Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
2,588 detections
Filters
Last updated
All Time
Detection languages
1,194
459
310
204
161
Contributors
487
455
323
171
123
Categories
1,884
937
415
400
360
Platforms
1,349
560
343
329
257
Products / Services
601
314
306
286
271
MITRE Techniques
18,127
15,648
12,596
8,407
6,231
2,588
CVEs
9
5
5
5
2
IDS Classtypes
142
96
49
11
6
IDS Protocols
177
49
44
17
4
This rule detects potential data exfiltration via email by identifying mass near-identical background-image URLs in the body of an email. The rule looks for a high volume (8+) of distinct URLs that contain short parameters (indicative of encoded or smuggled data) within a single email. It further refines the detection by checking for the presence of specific CSS-like syntax often used in CSS-injection keylogging techniques to force the browser or email client to send background-image requests.
This rule detects potential credential or sensitive file theft from VS Code extensions or configuration folders, followed closely by network exfiltration to the Telegram API. It specifically monitors file access to common paths (.ssh/id_rsa, .aws/credentials, .netrc, Local Storage, etc.) by VS Code processes, correlated with non-Telegram processes establishing network connections to api.telegram.org within a 10-minute window.
Detects a curl request to a tunnel-exposed /api/summary endpoint piped through jq extracting a 'spend' field, indicating automated exfiltration of financial data via a coding-agent-established tunnel.
This rule monitors GCP VPC Flow logs for potential command and control (C2) or data exfiltration activity by identifying internal-to-external network connections that exhibit high novelty. It calculates the percentage of new destination IP addresses contacted by internal source IPs within a single day. Alerts are triggered when a source IP initiates connections to at least 20 unique external destinations, where at least 20% of those destinations have never been contacted by that source in the previous day. This behavioral analysis is designed to detect reconnaissance, scanning, or large-scale data exfiltration patterns.
Detects macOS infostealer malware using the curl utility to exfiltrate archived data (zip, tar, tgz, etc.) via multipart POST requests to specific known staging endpoints associated with the ClickFix campaign.
Detects outbound network connections and DNS queries to a maintained list of known UNC6671 extortion-related domain names. The rule utilizes a temporal 'recency' window (aged IOCs) to filter out stale indicators, mitigating false positives caused by sinkholed or reassigned infrastructure.
This rule detects suspicious activity originating from the Node.js runtime environment. It specifically monitors for 'node.exe' or 'node' processes accessing sensitive credential-related files (such as .npmrc, .netrc, .git-credentials, or id_rsa) followed closely (within 5 minutes) by network connections to public IP addresses. This behavior is indicative of an npm-based malware or malicious package attempting to harvest locally stored credentials and exfiltrate them to an external, potentially attacker-controlled server.
This one's the interesting bit they're using Ethereum RPC calls as a C2channel and dumping stolen secrets to attacker-owned public repos tagged "Shai-Hulud: Here We Go Again". Also flags Claude/VS Code config file writes dropped by the worm for lateral persistence across dev machines.
This rule detects potential data exfiltration via email by identifying mass near-identical background-image URLs in the body of an email. The rule looks for a high volume (8+) of distinct URLs that contain short parameters (indicative of encoded or smuggled data) within a single email. It further refines the detection by checking for the presence of specific CSS-like syntax often used in CSS-injection keylogging techniques to force the browser or email client to send background-image requests.
Detects the UNC5142 DeviceManager RAT's DNS-based C2 channel: raw malformed/non-resolver UDP/53 packets and Base64-chunked o-/e-/r- prefixed query names to 91.92.240.100, plus microsoft.com-suffixed decoy queries.
This rule detects potential credential or sensitive file theft from VS Code extensions or configuration folders, followed closely by network exfiltration to the Telegram API. It specifically monitors file access to common paths (.ssh/id_rsa, .aws/credentials, .netrc, Local Storage, etc.) by VS Code processes, correlated with non-Telegram processes establishing network connections to api.telegram.org within a 10-minute window.
This rule detects potentially suspicious data exfiltration activities by identifying a pattern where a user authenticates or interacts with the Microsoft Purview service, followed closely (within 15 minutes) by a connection to an external domain that the organization has not established a historical communication pattern with in the last 30 days. This behavior may indicate an adversary is attempting to exfiltrate session tokens or sensitive data to an external server.
This rule detects high-volume outbound network traffic from a Google Cloud Platform (GCP) source IP to external (non-private/non-internal) destinations. It establishes a baseline of daily outbound traffic per source IP over time and triggers an alert if the current daily outbound volume exceeds 500MB and shows a significant increase (more than 3x the established average) compared to previous days. This behavior is often indicative of unauthorized data exfiltration or abnormal network activity.
This rule detects the creation of Google Cloud Platform (GCP) persistent disk snapshots. Monitoring for disk snapshot creation is critical as adversaries may use this functionality to create backups of sensitive virtual machine disks for exfiltration, staging, or offline analysis, effectively bypassing traditional network-based egress controls.
Detects a curl request to a tunnel-exposed /api/summary endpoint piped through jq extracting a 'spend' field, indicating automated exfiltration of financial data via a coding-agent-established tunnel.
Detects instances where a single user account performs a large volume of file downloads (50 or more unique files) within a 30-minute window from Office 365 services. This behavior may indicate unauthorized data exfiltration or mass collection of sensitive information.
UNC6671 - Outbound connection to known automated SaaS data exfiltration IP, excluding maintained allowlist of legitimate backup/sync/CDN and sanctioned bulk-transfer destinations
This rule detects suspicious activity originating from the Node.js runtime environment. It specifically monitors for 'node.exe' or 'node' processes accessing sensitive credential-related files (such as .npmrc, .netrc, .git-credentials, or id_rsa) followed closely (within 5 minutes) by network connections to public IP addresses. This behavior is indicative of an npm-based malware or malicious package attempting to harvest locally stored credentials and exfiltrate them to an external, potentially attacker-controlled server.
Detects the execution of the rclone utility followed by successful outbound network connections occurring outside of standard business hours (8:00 PM to 6:00 AM). Rclone is a legitimate cloud storage synchronization tool frequently abused by adversaries for data exfiltration. This pattern identifies potential unauthorized data staging and egress to remote cloud storage services during times when activity is less likely to be monitored.
Detects anomalous, high-volume access, download, or synchronization activity in SharePoint or OneDrive within a one-hour window. The rule focuses on non-interactive user agents typically used by scripting libraries (e.g., Python requests, PowerShell, Go-http-client) and flags activity that exceeds an individual user's 21-day rolling baseline, suggesting potential automated data exfiltration or mass collection.
This rule detects potential data exfiltration by monitoring for anomalous user behavior across integrated SaaS applications. It flags accounts that perform high-frequency read, search, or view activities across three or more monitored SaaS platforms (Confluence, Jira, Slack, Google Workspace, Microsoft 365, or Bitbucket) within a 5-minute window, followed immediately by an upload or post activity to an external, non-integrated domain.
Page 40 of 124