Resource index

Research

Original incident-response research, field evidence and analysis.

Published
62 entries
Coverage
Complete index
Order
Featured, then newest
  1. A blue mesa tower with a lit window rising from dark ground
    research01

    Could You Investigate Misuse of Claude in Your Own Environment?

    Can you investigate Claude misuse? Only if evidence collection was configured before the incident. See what Claude Enterprise, Team, Code, and Cowork actually record.

  2. A narrow street canyon between blue concrete blocks
    research02

    A GitHub token was stolen. Now what?

    A stolen GitHub Personal Access Token exposes every repo an identity can reach, and the secrets inside. How to scope the blast radius, investigate, and harden fast.

  3. Power lines disappearing into fog across a desert at night
    research03

    Forensic Frustrations in Google Workspace

    Licensing gaps, log delay, and OAuth correlation limits break IR in Google Workspace. Two real cases on what you can and can't investigate, plus fixes.

  4. A row of doorways with one lit warm and a crooked copy lit a colder shade
    research04

    Borrowed Trust, Badly Done

    A legitimate mailbox sends an AI-written invoice lure that chains AWS S3 to Cloudflare to a fake Google login. Analysis, IOCs, and how to defend.

  5. research05

    Frontier Forensics: Open AI

    Explore what forensic evidence OpenAI leaves behind, including logs, retention, auditability, and provenance across AI systems.

  6. research06

    Device code phishing in Entra ID: what the sign-in logs actually show

    Device code phishing leaves a distinctive pair in Entra ID sign-in logs — an authentication with no interactive client, then a redemption from elsewhere. Here is the query and why Conditional Access missed it.

  7. Silhouetted team planning around a lit table inside a glass pavilion
    research07

    The Silent SaaS Threat: Part 3 – The Response Playbook

    A five-step Entra ID response playbook for malicious OAuth apps: detect rogue consents, assess blast radius via Graph permissions, contain, analyze, eradicate.

  8. Rows of small blue concrete modules beneath a round building, one lit amber
    research08

    Incident Response in Kubernetes (AKS)

    Maps AKS log sources for IR: why Kubernetes Audit and Guard logs are off by default, Automatic mode's forensic limits, and detection via Entra ID/NSG flow logs.

  9. An aerial view of four round buildings joined by lit walkways at night
    research09

    Incident Response in Kubernetes (EKS)

    Details EKS log sources for IR: API server, audit and authenticator logs, CloudTrail/VPC Flow Log correlation, and pod forensics for fileless malware.

  10. The final span of a bridge lowered into place between two buildings
    research10

    The Missing Link: AADGraphActivityLogs Finally Arrives

    Entra ID's new AADGraphActivityLogs table closes the legacy Azure AD Graph API blind spot, exposing activity from tools like AADInternals and ROADtools.

  11. research11

    Atlas Lion: Tactics, History & Defenses

    Threat profile of Atlas Lion (Storm-0539): gift-card fraud via AiTM phishing and rogue device registration in Azure/M365, with MFA and Conditional Access.

  12. research12

    Incident Response in the Neocloud - Lambda Cloud (Part 2)

    Closing our Neocloud IR series, we map Lambda Cloud's identity model, structure, and audit logging, detailing what evidence responders can pull from lambda.ai.

  13. research13

    The Silent SaaS Threat: Part 2 – The Proactive Playbook

    Part 2 of this SaaS series covers hardening Entra ID against OAuth abuse: admin consent workflows, delegated vs application permissions, credential hygiene.

  14. research14

    The Poisoned Pipeline: Axios Supply Chain Attack

    Analysis of the Axios npm compromise: trojanized versions 1.14.1 and 0.30.4 deployed a cross-platform RAT via CI/CD pipelines, with IOCs and remediation steps.

  15. research15

    The Invisible Architecture of Modern Phishing

    How a 2026 AiTM campaign chained ConvertKit email delivery, a compromised small-business page, and Akamai/Linode-hosted scripts to harvest M365 sessions.

  16. research16

    The Silent SaaS Threat: The 5 Questions

    A five-question checklist for gauging OAuth and SaaS sprawl risk in Entra, drawing on patterns from the Salesloft Drift and Midnight Blizzard intrusions.

  17. research17

    Incident Response in the Neocloud - Nebius (Part I)

    Part 1 of our Neocloud IR series maps Nebius's AWS-like IAM and logging architecture, plus the open-source Invictus-Nebius script for forensic data acquisition.

  18. A building standing intact at night as a storm front passes, one window lit
    research18

    The story of how we almost got hacked

    How Invictus tracked a BEC actor's AiTM phishing campaign built on EvilProxy, from a WeTransfer HTML lure to infrastructure uncovered via passive DNS pivoting.

  19. research19

    Profiling Silk Typhoon: Tactics, History & Defenses

    Threat actor profile of Silk Typhoon (HAFNIUM): Exchange Server zero-days, the CVE-2025-3928 Commvault Metallic exploit on Azure, and OAuth/Graph/EWS abuse.

  20. research20

    Anatomy of a BEC in 2025

    A Microsoft 365 BEC case study: a SharePoint-hosted PDF phishing lure led to mass outbound emails, tracing 2025 threat-actor tactics and detection gaps.

  21. research21

    A Candid Perspective on the Cloud Threat Landscape: A Recap from fwd:cloudsec EU

    Recap of Invictus's fwd:cloudsec EU talk: cloud threat actor trends, attack patterns, and a catalog of 70+ cloud TTPs for defenders and detection engineers.

  22. research22

    Profiling Sea Turtle: Tactics, History & Defenses

    Threat actor profile of Sea Turtle: DNS hijacking, zero-day exploitation, and AWS/Azure credential theft via security-group edits enabling SSH access.

  23. research23

    Introducing KubeForenSys: A Kubernetes Forensic Collection Framework for Azure Kubernetes Service (AKS)

    KubeForenSys is Invictus's open-source AKS forensic tool, unifying Kubernetes API events, RBAC changes, and pod activity into a single incident timeline.

  24. research24

    Profiling TraderTraitor: Tactics, History & Defenses

    Threat actor profile of TraderTraitor: supply chain compromise and credential theft against AWS-hosted crypto platforms, tied to the Ronin and Bybit heists.

  25. research25

    Profiling Laundry Bear: Tactics, History & Defenses

    Threat actor profile of Laundry Bear (Void Blizzard): pass-the-cookie attacks, Evilginx AitM phishing, and password spraying against M365 and Entra ID.

  26. research26

    Profiling JavaGhost: Tactics, History & Defenses

    Threat actor profile of JavaGhost: stolen IAM keys abused for SES phishing, CloudTrail evasion via GetFederationToken, and API Gateway-Lambda persistence.

  27. research27

    Cloud Incident Readiness: Critical infrastructure for cloud incident response

    Part three of a cloud readiness series on IR infrastructure: Azure Log Analytics, Sentinel, and storage accounts set up for root cause analysis and containment.

  28. research28

    Cloud Heavy, Hybrid Ready: Lessons from BlackBasta and Scattered Spider

    Analysis of leaked BlackBasta chat logs and Scattered Spider TTPs: RaaS techniques for hybrid environments vs. cloud-native identity attacks, and IR takeaways.

  29. research29

    Cloud Incident Readiness: Key logs for cloud incidents

    Part two of a cloud readiness series ranking must-have, should-have, and nice-to-have logs across Azure, AWS, and GCP for investigating cloud incidents.

  30. research30

    Deep Dive: Forensic Analysis of eM Client

    Forensic analysis of eM Client abuse in M365 BEC cases: registration, sign-in and forwarding-rule artifacts in Entra ID and Unified Audit Logs, keyed to app ID.

  31. research31

    Locked Out, Dropboxed In: When BEC threats innovate

    An AiTM case where the threat actor abused eM Client's IMAP.AccessAsUser.All for persistence, then registered Dropbox and WeTransfer accounts to keep phishing.

  32. research32

    Cloud Incident Readiness: A Practical Playbook for CISOs and Teams

    Part one of a cloud readiness series: practical steps for provisioning emergency access across Azure, AWS, and GCP ahead of an IR engagement starting.

  33. research33

    Responding to Adversary in the Middle attacks

    How Adversary-in-the-Middle phishing bypasses MFA in Microsoft 365, with log-based detection, response steps, and prevention guidance from BEC casework.

  34. research34

    This Is How We Do It: Resolving Microsoft Cloud Incidents Like Pros

    How Invictus streamlined Microsoft 365 and Azure incident response using a multi-tenant Graph API app for one-click log acquisition and automated processing.

  35. research35

    Cloud native incident response in AWS - Part II

    Part two of our AWS IR series: using Athena and Glue to query CloudTrail, VPC Flow, S3, ELB and CloudFront logs, plus partitioning and nested field parsing.

  36. research36

    Cloud native incident response in AWS - Part I

    Part one of a series on native AWS incident response methodology: the investigative approach, log sources, and analysis tooling drawn from real AWS compromises.

  37. research37

    A deep dive into Entra ID Identity Protection for Incident Response

    A deep dive into Microsoft Entra ID Protection's risky user, sign-in, and workload identity reports, and how to use that data during incident response.

  38. research38

    The Azure log you probably didn’t know existed

    An overview of Azure's Directory Activity Log, the tenant-scoped log recording Global Administrator permission changes, useful in incident investigations.

  39. research39

    The Microsoft Graph for Unified Audit Log acquisitions is here!

    Microsoft's auditLogQuery Graph API now supports Unified Audit Log acquisition, with the Microsoft-Extractor-Suite automating the query and retrieval workflow.

  40. research40

    What DFIR experts need to know about the current state of the Unified Audit Log

    Post-Storm-0558 changes to the M365 Unified Audit Log: extended 180-day retention, new Premium events like MailItemsAccessed, and Graph API acquisition.

  41. research41

    The curious case of DangerDev@protonmail.me

    An AWS incident walkthrough: a threat actor abusing Simple Email Service sending limits after credential compromise, with a full MITRE-mapped timeline.

  42. research42

    Do not use the Get-MgAuditLogSignIn for your investigations!

    Get-MgAuditLogSignIn omits the Authentication Protocol field needed to spot Device Code phishing. Get-MgBetaAuditLogSignIn returns complete Entra sign-in data.

  43. Power pylons crossing a desert mesa landscape by day
    research43

    AWS CloudTrail cheat sheet

    A cheat sheet of high-signal AWS CloudTrail event names for faster incident triage, built from real IR engagements to help responders spot malicious activity.

  44. research44

    Ransomware in the cloud

    A ransomware case in AWS traced to an exposed long-term IAM key, with SES quota-call reconnaissance and a CloudTrail 90-day retention gap limiting visibility.

  45. research45

    A Defenders Guide to GraphRunner — Part II

    Part II of the GraphRunner defender's guide: pillage modules for SharePoint, OneDrive, mailbox, and Teams data theft, plus abuse prevention guidance.

  46. research46

    A Defenders Guide to GraphRunner — Part I

    Part I of a defender's guide to GraphRunner, the Entra ID post-exploitation tool: authentication, recon, and persistence modules, with IoCs and KQL queries.

  47. research47

    Everything you need to know about the MicrosoftGraphActivityLogs

    How Microsoft Graph Activity Logs close the Graph API auditing gaps exposed by Storm-0558 and SolarWinds, covering licensing, ingestion and KQL analysis.

  48. research48

    Automated AWS Incident Response — The next episode

    Feature update for Invictus-AWS, our open-source AWS IR tool: automated CloudTrail acquisition, service enumeration, and analysis for faster investigations.

  49. research49

    5 Tips to prevent or limit the impact of an incident in Azure

    Five low-cost Azure controls to limit incident impact: budget alerts, app registration limits, subscription policies, audit logging, collaboration controls.

  50. research50

    Automated First-Response in AWS using Sigma and Athena

    Research testing 32 AWS attack techniques via Stratus Red Team: Sigma rules queried via Athena outperformed GuardDuty at detecting malicious CloudTrail events.

  51. research51

    Importing Windows Event Log files into Splunk

    Configures Splunk to parse exported .evtx files via the preprocess-winevt sourcetype, avoiding the merged-event errors seen with the default wineventlog type.

  52. research52

    The Evolution of Business Email Compromise

    Business Email Compromise TTPs from a Microsoft 365 case: spearphishing link access, internal lateral phishing, and evidence collection via Extractor Suite.

  53. research53

    Welcome 👋 Microsoft Extractor Suite

    Introducing the Microsoft Extractor Suite: a modular PowerShell tool for acquiring Unified Audit Logs, Azure AD sign-in/audit logs, and OAuth app data.

  54. research54

    Responding to an attack in AWS

    Final part of an AWS investigation: CloudTrail analysis of an unauthorized IAM user, EC2 Session Manager lateral movement, and database exfiltration to S3.

  55. research55

    Incident Response in Azure

    Details how Entra ID cross-tenant sync can be abused for persistence, which Audit Log activities expose the attacker tenant ID, and a KQL hunting query.

  56. research56

    Set up Splunk for Incident Response in GCP in 15 minutes..

    A walkthrough for deploying a Windows Splunk instance on Google Cloud Platform in 15 minutes, for ingesting and analysing incident response log evidence.

  57. research57

    Responding to macOS attacks — Part II

    Second in the macOS attack series: Persistence and Privilege Escalation via Launch Agents and Launch Daemons, mapped to MITRE ATT&CK with forensic artefacts.

  58. research58

    Top 5 Windows Events for Incident Response

    A ranked walkthrough of five Windows Event IDs incident responders rely on most, including 1149 for RDP auth, 4698 for scheduled tasks, and 4624 for logons.

  59. research59

    Responding to a Cobalt Strike attack — Part III

    Part three covers memory forensics for in-memory Cobalt Strike payloads: full memory image versus process dump analysis of PowerShell-injected beacons.

  60. research60

    Responding to a Cobalt Strike attack — Part II

    Part two on Cobalt Strike delivery mechanisms: scripted web drive-by payloads, HTML applications, Office macros, and static analysis of a beacon executable.

  61. research61

    Responding to a Cobalt Strike attack — Part I

    Part one of a Cobalt Strike IR series: OSINT and threat intelligence methods to identify beacon infrastructure and C2 activity during an active investigation.

  62. research62

    Responding to macOS attacks — Part I

    First in a series mapping macOS attacks to MITRE ATT&CK: Initial Access and Execution techniques with artefacts from real malware and adversary emulation.