helps a lot!
This is why I created a Security Engineering *org* at my last place.
Included devsecops, sec architects, cloudsec, threat modeling, and sec engineers (devs) that could onboard with a production for months at a time.
Set up best practices and *help* mitigate priority issues.
Analysis of a real-world attack chain in AWS
https://sysdig.com/blog/cloud-breach-terraform-data-theft/
On the attack chain:
IAM users strike again. Your first step in securing an AWS environment should be nuking anything that has static, long-term credentials. See A retrospective of 2022 cloud data breaches
IMDSv2 only protects your application against SSRFs. If it gets compromised, it doesn't make any difference. The phrasing from Sysdig is misleading here, what should be done is to block access to IMDS from the pod.
If your pod needs to access AWS APIs, use IAM roles for service accounts, see https://infosec.exchange/@christophetd/109491506163807997
If your Lambda function needs to access AWS APIs, use a Lambda execution role, it's much easier and more secure!