Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
565 detections
Filters
Last updated
All Time
Detection languages
464
49
14
12
11
Contributors
217
119
88
54
41
Categories
20,443
11,831
5,875
5,151
4,864
565
Platforms
495
55
21
17
14
Products / Services
468
195
64
23
14
MITRE Techniques
180
93
66
63
60
CVEs
4
4
4
4
3
IDS Classtypes
303
37
23
23
21
IDS Protocols
449
3
Detects anomalous network traffic patterns characteristic of SectopRAT, specifically utilizing Binance Smart Chain (BSC) RPC nodes to facilitate fallback Command and Control (C2) communication. The rule monitors for POST requests to known BSC data seed domains containing 'eth_call' in the body, specifically looking for START and FINISH markers in the response body.
Detects a sequence of potentially malicious activity targeting LLM-related API endpoints. The rule identifies a client IP performing rapid, distinct endpoint discovery, followed by a POST request to sensitive paths, and concluding with a spike in 5xx server-side errors, indicating potential exploitation attempts or fuzzing against Large Language Model interfaces.
Detects potential credential stuffing activity by monitoring for a high volume of HTTP 401 (Unauthorized) or 403 (Forbidden) responses originating from a single source IP address within a 60-second window, specifically targeting URIs containing the path 'login'.
Detects network traffic associated with Adversary-in-the-Middle (AiTM) phishing kits that impersonate Microsoft and Google login pages. The rules identify the presence of specific HTML elements (like Cloudflare Turnstile anti-bot gates) and authentication session cookie headers served from non-authorized/non-official domains, indicating a reverse-proxy phishing attack.
Detects a sequence of potentially malicious activity targeting LLM-related API endpoints. The rule identifies a client IP performing rapid, distinct endpoint discovery, followed by a POST request to sensitive paths, and concluding with a spike in 5xx server-side errors, indicating potential exploitation attempts or fuzzing against Large Language Model interfaces.
Detects a sequence of potentially malicious activity targeting LLM-related API endpoints. The rule identifies a client IP performing rapid, distinct endpoint discovery, followed by a POST request to sensitive paths, and concluding with a spike in 5xx server-side errors, indicating potential exploitation attempts or fuzzing against Large Language Model interfaces.
Detects web servers serving significantly different content sizes to AI crawlers (e.g., ChatGPT-User, ClaudeBot) compared to human web browsers for the same URL, which is a technique used to serve benign content to automated scrapers while delivering malicious or deceptive content to human users.
Detects HTTP traffic associated with the NeedyMantis malware, specifically identifying initial beacons to a known C2 domain using a specific User-Agent, followed by command-and-control response headers that suggest the exfiltration of reconnaissance data via the Set-Cookie field.
Detects network patterns associated with NeedyMantis-style command-and-control (C2) activity. This includes specific HTTP GET requests with hardcoded User-Agent strings, cookie-based binary data transfer in server responses, and WebSocket communication attempting to establish binary protocol channels.
Detects HTTP response bodies containing indicators of 'ClickFix' social engineering lures. The rule monitors for JavaScript code designed to manipulate the user's clipboard (execCommand('copy') or clipboard.writeText) in conjunction with instructions for the user to execute the Windows Run dialog (Win+R), a common technique to trick users into executing malicious commands under the guise of solving a fake CAPTCHA or error.
Detects network traffic associated with Adversary-in-the-Middle (AiTM) phishing kits that impersonate Microsoft and Google login pages. The rules identify the presence of specific HTML elements (like Cloudflare Turnstile anti-bot gates) and authentication session cookie headers served from non-authorized/non-official domains, indicating a reverse-proxy phishing attack.
Detects a multi-stage communication pattern involving an initial request to a URL-mirroring or shortening service with a long base64-encoded path, followed by a request to a screenshot/rendering API, and culminating in the reception of an image file. This chain is indicative of command and control (C2) communication where command output or exfiltrated data is encoded into images rendered by external web services to bypass traditional network security filters.
This rule monitors for two related scenarios targeting the RubyGems legacy API-key endpoint. First, it identifies suspicious responses indicating an API key might have been served from a CDN cache, potentially leading to cross-user credential leaks. Second, it flags a high volume of requests to the same endpoint using gzip compression, which is a known technique associated with edge-cache poisoning attacks.
Detects HTTP responses containing patterns associated with a malicious HTA stager hosted on Cloudflare Pages. The rule looks for specific URL structures often used in phishing campaigns that attempt to deliver malicious payloads, specifically involving a cloned Gmail attachment widget.
Detects anomalous network traffic associated with Vidar Stealer 2.0, where an HTTP GET request for a .jpg file is directed to a bare IP address, and the server's response header fails to indicate an image content type despite the requested file extension.
Detects a sequence of potentially malicious activity targeting LLM-related API endpoints. The rule identifies a client IP performing rapid, distinct endpoint discovery, followed by a POST request to sensitive paths, and concluding with a spike in 5xx server-side errors, indicating potential exploitation attempts or fuzzing against Large Language Model interfaces.
Detects web servers serving significantly different content sizes to AI crawlers (e.g., ChatGPT-User, ClaudeBot) compared to human web browsers for the same URL, which is a technique used to serve benign content to automated scrapers while delivering malicious or deceptive content to human users.
Detects web servers serving significantly different content sizes to AI crawlers (e.g., ChatGPT-User, ClaudeBot) compared to human web browsers for the same URL, which is a technique used to serve benign content to automated scrapers while delivering malicious or deceptive content to human users.
Detects web servers serving significantly different content sizes to AI crawlers (e.g., ChatGPT-User, ClaudeBot) compared to human web browsers for the same URL, which is a technique used to serve benign content to automated scrapers while delivering malicious or deceptive content to human users.
Detects HTTP responses containing a file named 'ClaudeDesktop.exe' served from paths containing '/artifacts/' on the 'claude.ai' host where the request referrer is not 'claude.ai' or 'anthropic.com', suggesting a malvertising or supply chain delivery attempt of a malicious installer.
Detects outbound HTTP traffic to punycode-encoded domains (typically used to spoof legitimate sites) that contain characteristic boilerplate text often generated by LLMs in their responses. This combination is indicative of an automated phishing kit setup or a landing page generated by an AI assistant.
Page 2 of 27