Ethan Andrews
@eandrewsTrusted contributorCompletionist
13 followers311 downloads7,465 copies113 likes78,089 views
1,181 detections
Filters
Last updated
All Time
Detection languages
1,181
Categories
476
426
188
169
125
Platforms
820
300
136
59
49
Products / Services
504
78
66
60
56
MITRE Techniques
431
415
293
251
160
CVEs
11
5
5
5
4
Detects the creation of 'delete.me' files in critical Windows directories (specifically C:\Windows and SYSVOL paths). This behavior is associated with the Akira ransomware, which uses these marker files to coordinate domain-wide encryption tasks, often executed by the System account during the preparation phase of an attack.
Detects the creation or modification of a .vscode/tasks.json file, followed by the execution of a process (specifically node.exe) that targets a script file without a standard JavaScript extension (.js, .mjs, .cjs). This behavior is consistent with an adversary configuring a hidden VS Code task (e.g., using 'hide:true' and 'runOn:folderOpen') to silently execute arbitrary code or assets when a project folder is opened.
Detects persistence mechanisms associated with the UAC-0226 threat actor by identifying a LNK shortcut file located in the user Startup folder that executes a minimized cmd.exe process. This command shell subsequently launches a hidden PowerShell command that reads and executes code from C:\ProgramData\WC3 using Invoke-Expression (IEX). This activity is linked to the GIFTEDCROOK delivery chain.
Detects browser search traffic being hijacked to the typosquatted domain 'perplexity-ai.online' by the malicious Chromium extension 'Search for perplexity ai'. The rule monitors for network connections to the malicious domain and the presence of extension-related file artifacts on the system. The extension utilizes chrome_settings_overrides to redirect browser Omnibox queries and exfiltrate user search activity to attacker-controlled infrastructure.
Detects AI agent processes (e.g., claude, node) executing environment enumeration commands (env, printenv) following suspected prompt injection, combined with subsequent network traffic associated with credential exfiltration patterns.
Detects remote code execution exploitation targeting CVE-2026-20253 in Splunk. The rule monitors for unauthorized child processes, such as shells or scripting interpreters, being spawned by the Splunk daemon (splunkd). It also flags processes containing exploit-related strings in the command line as a secondary indicator.
Detects the execution of Node.js where the command line arguments reference a .woff2 font file. This behavior is indicative of an adversary attempting to execute a malicious payload disguised as a legitimate font file, a technique previously associated with the PolinRider campaign.
Detects Microsoft Exchange EWS (Exchange.asmx) requests containing a ReferenceAttachment call where the ProviderEndpointUrl parameter points to a non-Microsoft/external domain. This is consistent with SSRF exploitation techniques, such as those observed in CVE-2026-45504, where an attacker-controlled endpoint is provided to Exchange for fetching WOPI properties.
Detects the presence of malicious Chromium browser extension artifacts related to the 'Search for perplexity ai' extension. This extension abuses Manifest V3 declarativeNetRequest permissions (such as declarativeNetRequestFeedback and declarativeNetRequestWithHostAccess) to intercept and redirect network requests to the attacker-controlled domain 'perplexity-ai.online'. The rule specifically identifies the extension's rule resource files (e.g., perplexity-rules.json) and specific file paths associated with the extension ID.
Detects instances where the Claude Code AI agent or its subprocesses are used to establish persistence. This is achieved by monitoring for crontab modifications or direct writes to SSH authorized_keys files, which are common indicators of post-exploitation activity following an Axiom-style indirect prompt injection.
Page 29 of 119