Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
40,407 detections
Filters
Last updated
All Time
Detection languages
15,618
13,637
2,824
1,803
1,727
Contributors
7,678
5,999
5,315
4,504
4,206
Categories
18,287
9,897
3,918
3,796
3,721
Platforms
40,407
7,473
6,580
3,935
3,713
Products / Services
10,635
9,722
6,762
1,902
1,895
MITRE Techniques
13,801
13,234
8,133
6,014
4,458
CVEs
49
45
30
30
29
IDS Classtypes
214
56
43
23
19
IDS Protocols
183
172
20
16
8
The following analytic detects instances where Java, or Tomcat processes spawn a Linux shell, which may indicate exploitation attempts, such as those related to CVE-2021-44228 (Log4Shell).
This detection leverages Endpoint Detection and Response (EDR) telemetry, focusing on process names and parent-child process relationships.
This activity is significant as it can signify a compromised Java application, potentially leading to unauthorized shell access.
If confirmed malicious, attackers could execute arbitrary commands, escalate privileges, or maintain persistent access, posing a severe threat to the environment.
This detection leverages Endpoint Detection and Response (EDR) telemetry, focusing on process names and parent-child process relationships.
This activity is significant as it can signify a compromised Java application, potentially leading to unauthorized shell access.
If confirmed malicious, attackers could execute arbitrary commands, escalate privileges, or maintain persistent access, posing a severe threat to the environment.
Detects servers running vulnerable Java versions (JDK/JRE) with indicators of internet exposure, inbound connections, or active scanning, prioritizing risk levels for potential Java4Shell exploitation.
The following analytic identifies the execution of suspicious processes typically associated with WebShell activity on web servers.
It detects when processes like `cmd.exe`, `powershell.exe`, or `bash.exe` are spawned by web server processes such as `w3wp.exe` or `nginx.exe`.
This behavior is significant as it may indicate an adversary exploiting a web application vulnerability to install a WebShell, providing persistent access and command execution capabilities.
If confirmed malicious, this activity could allow attackers to maintain control over the compromised server, execute arbitrary commands, and potentially escalate privileges or exfiltrate sensitive data.
It detects when processes like `cmd.exe`, `powershell.exe`, or `bash.exe` are spawned by web server processes such as `w3wp.exe` or `nginx.exe`.
This behavior is significant as it may indicate an adversary exploiting a web application vulnerability to install a WebShell, providing persistent access and command execution capabilities.
If confirmed malicious, this activity could allow attackers to maintain control over the compromised server, execute arbitrary commands, and potentially escalate privileges or exfiltrate sensitive data.
Hunt: A Java-native callback, memshell beacon or reverse shell goes to a public IP that almost no other JVM contacts. Fleet prevalence replaces a baseline you would otherwise build by hand: destinations reached by 2 or fewer hosts, rarest first. This also covers Calif's client-side variant (a Java client connecting to a malicious server on an unusual port).
This rule performs a software inventory search to identify devices running specific versions of Nginx within a targeted range (1.9.5 up to 1.30.7, or 1.31.0 up to 1.31.9). This is typically used for vulnerability management and asset discovery.
hunt: JVM crash logs from failed heap grooming, which is Upwind's published indicator. The query returns any JVM that crashed two or more times, and any crash of a product Calif showed was exploitable pre-auth.
Detects java.exe or javaw.exe on Windows spawning a script host or download LOLBin directly, or spawning cmd.exe with download or discovery commands. Java4Shell (CVE-2026-114938) gives an unauthenticated remote attacker heap corruption in the OpenJDK AVX-512 Base64 decoder and Calif reached RCE in widely deployed Java products. Weaponized chains are not public, so this rule keys on the post-exploitation child process. Windows hosts need an AVX-512 VBMI CPU (Intel Ice Lake or newer, AMD Zen 4 or newer) and an affected OpenJDK build to be exposed.
This rule performs a hunting query for Java4Shell (CVE-2026-114938) by triaging Java versions from binary install paths and checking command-line flags for mitigation (such as -XX:-UseBASE64Intrinsics or -XX:UseAVX=2). It targets server and domain controller roles on Windows and Linux platforms.
Hunts naming artifacts reported in ARTEX driven intrusions: artex-probe-NNNNN probe files (5-digit incrementing), accounts containing ArtexProbe (often plaintext passwords, sometimes admin flags), art18- prefixed admin-command markers, and the C:\ProgramData\artx5v\ directory reported for VAPOR PANDA. A hit means an agentic harness touched the host. Forks and rebrands (ScopeWeaver, "ARTEX PRO") may change the names, so a miss proves nothing.
Previdian's honeypot also shows non-numeric suffixes (artex-probe-o20v0gt9). Coverage is narrow by design: forks and rebrands can rename artifacts, so a miss is not evidence of absence.
Previdian's honeypot also shows non-numeric suffixes (artex-probe-o20v0gt9). Coverage is narrow by design: forks and rebrands can rename artifacts, so a miss is not evidence of absence.
Identifies devices with a confirmed Microsoft Defender Vulnerability Management finding for Java4Shell (CVE-2026-114938).
Detects instances where the Nginx web server process initiates a shell or scripting interpreter, which may indicate web application compromise, remote code execution, or exploitation of a public-facing application.
The analytic associates each Zeek request containing a malformed Base64 value or a scripted decoder trigger with a nearby Base64-related HotSpot argument exception on the same server and Java PID.
Sysmon child process, file creation, or process termination events from that PID add context and can raise confidence.
Sysmon child process, file creation, or process termination events from that PID add context and can raise confidence.
Identifies servers running vulnerable or unknown versions of Java (JDK/JRE) with evidence of internet exposure, such as inbound public connections or internet scanning indicators, to prioritize patching for CVE-2026-114938.
Detects network-facing Java processes (or related applications like Tomcat, Jenkins) spawning common command shells, scripting engines, or network utilities, which often indicates remote code execution (RCE) exploitation such as Log4j.
Hunt: A JVM writes a JSP file (web shell) or writes to SSH, cron, systemd, ld.so.preload, rc.local, profile.d or the Windows Startup folder.
A JVM directly runs a command with reverse-shell, download and execute, base64 to shell or encoded PowerShell syntax. This catches RCE that succeeds on the first attempt, without a crash.
Detects the Nginx web server spawning suspicious child processes such as command shells, scripting interpreters, or reconnaissance tools, which often indicates exploitation of a web application vulnerability (RCE).
Detects Java processes spawning shells (sh, bash, cmd, powershell) or reconnaissance utilities (whoami, id, curl, wget), commonly associated with exploitation (such as Java4Shell / CVE-2026-114938), excluding known build tooling and routine host activity.
Detects outbound network connections to webhook.site from non-browser processes, which can indicate data exfiltration or automated callbacks by scripts or tools like Python.
Detects a Java service that is repeatedly relaunched on a host (consistent with JVM crashes from failed Java4Shell heap-corruption attempts), followed by the java process spawning a shell child, indicating successful pre-auth remote code execution.
Previdian's WebLogic honeypot captured ARTEX operators embedding a random marker in the first command they run (echo ARTEX479630; id; hostname; uname -r; nproc, base64-encoded in transit). If a payload like this ever executes on a real host, the marker or its encoded form lands in a process command line. Arm 1 matches the marker with its digit suffix, not the bare word. Arm 2 matches the base64 of echo ARTEX at all three alignments; case-sensitive, 13 characters, so collisions are negligible. Arm 3 matches the id; hostname; uname -r; nproc chain only from a server-side lineage, because that sequence alone is ordinary admin habit. Limits: the marker is attacker-controlled and spoofable, a nearby operator may drop it, and Previdian confirmed no host-level execution (the honeypot returned canned output), so this is a single-source tripwire, not a proven artifact.
Page 1 of 1925