RAG, ICL, and Windows Events: Building a Human-Guided Security Analyst
Leveraging AI in the defensive/offensive space has taken off the past couple of years.
Leveraging AI in the defensive/offensive space has taken off the past couple of years.
If you’re an offensive or defensive engineer, a Windows endpoint engineer, or a Windows researcher, chances are you’ve come across Microsoft’s Threat-Intelligence ETW provider and understand the immense value it offers for telemetry.
As the reader, I’m sure you’re thinking — “oh great, another EDR internals or bypass post”.
Detection Engineers, Threat Hunters, and SOC Analysts all rely on one critical thing to do their jobs effectively — telemetry.
Microsoft continuously marches forward in providing us security events that uncovers activity that has long been used by attackers.
A walkthrough of how attackers use Windows Filtering Platform rules to silence EDR network traffic, and how products can protect themselves from this attack.
Not long ago I wrote a blog called Understanding ETW Patching where I walked through how ETW patching is a hyper-focused version of a function patch.
As of late, I have gotten a lot of questions around Event Tracing for Windows (ETW) patching, specifically the following questions:
A walk through process forking on Windows, how an adversary might leverage it, how you can identify this behavior, and where the 4688 event's metadata falls short.
A deep dive into adversarial LDAP tradecraft - exposing the telemetry available for LDAP activity and offering guidance on detecting malicious behavior, co-authored with the TrustedSec research team.
A walk through the client/server relationship at the heart of RPC and COM activity, and why that context is critical for detection engineering and incident response.
A deep dive into DLL hijacking - the basics, the different types of hijacks, and detection ideas for an attack that's historically been considered hard to spot.
Not all manifest-based Event Tracing for Windows (ETW) providers that are exposed through Windows are ingested into telemetry sensors/EDR’s.
Creating detections can be challenging.
Part 2 of the Defender's Guide series: how Windows services work under the hood, the metadata available for them, and how to spot service abuse in telemetry.
In part 1 of this series, I touched on how data is the foundation for defensive capabilities and the importance for defenders to understand where and how telemetry is being generated.
Data is the foundation by which defense is built upon.
A walk through Windows logon sessions - what they are, how they're tied to tokens and processes, and how defenders can use that link to tell the full story around suspicious activity.
A look at the Kerberos Relaying attacks made popular by KrbRelayUp - what's actually being relayed, what defenders should be looking for, and where existing detections might be missing the mark.
Detection engineers are frequently beset with the challenge of detecting a technique for which optics are poor, non-existent, or difficult to collect at scale.
This past week I briefly talked about Process Access data within a talk that Olaf and I gave at ATT&CKCON 3.0 (YouTube link isn’t live yet).
In this second installment of our Better know a data source series, we’re showcasing process integrity levels.
A look at how adversaries abuse Microsoft RPC (MSRPC) for privilege escalation - and where the detection opportunities sit, from PetitPotam to PrintNightmare.
A common issue within the investigation process is alert fatigue.
The term evasion is derived from the Latin word “evadere” which means — “To escape, to get away.” The DOD defines evasion as — “The process whereby isolated personnel avoid capture with the goal of successfully returning to areas under friendly control.”
Applying Capability Abstraction to scheduled tasks: walking the static and dynamic analysis used to map the behavior end-to-end and turn it into reliable detection.
A few months ago, Jared Atkinson released a blog post that introduced a detection engineering methodology he referred to as Capability Abstraction.
During Part 1 of this blog series: Engineering Process Injection Detections — Part 1: Research, I covered how you can maximize your detection efforts by following a concept outlined by Jared Atkinson: Capability Abstraction.
One thing that the SpecterOps defensive team likes to pride ourselves in, is our ability to manipulate data in a way to best help our client’s needs.
Often within detection engineering, we come across an attack technique that we want to create a detection for but don’t know where to start the process to effectively do so.
From a defensive perspective, one of the most dangerous things we apply to security is assumptions.
Around 3 months ago, a new attack technique was introduced to the InfoSec community known as “Process Reimaging.” This technique was released by the McAfee Security team in a blog titled — “In NTDLL I Trust — Process Reimaging and Endpoint Security Solution Bypass.” A few days after this attack technique was released, a co-worker and friend of mine — Dwight Hohnstein — came out with proof of concept code demonstrating this technique, which can be found on his GitHub.
As an adversary, one of the goals is to capture Domain Admin (DA) credentials, change/modify objects inside of Active Directory, and to be able to evade any detection systems that an environment may have in place.
Process Injection is a very common Defense Evasion/Privilege Escalation technique.
A walk through the IOC differences between Kerberoasting and AS-REP Roasting - two attack techniques that look similar at a glance but leave distinctly different traces.