Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
3,624 detections
Filters
Last updated
All Time
Detection languages
1,749
374
349
273
255
Contributors
908
661
593
359
164
Categories
1,486
1,150
779
525
515
Platforms
39,894
7,228
6,523
3,864
3,624
Products / Services
802
512
432
340
297
MITRE Techniques
1,177
572
523
506
437
CVEs
17
16
9
8
7
IDS Classtypes
41
15
10
6
5
IDS Protocols
47
24
3
2
1
Detects the unauthorized injection of the 'GEMINI_TRUST_WORKSPACE=true' environment variable via process command lines or shell environments. This rule identifies potential exploitation attempts related to CVE-2026-12537 by flagging instances where this variable is set outside of defined, trusted CI/CD runner environments.
Detects Python processes that combine potentially dangerous pickle deserialization (pickle.loads) with network-fetching libraries (urllib, requests, socket) and decoding methods (b64decode, fromhex). This pattern is consistent with backdoors or remote access tools (RATs) that fetch encoded payloads over the network and execute them directly in memory via deserialization.
Detects Python processes executing complex, multi-layered decoding chains (Base64, ROT13, and Hex) in their command line arguments. This pattern is often used by malicious AI skill backdoors, such as ClawHub, to obfuscate payload execution and evade signature-based detection.
Detects outbound network connections initiated by common AI agent runtime processes (Python, Node.js, OpenClaw) to external destinations that are not identified as known AI service providers, specifically focusing on non-standard ports or the use of WebSocket protocols. This behavior may indicate an AI agent being used for unauthorized data exfiltration, command-and-control communication, or interaction with an unverified external service.
Detects high volumes of file operations (creation, rename, or modification) where files are given extensions typically associated with ransomware (e.g., .locked, .encrypted). The rule aggregates activity by process and endpoint and triggers an alert when the number of such operations exceeds a defined threshold, while excluding known legitimate backup and security software.
Detects recursive file and directory discovery activity performed by common command-line utilities (dir, gci, find, ls) when initiated by suspicious parent processes such as interpreters (python, node, wscript) or shells, excluding processes signed by known trusted publishers.
Detects recursive file and directory discovery activity performed by common command-line utilities (dir, gci, find, ls) when initiated by suspicious parent processes such as interpreters (python, node, wscript) or shells, excluding processes signed by known trusted publishers.
Detects anomalous authentication patterns where a single source IP address exhibits multiple failed login attempts followed by at least one successful login within a 10-minute time window, indicative of credential stuffing or password spraying.
Detects unauthorized processes, such as shells, scripting engines, or command-line utilities, that attempt to display or access environment variables containing sensitive cloud credential patterns (e.g., AWS_ACCESS, API_KEY). This behavior often indicates an attempt to steal cloud provider credentials from the host environment.
Detects the execution of a custom Mach-O binary designed to fake system password prompts (UNK_DeadDrop). This activity is indicative of credential phishing attempts on macOS systems.
This rule monitors DNS queries for potential exfiltration or C2 communication channels. It flags DNS queries that exceed 100 characters in length or contain labels with at least 30 alphanumeric characters, excluding activity from common web browsers.
Detects the download of a standalone CPython 3.10.18 interpreter from github.com/astral-sh/python-build-standalone using curl, followed by the extraction of the compressed archive using tar. This behavior is associated with the macOS.Gaslight malware's secondary collection module installer, which likely prepares a localized Python environment to execute further malicious tasks.
Detects the AMOS infostealer staging collected system data by using the macOS ditto utility to compress information into a hidden .hipr directory. The rule identifies the execution of ditto with .hipr in the command line in conjunction with the creation of User_Name.txt and System_Information.txt within that same directory.
Detects the AMOS infostealer attempting to access and enumerate browser extension directories on macOS to harvest cryptocurrency wallet data from Chrome and Firefox profiles.
Detects potential information stealing activity, specifically the Atomic macOS Stealer (AMOS) malware, by monitoring unauthorized processes accessing Safari cookies and Apple Notes databases, excluding legitimate applications.
Detects potential activity associated with the AMOS infostealer, where common terminal commands (cp or mv) are used to manipulate or replace legitimate Ledger Live or Trezor Suite cryptocurrency wallet applications located in the /Applications directory on macOS.
Detects unauthorized processes reading, copying, or renaming Kerberos ticket cache files (ccache) located in common temporary directories. This behavior is often associated with credential theft and potential Pass the Ticket attacks where adversaries attempt to extract session credentials to perform lateral movement or privilege escalation.
Detects modification of scripts in extension folders.
Detects Python executing a web-administration exploit with valid credential input and explicit two-factor bypass logic, matching the AI-assisted semantic trust-flaw exploitation described by GTIG.
Detects Python or a shell replacing the installed Chrome build with an older package before installing a wallet extension that requires Manifest V2 capabilities. This activity is associated with the InvisibleFerret malware.
Detects instances where VS Code extension-related processes (such as amazonq-lsp or extensionHost) spawn shell processes (bash/sh) that subsequently initiate network activity via curl or wget. This pattern is indicative of potential exploitation where malicious workspace configuration files (e.g., .amazonq/mcp.json) are used to execute shell commands with unauthorized access to cloud credentials.
Page 128 of 173