Detections
Explore public detection logic contributed by the community across SIEM and rule languages.
159 detections
Filters
Last updated
All Time
Detection languages
91
23
18
7
7
Contributors
46
40
17
14
9
Categories
20,402
11,796
5,865
5,145
4,864
Platforms
73
63
17
16
10
Products / Services
52
41
27
26
25
MITRE Techniques
48
46
37
28
27
CVEs
11
3
3
2
1
IDS Classtypes
6
1
IDS Protocols
7
Detects potential embedding inversion or extraction attempts by identifying callers exhibiting abnormally high volumes of similarity or vector search queries against RAG-enabled vector stores or cognitive search APIs within a short timeframe.
Detects potential embedding inversion or extraction attempts by identifying callers exhibiting abnormally high volumes of similarity or vector search queries against RAG-enabled vector stores or cognitive search APIs within a short timeframe.
Detects anomalous Agentforce session activity where a user, via the General CRM subagent, queries Lead records followed shortly by an access request to Account records. This pattern is indicative of a potentially hijacked agent session attempting to pivot from lead review into unauthorized data access of broader customer information.
Detects AI agent tooling (e.g., LangChain, Copilot, Claude Code) performing read operations on databases containing sensitive data patterns, or performing suspicious bulk data enumeration. The rule monitors Azure SQL Database security audit events for suspicious application name signatures and non-schema introspective queries.
Detects AI agent tooling (e.g., LangChain, Copilot, Claude Code) performing read operations on databases containing sensitive data patterns, or performing suspicious bulk data enumeration. The rule monitors Azure SQL Database security audit events for suspicious application name signatures and non-schema introspective queries.
Detects high-frequency search queries in SharePoint that utilize internal document identifiers (indexdocid), which may indicate an adversary attempting to enumerate or scrape content across a site collection.
Detects anomalous Agentforce session activity where a user, via the General CRM subagent, queries Lead records followed shortly by an access request to Account records. This pattern is indicative of a potentially hijacked agent session attempting to pivot from lead review into unauthorized data access of broader customer information.
Detects anomalous Agentforce session activity where a user, via the General CRM subagent, queries Lead records followed shortly by an access request to Account records. This pattern is indicative of a potentially hijacked agent session attempting to pivot from lead review into unauthorized data access of broader customer information.
Detects anomalous Agentforce session activity where a user, via the General CRM subagent, queries Lead records followed shortly by an access request to Account records. This pattern is indicative of a potentially hijacked agent session attempting to pivot from lead review into unauthorized data access of broader customer information.
Detects high-risk destructive database operations (DROP, TRUNCATE, or mass DELETE without a WHERE clause) performed by automated identities (AI agents, service accounts, or automation scripts) in a production environment outside of an authorized change window. This rule is designed to identify potential excessive-agency incidents where automated systems perform unauthorized destructive actions.
This rule detects a sequence of events where a database user record is updated (inserting a new password hash) followed by a successful authentication for that same user, immediately following a recent login failure. This pattern is indicative of an attacker modifying credentials and then testing the new access.
Detects anomalous database activity involving the creation of 'ZQ'-prefixed staging tables followed by a high-volume drop operation across multiple tables within a short timeframe. This behavior is indicative of data staging for potential exfiltration followed by an attempt to delete evidence or clear tracks.
Detects the presence of natural-language self-narration in code comments within Python or SQL scripts. This pattern is indicative of LLM-driven agentic execution, where automated agents narrate their own malicious actions, such as enumeration, data exfiltration, or database manipulation, as observed in the JADEPUFFER campaign.
Detects the execution of MySQL process command lines that simultaneously contain the strings 'AES_ENCRYPT' and 'UPDATE'. This pattern often indicates attempts to manipulate data within the database via SQL injection or unauthorized database modification.
Detects potential SQL injection attempts by monitoring for 'INSERT INTO' SQL commands executed via command-line arguments. The rule specifically looks for operations targeting user or role tables and including administrative keywords, originating from database or application-related processes like mysql or java.
Detects the use of database or archiving command-line utilities to export data related to RADIUS services. The rule filters out known backup service accounts and authorized backup processes, flagging activity that occurs outside of standard business hours or is performed by accounts not explicitly designated for administrative or database maintenance tasks.
This rule monitors for potential suspicious database operations executed via MySQL command-line utilities. It specifically detects the use of the AES_ENCRYPT function in conjunction with UPDATE statements, which is a technique often used by ransomware to encrypt database tables or data.
Detects a sequence of activity involving the use of the MySQL 'AES_ENCRYPT' function followed by a 'DROP TABLE', 'DROP SCHEMA', or 'DROP DATABASE' command on the same host within a 30-minute window. This behavior is consistent with extortion tactics, such as the JADEPUFFER campaign, where attackers encrypt sensitive database records and delete the original tables to compel ransom payments.
Detects potential SQL injection attempts by monitoring for 'INSERT INTO' SQL commands executed via command-line arguments. The rule specifically looks for operations targeting user or role tables and including administrative keywords, originating from database or application-related processes like mysql or java.
Detects unauthorized attempts to modify WordPress user tables (wp_usermeta) by web server processes to escalate privileges, specifically targeting the 'wp_capabilities' or 'administrator' flags. This activity is indicative of an exploit or post-exploitation behavior following a web-based attack.
Detects the creation of a new administrator account in WordPress by monitoring for SQL INSERT operations targeting the wp_users and wp_usermeta tables that involve setting the wp_capabilities to administrator. This behavior is indicative of unauthorized account creation often associated with compromised or malicious WordPress plugin updates.
Page 2 of 8