Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
180 detections
Filters
Last updated
All Time
Detection languages
91
29
14
14
11
Contributors
66
42
28
6
6
Categories
125
119
42
35
32
Platforms
120
95
36
27
8
Products / Services
85
38
27
21
18
MITRE Techniques
18,108
15,613
12,573
8,384
6,203
CVEs
4
3
3
2
2
IDS Classtypes
3
1
IDS Protocols
2
2
Detects various techniques used by an attacker to escape a container environment or gain unauthorized access to host-level resources. This includes the execution of namespace manipulation tools (nsenter, unshare, chroot), unauthorized access to the docker.sock Unix socket, mounting of host filesystems from within a container, and direct attempts to access the host's /proc/1 namespace files.
This rule detects suspicious activity patterns on Linux systems that may indicate privilege escalation or container escape attempts. It monitors three main behaviors: the execution of network configuration tools (tc, insmod, modprobe) for kernel-level manipulations (pedit/actions), the use of 'unshare' to create isolated namespaces, and typical privilege escalation commands (su, id, whoami, bash) executed by non-root users. The rule correlates these events and alerts based on the combination and intensity of these behaviors, flagging high-severity activity when kernel-level tools are combined with namespace manipulation or privilege elevation.
This rule detects suspicious manipulation of Linux namespaces using the 'unshare' utility or manual modification of user ID mapping files ('uid_map', 'gid_map', 'setgroups') in '/proc/self/'. Such activities are characteristic of container escape attempts or privilege escalation, where an attacker seeks to gain root access on the host or map unprivileged users to privileged IDs within a new namespace.
Detects potential container breakout attempts on Windows systems by monitoring for the launch of privileged containers with sensitive host paths mounted, the execution of host-level processes directly spawned by container runtime binaries (docker.exe, containerd.exe), and direct access to sensitive host filesystems (e.g., C:\windows, C:\etc, C:\programdata) by container runtimes.
This rule detects suspicious activity involving Docker containers, specifically focusing on path traversal attempts via the 'docker cp' command, execution of potential proof-of-concept scripts, or the creation of specific marker files indicating attempts to interact with or escape from the containerized environment to the underlying host.
Detects attempts to perform a container escape using path traversal via the 'docker cp' command. This rule monitors for malicious directory traversal sequences, the creation of specific artifacts used in known PoC exploits for container escapes, and the subsequent execution of scripts or shell interpreters by container management processes.
This rule detects potentially malicious attempts to escape a container or manipulate the host filesystem using the 'docker cp' command. It identifies commands that contain directory traversal patterns (e.g., '../', '..\') or direct references to the host filesystem via '/proc/self/root'.
Detects attempts to perform a container escape using the 'docker cp' command, where a maliciously crafted source path with traversal sequences allows the container daemon to write files outside of the container boundary. The rule monitors for suspicious 'docker cp' command executions and the creation of artifacts associated with known path traversal escape techniques in the host filesystem by the Docker daemon.
Detects the use of 'docker cp' to copy files to or from sensitive host paths, such as cron, password files, SSH keys, or system directories, which often indicates attempts to escape container isolation or establish persistence on the host.
Detects attempts to modify or bypass AppArmor security profiles using utilities such as aa-complain, aa-disable, or apparmor_parser, correlated with process execution involving user namespace unsharing (e.g., via unshare or CLONE_NEWUSER/CLONE_NEWNET). This combination suggests an attacker attempting to weaken host security controls or perform a container breakout or privilege escalation by manipulating kernel-level namespaces and security policies.
Detects the use of the 'unshare' utility to create new namespaces on Linux systems. Adversaries often use 'unshare' to perform container escapes, escalate privileges, or isolate processes from host-level monitoring. This rule specifically filters out legitimate administrative or container management processes such as systemd, containerd, and runc to reduce noise.
Detects the creation of Kubernetes pods where the 'hostPID' or 'hostNetwork' settings are enabled. These configurations allow a pod to share the host's process namespace or network stack, respectively, which are common vectors for container escapes and host-level privilege escalation.
Detects the use of 'docker cp' commands containing path traversal sequences (e.g., '..') in the destination argument. This behavior is indicative of an attempt to escape the container environment and write files to the host filesystem, potentially exploiting the 'docker-cp-copyout' vulnerability.
Detects the execution of container-related binaries (act_runner, docker, dockerd) using sensitive command-line flags that indicate potential container escape or high-privilege configuration, such as mounting the host namespace, running in privileged mode, or leaking CI/CD tokens.
Detects Gitea Act Runner spawning Docker containers with high-risk security options (e.g., --privileged, --pid=host, --ipc=host, --cap-add=ALL, or security profile unconfined). These flags are often used to attempt a container escape, allowing a process running within a Gitea CI/CD workflow to gain unauthorized access to the underlying host or other container resources.
Detects anomalous process execution and filesystem access patterns originating from Gitea Act Runner that indicate an attempted container-to-host escape. This includes the injection of dangerous container flags (e.g., --pid=host, --privileged, --cap-add=ALL) into workflow commands, the execution of nsenter, or direct access to the host's filesystem structures (e.g., /host or /proc/1/root).
Detects the Gitea Act Runner CI/CD tool executing Docker containers with dangerous security flags (such as --privileged, --pid=host, or --cap-add=ALL) or spawning nsenter to interact with host namespaces. This behavior indicates a potential attempt to escape the container boundary and gain unauthorized root-level access on the underlying CI/CD runner host.
Detects anomalous process execution patterns associated with the Gitea Act Runner that indicate attempts to break out of container isolation. This includes the use of sensitive Docker flags like --privileged, --pid=host, or --ipc=host, the mounting of the Docker daemon socket, or the execution of nsenter to gain access to the host namespace, which are indicative of container escape techniques.
This rule detects suspicious activity involving container runtimes (e.g., Docker) that indicates potential container escape attempts or unauthorized host interaction. It flags scenarios where container processes spawn administrative shells (cmd.exe, powershell.exe) with SYSTEM privileges, instances of privileged container creation, or direct access to physical drives from container management tools, which are common indicators of host filesystem manipulation.
This rule detects suspicious activity involving container runtimes (e.g., Docker) that indicates potential container escape attempts or unauthorized host interaction. It flags scenarios where container processes spawn administrative shells (cmd.exe, powershell.exe) with SYSTEM privileges, instances of privileged container creation, or direct access to physical drives from container management tools, which are common indicators of host filesystem manipulation.
Detects the creation of Kubernetes pods with dangerous security settings that could lead to container escape or privilege escalation. This includes configurations such as hostNetwork enabled, allowPrivilegeEscalation set to true, or running as root (runAsUser = 0). The rule specifically monitors 'io.k8s.core.v1.pods.create' operations in cloud audit logs.
Page 7 of 9