Tool
Hunt pack: Agrius
930 vendor-native detections · ready to paste into your SIEM · cross-linked to ATT&CK
Vendor-native detections covering the ATT&CK techniques attributed to Agrius - a ready-to-deploy hunt pack across Splunk, Elastic and Sentinel.
◈
Detections
50 shown of 930OMI Vulnerability Exploitation
Following the September 14th, 2021 release of three Elevation of Privilege (EoP) vulnerabilities (CVE-2021-38645, CVE-2021-38649, CVE-2021-38648) and one unauthenticated Remote Code Execution (RCE) vulnerability (CVE-2021-38647) in the Open Management Infrastructure (OMI) Framework.
This detection validates that any OMS-agent that is reporting to the Microsoft Sentinel workspace is updated with the patch. The detection will go over the heartbeats received from all agents over the last day and wi
Show query
let OMIVulnerabilityPatchVersion = "OMIVulnerabilityPatchVersion:1.13.40-0";
Heartbeat
| where Category == "Direct Agent"
| summarize arg_max(TimeGenerated,*) by Computer
| parse strcat("Version:" , Version) with * "Version:" Major:long "."
Minor:long "." Patch:long "-" *
| parse OMIVulnerabilityPatchVersion with * "OMIVulnerabilityPatchVersion:"
OMIVersionMajor:long "." OMIVersionMinor:long "." OMIVersionPatch:long "-" *
| where Major <OMIVersionMajor or (Major==OMIVersionMajor and Minor
<OMIVersionMinor) or (Major==OMIVersionMajor and Minor==OMIVersionMinor and
Patch<OMIVersionPatch)
| project Version, Major,Minor,Patch,
Computer,ComputerIP,OSType,OSName,ResourceId
Potential re-named sdelete usage (ASIM Version)
'This detection looks for command line parameters associated with the use of Sysinternals sdelete (https://docs.microsoft.com/sysinternals/downloads/sdelete) to delete multiple files on a host's C drive.
A threat actor may re-name the tool to avoid detection and then use it for destructive attacks on a host.
This detection uses the ASIM imProcess parser, this will need to be deployed before use - https://docs.microsoft.com/azure/sentinel/normalization'
Show query
imProcess
| where CommandLine has_all ("accepteula", "-s", "-r", "-q")
| where Process !endswith "sdelete.exe"
| where CommandLine !has "sdelete"
| extend AccountName = tostring(split(ActorUsername, @'\')[1]), AccountNTDomain = tostring(split(ActorUsername, @'\')[0])
Privileged User Logon from new ASN
'Detects a successful logon by a privileged account from an ASN not logged in from in the last 14 days.
Monitor these logons to ensure they are legitimate and identify if there are any similar sign ins.'
Show query
let admins=(IdentityInfo | where AssignedRoles contains "admin" or GroupMembership has "Admin" | summarize by tolower(AccountUPN)); let known_asns = ( SigninLogs | where TimeGenerated between(ago(14d)..ago(1d)) | where ResultType == 0 | summarize by AutonomousSystemNumber); SigninLogs | where TimeGenerated > ago(1d) | where ResultType == 0 | where tolower(UserPrincipalName) in (admins) | where AutonomousSystemNumber !in (known_asns) | project-reorder TimeGenerated, UserPrincipalName, UserAgent, IPAddress, AutonomousSystemNumber | extend AccountName = tostring(split(UserPrincipalName, "@")[0]), AccountUPNSuffix = tostring(split(UserPrincipalName, "@")[1])
Probable AdFind Recon Tool Usage (Normalized Process Events)
'Identifies the host and account that executed AdFind by hash and filename in addition to common and unique flags that are used by many threat actors in discovery.
To use this analytics rule, make sure you have deployed the [ASIM normalization parsers](https://aka.ms/ASimProcessEvent)'
Show query
let args = dynamic(["objectcategory","domainlist","dcmodes","adinfo","trustdmp","computers_pwdnotreqd","Domain Admins", "objectcategory=person", "objectcategory=computer", "objectcategory=*","dclist"]); let parentProcesses = dynamic(["pwsh.exe","powershell.exe","cmd.exe"]); imProcessCreate //looks for execution from a shell | where ActingProcessName has_any (parentProcesses) | extend ActingProcessFileName = tostring(split(ActingProcessName, '\\')[-1]) | where ActingProcessFileName in~ (parentProcesses) // main filter | where Process hassuffix "AdFind.exe" or TargetProcessSHA256 == "c92c158d7c37fea795114fa6491fe5f145ad2f8c08776b18ae79db811e8e36a3" // AdFind common Flags to check for from various threat actor TTPs or CommandLine has_any (args) | extend AlgorithmType = "SHA256" | extend AccountName = tostring(split(User, @'\')[1]), AccountNTDomain = tostring(split(User, @'\')[0]) | extend HostName = tostring(split(Dvc, ".")[0]), DomainIndex = toint(indexof(Dvc, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(Dvc, DomainIndex + 1), Dvc) | project-away DomainIndex
PulseConnectSecure - CVE-2021-22893 Possible Pulse Connect Secure RCE Vulnerability Attack
'This query identifies exploitation attempts using Pulse Connect Secure(PCS) vulnerability (CVE-2021-22893) to the VPN server'
Show query
let threshold = 3; PulseConnectSecure | where Messages contains "Unauthenticated request url /dana-na/" | summarize StartTime = min(TimeGenerated), EndTime = max(TimeGenerated), count() by Source_IP | where count_ > threshold
RDP Nesting
'Query detects potential lateral movement within a network by identifying when an RDP connection (EventID 4624, LogonType 10) is made to an initial system, followed by a subsequent RDP connection from that system to another, using the same account within a 60-minute window.
To reduce false positives, it excludes scenarios where the same account has made 5 or more connections to the same set of computers in the previous 7 days. This approach focuses on highlighting unusual RDP behaviour that sug
Show query
let endtime = 1d;
// Function to resolve hostname to IP address using DNS logs or a lookup table (example syntax)
let rdpConnections =
(union isfuzzy=true
(
SecurityEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and LogonType == 10
// Labeling the first RDP connection time, computer and ip
| extend
FirstHop = bin(TimeGenerated, 1m),
FirstComputer = toupper(Computer),
FirstRemoteIPAddress = IpAddress,
Account = tolower(Account)
),
(
WindowsEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and EventData has ("10")
| extend LogonType = tostring(EventData.LogonType)
| where LogonType == 10 // Labeling the first RDP connection time, computer and ip
| extend Account = strcat(tostring(EventData.TargetDomainName), "", tostring(EventData.TargetUserName))
| extend IpAddress = tostring(EventData.IpAddress)
| extend
FirstHop = bin(TimeGenerated, 1m),
FirstComputer = toupper(Computer),
FirstRemoteIPAddress = IpAddress,
Account = tolower(Account)
))
| join kind=inner (
(union isfuzzy=true
(
SecurityEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and LogonType == 10
// Labeling the second RDP connection time, computer and ip
| extend
SecondHop = bin(TimeGenerated, 1m),
SecondComputer = toupper(Computer),
SecondRemoteIPAddress = IpAddress,
Account = tolower(Account)
),
(
WindowsEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and EventData has ("10")
| extend LogonType = toint(EventData.LogonType)
| where LogonType == 10 // Labeling the second RDP connection time, computer and ip
| extend Account = strcat(tostring(EventData.TargetDomainName), "", tostring(EventData.TargetUserName))
| extend IpAddress = tostring(EventData.IpAddress)
| extend
SecondHop = bin(TimeGenerated, 1m),
SecondComputer = toupper(Computer),
SecondRemoteIPAddress = IpAddress,
Account = tolower(Account)
))
)
on Account
| distinct
Account,
FirstHop,
FirstComputer,
FirstRemoteIPAddress,
SecondHop,
SecondComputer,
SecondRemoteIPAddress,
AccountType,
Activity,
LogonTypeName,
ProcessName;
// Resolve hostnames to IP addresses device network Ip's
let listOfFirstComputer = rdpConnections | distinct FirstComputer;
let listOfSecondComputer = rdpConnections | distinct SecondComputer;
let resolvedIPs =
DeviceNetworkInfo
| where TimeGenerated >= ago(endtime)
| where isnotempty(ConnectedNetworks) and NetworkAdapterStatus == "Up"
| extend ClientIP = tostring(parse_json(IPAddresses[0]).IPAddress)
| where isnotempty(ClientIP)
| where DeviceName in~ (listOfFirstComputer) or DeviceName in~ (listOfSecondComputer)
| summarize arg_max(TimeGenerated, ClientIP) by Computer= DeviceName
| project Computer=toupper(Computer), ResolvedIP = ClientIP;
// Join resolved IPs with the RDP connections
rdpConnections
| join kind=inner (resolvedIPs) on $left.FirstComputer == $right.Computer
| join kind=inner (resolvedIPs) on $left.SecondComputer == $right.Computer
// | where ResolvedIP != ResolvedIP1
| distinct
Account,
FirstHop,
FirstComputer,
FirstComputerIP = ResolvedIP,
FirstRemoteIPAddress,
SecondHop,
SecondComputer,
SecondComputerIP = ResolvedIP1,
SecondRemoteIPAddress,
AccountType,
Activity,
LogonTypeName,
ProcessName
// Ensure the first connection is before the second connection
// Identify only RDP to another computer from within the first RDP connection by only choosing matches where the Computer names do not match
// Ensure the IPAddresses do not match by excluding connections from the same computers with first hop RDP connections to multiple computers
| where FirstComputer != SecondComputer
and FirstRemoteIPAddress != SecondRemoteIPAddress
and SecondHop > FirstHop
// Ensure the second hop occurs within 30 minutes of the first hop
| where SecondHop <= FirstHop + 30m
| where SecondRemoteIPAddress == FirstComputerIP
| summarize FirstHopFirstSeen = min(FirstHop), FirstHopLastSeen = max(FirstHop)
by
Account,
FirstComputer,
FirstComputerIP,
FirstRemoteIPAddress,
SecondHop,
SecondComputer,
SecondComputerIP,
SecondRemoteIPAddress,
AccountType,
Activity,
LogonTypeName,
ProcessName
| extend
AccountName = tostring(split(Account, @"")[1]),
AccountNTDomain = tostring(split(Account, @"")[0])
| extend
HostName1 = tostring(split(FirstComputer, ".")[0]),
DomainIndex = toint(indexof(FirstComputer, '.'))
| extend HostNameDomain1 = iff(DomainIndex != -1, substring(FirstComputer, DomainIndex + 1), FirstComputer)
| extend
HostName2 = tostring(split(SecondComputer, ".")[0]),
DomainIndex = toint(indexof(SecondComputer, '.'))
| extend HostNameDomain2 = iff(DomainIndex != -1, substring(SecondComputer, DomainIndex + 1), SecondComputer)
| project-away DomainIndex
Rare RDP Connections
'Identifies when an RDP connection is new or rare related to any logon type by a given account today compared with the previous 14 days.
RDP connections are indicated by the EventID 4624 with LogonType = 10'
Show query
let starttime = 14d;
let endtime = 1d;
(union isfuzzy=true
(SecurityEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and LogonType == 10
| summarize StartTime = min(TimeGenerated), EndTime = max(TimeGenerated), ConnectionCount = count()
by Account = tolower(Account), Computer = toupper(Computer), IpAddress, AccountType, Activity, LogonTypeName, ProcessName
// use left anti to exclude anything from the previous 14 days that is not rare
),
(WindowsEvent
| where TimeGenerated >= ago(endtime)
| where EventID == 4624 and EventData has ("10")
| extend LogonType = tostring(EventData.LogonType)
| where LogonType == 10
| extend Account = strcat(tostring(EventData.TargetDomainName),"\\", tostring(EventData.TargetUserName))
| extend ProcessName = tostring(EventData.ProcessName)
| extend IpAddress = tostring(EventData.IpAddress)
| extend TargetUserSid = tostring(EventData.TargetUserSid)
| extend AccountType=case(Account endswith "$" or TargetUserSid in ("S-1-5-18", "S-1-5-19", "S-1-5-20"), "Machine", isempty(TargetUserSid), "", "User")
| extend Activity="4624 - An account was successfully logged on."
| extend LogonTypeName="10 - RemoteInteractive"
| summarize StartTime = min(TimeGenerated), EndTime = max(TimeGenerated), ConnectionCount = count()
by Account = tolower(Account), Computer = toupper(Computer), IpAddress, AccountType, Activity, LogonTypeName, ProcessName
))
| join kind=leftanti (
(union isfuzzy=true
(SecurityEvent
| where TimeGenerated between (ago(starttime) .. ago(endtime))
| where EventID == 4624
| summarize by Computer = toupper(Computer), IpAddress, Account = tolower(Account)
),
( WindowsEvent
| where TimeGenerated between (ago(starttime) .. ago(endtime))
| where EventID == 4624
| extend IpAddress = tostring(EventData.IpAddress)
| extend Account = strcat(tostring(EventData.TargetDomainName),"\\", tostring(EventData.TargetUserName))
| summarize by Computer = toupper(Computer), IpAddress, Account = tolower(Account)
))
) on Account, Computer
| summarize StartTime = min(StartTime), EndTime = max(EndTime), ConnectionCount = sum(ConnectionCount)
by Account, Computer, IpAddress, AccountType, Activity, LogonTypeName, ProcessName
| extend HostName = tostring(split(Computer, ".")[0]), DomainIndex = toint(indexof(Computer, '.'))
| extend HostNameDomain = iff(DomainIndex != -1, substring(Computer, DomainIndex + 1), Computer)
| extend AccountName = tostring(split(Account, @"\")[1]), AccountNTDomain = tostring(split(Account, @"\")[0])
| project-away DomainIndex
Risky user signin observed in non-Microsoft network device
'This content is utilized to identify instances of successful login by risky users, who have been observed engaging in potentially suspicious network activity on non-Microsoft network devices.'
Show query
SigninLogs
//Find risky Signin
| where RiskState == "atRisk" and ResultType == 0
| extend Signin_Time = TimeGenerated
| summarize
AppDisplayName=make_set(AppDisplayName),
ClientAppUsed=make_set(ClientAppUsed),
UserAgent=make_set(UserAgent),
CorrelationId=make_set(CorrelationId),
Signin_Time= min(Signin_Time),
RiskEventTypes=make_set(RiskEventTypes)
by
ConditionalAccessStatus,
IPAddress,
IsRisky,
ResourceDisplayName,
RiskDetail,
ResultType,
RiskLevelAggregated,
RiskLevelDuringSignIn,
RiskState,
UserPrincipalName=tostring(tolower(UserPrincipalName)),
SourceSystem
| join kind=inner (
CommonSecurityLog
| where DeviceVendor has_any ("Palo Alto Networks", "Fortinet", "Check Point", "Zscaler")
| where DeviceProduct startswith "FortiGate" or DeviceProduct startswith "PAN" or DeviceProduct startswith "VPN" or DeviceProduct startswith "FireWall" or DeviceProduct startswith "NSSWeblog" or DeviceProduct startswith "URL"
| where DeviceAction != "Block"
| where isnotempty(RequestURL)
| where isnotempty(SourceUserName)
| extend SourceUserName = tolower(SourceUserName)
| summarize
min(TimeGenerated),
max(TimeGenerated),
Activity=make_set(Activity)
by DestinationHostName, DestinationIP, RequestURL, SourceUserName=tostring(tolower(SourceUserName)),DeviceVendor,DeviceProduct
| extend 3p_observed_Time= min_TimeGenerated,Name = tostring(split(SourceUserName,"@")[0]),UPNSuffix =tostring(split(SourceUserName,"@")[1]))
on $left.IPAddress == $right.DestinationIP and $left.UserPrincipalName == $right.SourceUserName
| extend Timediff = datetime_diff('day', 3p_observed_Time, Signin_Time)
| where Timediff <= 1 and Timediff >= 0
RunningRAT request parameters
'This detection will alert when RunningRAT URI parameters or paths are detect in an HTTP request.
Id the device blocked this communication presence of this alert means the RunningRAT implant is likely still executing on the source host.'
Show query
let runningRAT_parameters = dynamic(['/ui/chk', 'mactok=', 'UsRnMe=', 'IlocalP=', 'kMnD=']); CommonSecurityLog | where RequestMethod == "GET" | project TimeGenerated, DeviceVendor, DeviceProduct, DeviceAction, DestinationDnsDomain, DestinationIP, RequestURL, SourceIP, SourceHostName, RequestClientApplication | where RequestURL has_any (runningRAT_parameters)
SUNBURST and SUPERNOVA backdoor hashes (Normalized File Events)
Identifies SolarWinds SUNBURST and SUPERNOVA backdoor file hash IOCs in File Events
To use this analytics rule, make sure you have deployed the [ASIM normalization parsers](https://aka.ms/ASimFileEvent)
References:
- https://www.fireeye.com/blog/threat-research/2020/12/evasive-attacker-leverages-solarwinds-supply-chain-compromises-with-sunburst-backdoor.html
- https://gist.github.com/olafhartong/71ffdd4cab4b6acd5cbcd1a0691ff82f
Show query
let SunburstMD5=dynamic(["b91ce2fa41029f6955bff20079468448","02af7cec58b9a5da1c542b5a32151ba1","2c4a910a1299cdae2a4e55988a2f102e","846e27a652a5e1bfbd0ddd38a16dc865","4f2eb62fa529c0283b28d05ddd311fae"]); let SupernovaMD5="56ceb6d0011d87b6e4d7023d7ef85676"; imFileEvent | where TargetFileMD5 in (SunburstMD5) or TargetFileMD5 in (SupernovaMD5) | extend AccountName = tostring(split(User, @'\')[1]), AccountNTDomain = tostring(split(User, @'\')[0]) | extend AlgorithmType = "MD5"
SUNBURST suspicious SolarWinds child processes (Normalized Process Events)
Identifies suspicious child processes of SolarWinds.Orion.Core.BusinessLayer.dll that may be evidence of the SUNBURST backdoor
References:
- https://www.fireeye.com/blog/threat-research/2020/12/evasive-attacker-leverages-solarwinds-supply-chain-compromises-with-sunburst-backdoor.html
- https://gist.github.com/olafhartong/71ffdd4cab4b6acd5cbcd1a0691ff82f
To use this analytics rule, make sure you have deployed the [ASIM normalization parsers](https://aka.ms/ASimProcessEvent)'
Show query
let excludeProcs = dynamic([@"\SolarWinds\Orion\APM\APMServiceControl.exe", @"\SolarWinds\Orion\ExportToPDFCmd.Exe", @"\SolarWinds.Credentials\SolarWinds.Credentials.Orion.WebApi.exe", @"\SolarWinds\Orion\Topology\SolarWinds.Orion.Topology.Calculator.exe", @"\SolarWinds\Orion\Database-Maint.exe", @"\SolarWinds.Orion.ApiPoller.Service\SolarWinds.Orion.ApiPoller.Service.exe", @"\Windows\SysWOW64\WerFault.exe"]); imProcessCreate | where Process hassuffix 'solarwinds.businesslayerhost.exe' | where not(Process has_any (excludeProcs)) | extend AccountName = tostring(split(ActorUsername, @'\')[1]), AccountNTDomain = tostring(split(ActorUsername, @'\')[0]) | extend HostName = tostring(split(Dvc, ".")[0]), DomainIndex = toint(indexof(Dvc, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(Dvc, DomainIndex + 1), Dvc) | project-away DomainIndex
Sdelete deployed via GPO and run recursively (ASIM Version)
'This query looks for the Sdelete process being run recursively after being deployed to a host via GPO. Attackers could use this technique to deploy Sdelete to multiple host and delete data on them.
This query uses the Advanced Security Information Model. Parsers will need to be deployed before use: https://docs.microsoft.com/azure/sentinel/normalization'
Show query
_Im_ProcessEvent
| where EventType =~ "ProcessCreated"
| where Process endswith "svchost.exe"
| where CommandLine has "-k GPSvcGroup" or CommandLine has "-s gpsvc"
| extend timekey = bin(TimeGenerated, 1m)
| project timekey, ActingProcessId, Dvc
| join kind=inner (
_Im_ProcessEvent
| where EventType =~ "ProcessCreated"
| where Process =~ "sdelete.exe" or CommandLine has "sdelete"
| where ActingProcessName endswith "svchost.exe"
| where CommandLine has_all ("-s", "-r")
| extend timekey = bin(TimeGenerated, 1m)
)
on $left.ActingProcessId == $right.ParentProcessId, timekey, Dvc
| extend AccountName = tostring(split(ActorUsername, @'\')[1]), AccountNTDomain = tostring(split(ActorUsername, @'\')[0])
| extend HostName = tostring(split(Dvc, ".")[0]), DomainIndex = toint(indexof(Dvc, '.'))
| extend HostNameDomain = iff(DomainIndex != -1, substring(Dvc, DomainIndex + 1), Dvc)
| project-away DomainIndex
Service Principal Assigned App Role With Sensitive Access
'Detects a Service Principal being assigned an app role that has sensitive access such as Mail.Read.
A threat actor who compromises a Service Principal may assign it an app role to allow it to access sensitive data, or to perform other actions.
Ensure that any assignment to a Service Principal is valid and appropriate.
Ref: https://docs.microsoft.com/azure/active-directory/fundamentals/security-operations-applications#application-granted-highly-privileged-permissions'
Show query
// Add other permissions to this list as needed
let permissions = dynamic([".All", "ReadWrite", "Mail.", "offline_access", "Files.Read", "Notes.Read", "ChannelMessage.Read", "Chat.Read", "TeamsActivity.Read",
"Group.Read", "EWS.AccessAsUser.All", "EAS.AccessAsUser.All"]);
let auditList =
AuditLogs
| where OperationName =~ "Add app role assignment to service principal"
| mv-expand TargetResources[0].modifiedProperties
| extend TargetResources_0_modifiedProperties = column_ifexists("TargetResources_0_modifiedProperties", '')
| where isnotempty(TargetResources_0_modifiedProperties)
;
let detailsList = auditList
| where TargetResources_0_modifiedProperties.displayName =~ "AppRole.Value" or TargetResources_0_modifiedProperties.displayName =~ "DelegatedPermissionGrant.Scope"
| extend Permissions = split((parse_json(tostring(TargetResources_0_modifiedProperties.newValue))), " ")
| where Permissions has_any (permissions)
| summarize AddedPermissions=make_set(Permissions,200) by CorrelationId
| join kind=inner auditList on CorrelationId
| extend InitiatingAppName = tostring(InitiatedBy.app.displayName)
| extend InitiatingAppServicePrincipalId = tostring(InitiatedBy.app.servicePrincipalId)
| extend InitiatingUserPrincipalName = tostring(InitiatedBy.user.userPrincipalName)
| extend InitiatingAadUserId = tostring(InitiatedBy.user.id)
| extend InitiatingIPAddress = tostring(InitiatedBy.user.ipAddress)
| extend InitiatedBy = tostring(iff(isnotempty(InitiatingUserPrincipalName),InitiatingUserPrincipalName, InitiatingAppName))
| extend displayName = tostring(TargetResources_0_modifiedProperties.displayName), newValue = tostring(parse_json(tostring(TargetResources_0_modifiedProperties.newValue)))
| where displayName == "ServicePrincipal.ObjectID" or displayName == "ServicePrincipal.DisplayName"
| extend displayName = case(displayName == "ServicePrincipal.ObjectID", "ServicePrincipalObjectID", displayName == "ServicePrincipal.DisplayName", "ServicePrincipalDisplayName", displayName)
| project TimeGenerated, CorrelationId, Id, AddedPermissions = tostring(AddedPermissions), InitiatingAadUserId, InitiatingAppName, InitiatingAppServicePrincipalId, InitiatingIPAddress, InitiatingUserPrincipalName, InitiatedBy, displayName, newValue
;
detailsList | project Id, displayName, newValue
| evaluate pivot(displayName, make_set(newValue))
| join kind=inner detailsList on Id
| extend ServicePrincipalObjectID = todynamic(column_ifexists("ServicePrincipalObjectID", "")), ServicePrincipalDisplayName = todynamic(column_ifexists("ServicePrincipalDisplayName", ""))
| mv-expand ServicePrincipalObjectID, ServicePrincipalDisplayName
| project-away Id1, displayName, newValue
| extend ServicePrincipalObjectID = tostring(ServicePrincipalObjectID), ServicePrincipalDisplayName = tostring(ServicePrincipalDisplayName)
| summarize FirstSeen = min(TimeGenerated), LastSeen = max(TimeGenerated), EventIds = make_set(Id,200) by CorrelationId, AddedPermissions, InitiatingAadUserId, InitiatingAppName, InitiatingAppServicePrincipalId, InitiatingIPAddress, InitiatingUserPrincipalName, InitiatedBy, ServicePrincipalDisplayName, ServicePrincipalObjectID
| extend InitiatingAccountName = tostring(split(InitiatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(InitiatingUserPrincipalName, "@")[1])
Service Principal Assigned Privileged Role
'Detects a privileged role being added to a Service Principal.
Ensure that any assignment to a Service Principal is valid and appropriate - Service Principals should not be assigned to very highly privileged roles such as Global Admin.
Ref: https://docs.microsoft.com/azure/active-directory/fundamentals/security-operations-privileged-accounts#changes-to-privileged-accounts'
Show query
AuditLogs
| where OperationName has_all ("member to role", "add")
| where Result =~ "Success"
| extend type_ = tostring(TargetResources[0].type)
| where type_ =~ "ServicePrincipal"
| where isnotempty(TargetResources)
| extend InitiatingAppName = tostring(InitiatedBy.app.displayName)
| extend InitiatingAppServicePrincipalId = tostring(InitiatedBy.app.servicePrincipalId)
| extend InitiatingUserPrincipalName = tostring(InitiatedBy.user.userPrincipalName)
| extend InitiatingAadUserId = tostring(InitiatedBy.user.id)
| extend InitiatingIPAddress = tostring(InitiatedBy.user.ipAddress)
| extend InitiatedBy = tostring(iff(isnotempty(InitiatingUserPrincipalName),InitiatingUserPrincipalName, InitiatingAppName))
| extend ServicePrincipalName = tostring(TargetResources[0].displayName)
| extend ServicePrincipalId = tostring(TargetResources[0].id)
| mv-expand TargetResources[0].modifiedProperties
| extend TargetResources_0_modifiedProperties = columnifexists("TargetResources_0_modifiedProperties", '')
| where isnotempty(TargetResources_0_modifiedProperties)
| extend displayName = tostring(TargetResources_0_modifiedProperties.displayName), newValue = tostring(parse_json(tostring(TargetResources_0_modifiedProperties.newValue)))
| where displayName == "Role.DisplayName" and newValue contains "admin"
| extend InitiatingAccountName = tostring(split(InitiatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(InitiatingUserPrincipalName, "@")[1])
| extend TargetRole = newValue
| project-reorder TimeGenerated, ServicePrincipalName, ServicePrincipalId, InitiatedBy, TargetRole, InitiatingIPAddress
Service Principal Authentication Attempt from New Country
'Detects when there is a Service Principal login attempt from a country that has not seen a successful login in the previous 14 days.
Threat actors may attempt to authenticate with credentials from compromised accounts - monitoring attempts from anomalous locations may help identify these attempts.
Authentication attempts should be investigated to ensure the activity was legitimate and if there is other similar activity.
Ref: https://docs.microsoft.com/azure/active-directory/fundamentals/s
Show query
let known_locations = ( AADServicePrincipalSignInLogs | where TimeGenerated between(ago(14d)..ago(1d)) | where ResultType == 0 | summarize by Location); AADServicePrincipalSignInLogs | where TimeGenerated > ago(1d) | where ResultType != 50126 | where Location !in (known_locations) | extend City = tostring(parse_json(LocationDetails).city) | extend State = tostring(parse_json(LocationDetails).state) | extend Place = strcat(City, " - ", State) | extend Result = strcat(tostring(ResultType), " - ", ResultDescription) | summarize FirstSeen=min(TimeGenerated), LastSeen=max(TimeGenerated), make_set(Result), make_set(IPAddress), make_set(Place) by ServicePrincipalName, Location
Sign-ins from IPs that attempt sign-ins to disabled accounts (Uses Authentication Normalization)
'Identifies IPs with failed attempts to sign in to one or more disabled accounts signed in successfully to another account.
To use this analytics rule, make sure you have deployed the [ASIM normalization parsers](https://aka.ms/ASimAuthentication)'
Show query
imAuthentication
| where EventResult =='Failure'
| where EventResultDetails == 'User disabled'
| summarize StartTime=min(EventStartTime), EndTime=max(EventEndTime), disabledAccountLoginAttempts = count()
, disabledAccountsTargeted = dcount(TargetUsername), disabledAccountSet = make_set(TargetUsername)
, applicationsTargeted = dcount(TargetAppName)
, applicationSet = make_set(TargetAppName)
by SrcDvcIpAddr, Type
| order by disabledAccountLoginAttempts desc
| join kind=leftouter
(
// Consider these IPs suspicious - and alert any related successful sign-ins
imAuthentication
| where EventResult=='Success'
| summarize successfulAccountSigninCount = dcount(TargetUsername), successfulAccountSigninSet = makeset(TargetUsername, 15) by SrcDvcIpAddr, Type
// Assume IPs associated with sign-ins from 100+ distinct user accounts are safe
| where successfulAccountSigninCount < 100
)
on SrcDvcIpAddr
| where isnotempty(successfulAccountSigninCount)
| project StartTime, EndTime, SrcDvcIpAddr, disabledAccountLoginAttempts, disabledAccountsTargeted, disabledAccountSet, applicationSet,
successfulAccountSigninCount, successfulAccountSigninSet, Type
| order by disabledAccountLoginAttempts
Silk Typhoon New UM Service Child Process
'This query looks for new processes being spawned by the Exchange UM service where that process has not previously been observed before.
Reference: https://www.microsoft.com/security/blog/2021/03/02/hafnium-targeting-exchange-servers/'
Show query
let lookback = 14d;
let timeframe = 1d;
(union isfuzzy=true
(SecurityEvent
| where TimeGenerated > ago(lookback) and TimeGenerated < ago(timeframe)
| where EventID == 4688
| where ParentProcessName has_any ("umworkerprocess.exe", "UMService.exe")
| join kind=rightanti (
SecurityEvent
| where TimeGenerated > ago(timeframe)
| where ParentProcessName has_any ("umworkerprocess.exe", "UMService.exe")
| where EventID == 4688) on NewProcessName
),
(WindowsEvent
| where TimeGenerated > ago(lookback) and TimeGenerated < ago(timeframe)
| where EventID == 4688 and EventData has_any ("umworkerprocess.exe", "UMService.exe")
| extend ParentProcessName = tostring(EventData.ParentProcessName)
| where ParentProcessName has_any ("umworkerprocess.exe", "UMService.exe")
| extend NewProcessName = tostring(EventData.NewProcessName)
| extend Account = strcat(tostring(EventData.SubjectDomainName),"\\", tostring(EventData.SubjectUserName))
| extend IpAddress = tostring(EventData.IpAddress)
| join kind=rightanti (
WindowsEvent
| where TimeGenerated > ago(timeframe)
| where EventID == 4688 and EventData has_any ("umworkerprocess.exe", "UMService.exe")
| extend ParentProcessName = tostring(EventData.ParentProcessName)
| where ParentProcessName has_any ("umworkerprocess.exe", "UMService.exe")
| extend NewProcessName = tostring(EventData.NewProcessName)
| extend Account = strcat(tostring(EventData.SubjectDomainName),"\\", tostring(EventData.SubjectUserName))
| extend IpAddress = tostring(EventData.IpAddress)) on NewProcessName
| extend HostName = tostring(split(Computer, ".")[0]), DomainIndex = toint(indexof(Computer, '.'))
| extend HostNameDomain = iff(DomainIndex != -1, substring(Computer, DomainIndex + 1), Computer)
| extend SubjectUserName = tostring(EventData.SubjectUserName), SubjectDomainName = tostring(EventData.SubjectDomainName)
| project-away DomainIndex
))
Silk Typhoon Suspicious Exchange Request
'This query looks for suspicious request patterns to Exchange servers that fit a pattern observed by Silk Typhoon actors.
The same query can be run on HTTPProxy logs from on-premise hosted Exchange servers.
Reference: https://www.microsoft.com/security/blog/2021/03/02/hafnium-targeting-exchange-servers/'
Show query
let exchange_servers = ( W3CIISLog | where TimeGenerated > ago(14d) | where sSiteName =~ "Exchange Back End" | summarize by Computer); W3CIISLog | where TimeGenerated > ago(1d) | where Computer in (exchange_servers) | where csUriQuery startswith "t=" | project-reorder TimeGenerated, Computer, csUriStem, csUriQuery, csUserName, csUserAgent, cIP | extend HostName = tostring(split(Computer, ".")[0]), DomainIndex = toint(indexof(Computer, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(Computer, DomainIndex + 1), Computer) | extend AccountName = tostring(split(csUserName, "@")[0]), AccountUPNSuffix = tostring(split(csUserName, "@")[1])
Silk Typhoon Suspicious File Downloads.
'This query looks for messages related to file downloads of suspicious file types. This query uses the Exchange HttpProxy AOBGeneratorLog, you will need to onboard this log as a custom log under the table http_proxy_oab_CL before using this query.
Reference: https://www.microsoft.com/security/blog/2021/03/02/hafnium-targeting-exchange-servers/'
Show query
let scriptExtensions = dynamic([".php", ".jsp", ".js", ".aspx", ".asmx", ".asax", ".cfm", ".shtml"]);
http_proxy_oab_CL
| where RawData contains "Download failed and temporary file"
| extend File = extract("([^\\\\]*)(\\\\[^']*)",2,RawData)
| extend Extension = strcat(".",split(File, ".")[-1])
| extend InteractiveFile = iif(Extension in (scriptExtensions), "Yes", "No")
// Uncomment the following line to alert only on interactive file download type
//| where InteractiveFile =~ "Yes"
| extend HostName = tostring(split(Computer, ".")[0]), DomainIndex = toint(indexof(Computer, '.'))
| extend HostNameDomain = iff(DomainIndex != -1, substring(Computer, DomainIndex + 1), Computer)
Silk Typhoon Suspicious UM Service Error
'This query looks for errors that may indicate that an attacker is attempting to exploit a vulnerability in the service.
Reference: https://www.microsoft.com/security/blog/2021/03/02/hafnium-targeting-exchange-servers/'
Show query
Event | where EventLog =~ "Application" | where Source startswith "MSExchange" | where EventLevelName =~ "error" | where (RenderedDescription startswith "Watson report" and RenderedDescription contains "umworkerprocess" and RenderedDescription contains "TextFormattingRunProperties") or RenderedDescription startswith "An unhandled exception occurred in a UM worker process" or RenderedDescription startswith "The Microsoft Exchange Unified Messaging service" or RenderedDescription contains "MSExchange Unified Messaging" | where RenderedDescription !contains "System.OutOfMemoryException" | extend HostName = tostring(split(Computer, ".")[0]), DomainIndex = toint(indexof(Computer, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(Computer, DomainIndex + 1), Computer) | project-away DomainIndex
Solorigate Defender Detections
'Surfaces any Defender Alert for Solorigate Events. In Microsoft Sentinel the SecurityAlerts table includes only the Device Name of the affected device, this query joins the DeviceInfo table to clearly connect other information such as
Device group, ip, logged on users etc. This way, the Microsoft Sentinel user can have all the pertinent device info in one view for all the the Solarigate Defender alerts.'
Show query
DeviceInfo | extend DeviceName = tolower(DeviceName) | join (SecurityAlert | where ProviderName =~ "MDATP" | extend ThreatName = tostring(parse_json(ExtendedProperties).ThreatName) | where ThreatName has "Solorigate" ) on $left.DeviceName == $right.CompromisedEntity | project TimeGenerated, DisplayName, ThreatName, CompromisedEntity, PublicIP, MachineGroup, AlertSeverity, Description, LoggedOnUsers, DeviceId, TenantId | extend HostName = tostring(split(CompromisedEntity, ".")[0]), DomainIndex = toint(indexof(CompromisedEntity, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(CompromisedEntity, DomainIndex + 1), CompromisedEntity) | project-away DomainIndex
Suspicious Login from deleted guest account
' This query will detect logins from guest account which was recently deleted.
For any successful logins from deleted identities should be investigated further if any existing user accounts have been altered or linked to such identity prior deletion'
Show query
let query_frequency = 1h;
let query_period = 1d;
AuditLogs
| where TimeGenerated > ago(query_frequency)
| where Category =~ "UserManagement" and OperationName =~ "Delete user"
| mv-expand TargetResource = TargetResources
| where TargetResource["type"] == "User" and TargetResource["userPrincipalName"] has "#EXT#"
| extend ParsedDeletedUserPrincipalName = extract(@"^[0-9a-f]{32}([^\#]+)\#EXT\#", 1, tostring(TargetResource["userPrincipalName"]))
| extend
Initiator = iif(isnotempty(InitiatedBy["app"]), tostring(InitiatedBy["app"]["displayName"]), tostring(InitiatedBy["user"]["userPrincipalName"])),
InitiatorId = iif(isnotempty(InitiatedBy["app"]), tostring(InitiatedBy["app"]["servicePrincipalId"]), tostring(InitiatedBy["user"]["id"])),
Delete_IPAddress = tostring(InitiatedBy[tostring(bag_keys(InitiatedBy)[0])]["ipAddress"])
| project Delete_TimeGenerated = TimeGenerated, Category, Identity, Initiator, Delete_IPAddress, OperationName, Result, ParsedDeletedUserPrincipalName, InitiatedBy, AdditionalDetails, TargetResources, InitiatorId, CorrelationId
| join kind=inner (
SigninLogs
| where TimeGenerated > ago(query_period)
| where ResultType == 0
| summarize take_any(*) by UserPrincipalName
| extend ParsedUserPrincipalName = translate("@", "_", UserPrincipalName)
| project SigninLogs_TimeGenerated = TimeGenerated, UserPrincipalName, UserDisplayName, ResultType, ResultDescription, IPAddress, LocationDetails, AppDisplayName, ResourceDisplayName, ClientAppUsed, UserAgent, DeviceDetail, UserId, UserType, OriginalRequestId, ParsedUserPrincipalName
) on $left.ParsedDeletedUserPrincipalName == $right.ParsedUserPrincipalName
| where SigninLogs_TimeGenerated > Delete_TimeGenerated
| project-away ParsedDeletedUserPrincipalName, ParsedUserPrincipalName
| extend
AccountName = tostring(split(UserPrincipalName, "@")[0]),
AccountUPNSuffix = tostring(split(UserPrincipalName, "@")[1])
Suspicious Sign In by Entra ID Connect Sync Account
'This query looks for sign ins by the Microsoft Entra ID Connect Sync account to Azure where properties about the logon are anomalous.
This query uses Microsoft Sentinel's UEBA features to detect these suspicious properties.
A threat actor may attempt to steal the Sync account credentials and use them to access Azure resources. This alert should be
reviewed to ensure that the log in came was from a legitimate source.'
Show query
BehaviorAnalytics // User modification is expected from this account so focus on logons | where ActivityType =~ "LogOn" | where UserName startswith "Sync_" and UsersInsights.AccountDisplayName =~ "On-Premises Directory Synchronization Service Account" // Filter out this expected activity | where ActivityInsights.App !~ "Microsoft Azure Active Directory Connect" | where InvestigationPriority > 0 | extend Name = split(UserPrincipalName, "@")[0], UPNSuffix = split(UserPrincipalName, "@")[1]
Suspicious VM Instance Creation Activity Detected
'This detection identifies high-severity alerts across various Microsoft security products, including Microsoft Defender XDR and Microsoft Entra ID, and correlates them with instances of Google Cloud VM creation. It focuses on instances where VMs were created within a short timeframe of high-severity alerts, potentially indicating suspicious activity.'
Show query
// Filter alerts from specific Microsoft security products with medium and high severity
SecurityAlert
| where ProductName in ("Microsoft 365 Defender", "Azure Active Directory", "Microsoft Defender Advanced Threat Protection", "Microsoft Cloud App Security", "Azure Active Directory Identity Protection", "Microsoft Defender ATP")
| where AlertSeverity has_any ("Medium", "High")
// Parse JSON entities and extend AlertTimeGenerated
| extend Entities = parse_json(Entities), AlertTimeGenerated=TimeGenerated
// Extract and process IP entities
| mv-apply Entity = Entities on
(
where Entity.Type == 'ip'
| extend EntityIp = tostring(Entity.Address)
)
// Extract and process account entities
| mv-apply Entity = Entities on
(
where Entity.Type == 'account'
| extend AccountObjectId = tostring(Entity.AadUserId)
)
// Filter out records with empty EntityIp
| where isnotempty(EntityIp)
// Summarize data and create sets of entities and system alert IDs
| summarize Entitys=make_set(Entity), SystemAlertIds=make_set(SystemAlertId)
by
AlertName,
ProductName,
AlertSeverity,
EntityIp,
Tactics,
Techniques,
ProviderName,
AlertTime= bin(AlertTimeGenerated, 1d),
AccountObjectId
// Join with GCPAuditLogs for VM instance creation
| join kind=inner (
GCPAuditLogs
| where ServiceName == "compute.googleapis.com" and MethodName endswith "instances.insert"
| extend
GCPUserUPN= tostring(parse_json(AuthenticationInfo).principalEmail),
GCPUserIp = tostring(parse_json(RequestMetadata).callerIp),
GCPUserUA= tostring(parse_json(RequestMetadata).callerSuppliedUserAgent),
VMStatus = tostring(parse_json(Response).status),
VMOperation=tostring(parse_json(Response).operationType),
VMName= tostring(parse_json(Request).name),
VMType = tostring(split(parse_json(Request).machineType, "/")[-1])
| where GCPUserUPN !has "gserviceaccount.com"
| where VMOperation == "insert" and isnotempty(GCPUserIp) and GCPUserIp != "private"
| project
GCPOperationTime=TimeGenerated,
VMName,
VMStatus,
MethodName,
GCPUserUPN,
ProjectId,
GCPUserIp,
GCPUserUA,
VMOperation,
VMType
)
on $left.EntityIp == $right.GCPUserIp
// Join with IdentityInfo to enrich user identity details
| join kind=inner (IdentityInfo
| distinct AccountObjectId, AccountUPN, JobTitle
)
on AccountObjectId
// Calculate the time difference between the alert and VM creation for further analysis
| extend TimeDiff= datetime_diff('day', AlertTime, GCPOperationTime),Name = split(GCPUserUPN, "@")[0], UPNSuffix = split(GCPUserUPN, "@")[1]
Suspicious linking of existing user to external User
' This query will detect when an attempt is made to update an existing user and link it to an guest or external identity. These activities are unusual and such linking of external
identities should be investigated. In some cases you may see internal Entra ID sync accounts (Sync_) do this which may be benign'
Show query
AuditLogs | where OperationName=~ "Update user" | where Result =~ "success" | mv-expand TargetResources | mv-expand TargetResources.modifiedProperties | extend displayName = tostring(TargetResources_modifiedProperties.displayName), TargetUPN_oldValue = tostring(parse_json(tostring(TargetResources_modifiedProperties.oldValue))[0]), TargetUPN_newValue = tostring(parse_json(tostring(TargetResources_modifiedProperties.newValue))[0]) | where displayName == "UserPrincipalName" and TargetUPN_oldValue !has "#EXT" and TargetUPN_newValue has "#EXT" | extend InitiatingAppName = tostring(InitiatedBy.app.displayName) | extend InitiatingAppServicePrincipalId = tostring(InitiatedBy.app.servicePrincipalId) | extend InitiatingUserPrincipalName = tostring(InitiatedBy.user.userPrincipalName) | extend InitiatingAadUserId = tostring(InitiatedBy.user.id) | extend InitiatingIPAddress = tostring(InitiatedBy.user.ipAddress) | extend InitiatedBy = tostring(iff(isnotempty(InitiatingUserPrincipalName),InitiatingUserPrincipalName, InitiatingAppName)) | summarize arg_max(TimeGenerated, *) by CorrelationId | project-reorder TimeGenerated, InitiatedBy, InitiatingAppName, InitiatingAppServicePrincipalId, InitiatingUserPrincipalName, InitiatingAadUserId, InitiatingIPAddress, TargetUPN_oldValue, TargetUPN_newValue | extend InitiatingAccountName = tostring(split(InitiatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(InitiatingUserPrincipalName, "@")[1]) | extend TargetAccountName = tostring(split(TargetUPN_oldValue, "@")[0]), TargetUPNSuffix = tostring(split(TargetUPN_oldValue, "@")[1])
Suspicious modification of Global Administrator user properties
'This query will detect if user properties of Global Administrator are updated by an existing user. Usually only user administrator or other global administrator can update such properties.
Investigate if such user change is an attempt to elevate an existing low privileged identity or rogue administrator activity'
Show query
let query_frequency = 1h;
let query_period = 14d;
IdentityInfo
| where TimeGenerated > ago(query_period)
| where set_has_element(AssignedRoles, "Global Administrator")
| distinct AccountUPN, AccountObjectId
| join kind=inner (
AuditLogs
| where TimeGenerated > ago(query_frequency)
| where OperationName=~ "Update user" and Result =~ "success"
// | where isnotempty(InitiatedBy["user"])
| mv-expand TargetResource = TargetResources
| where TargetResource["type"] == "User"
| extend AccountObjectId = tostring(TargetResource["id"])
| where tostring(TargetResource["modifiedProperties"]) != "[]"
| mv-apply modifiedProperty = TargetResource["modifiedProperties"] on (
summarize modifiedProperties = make_bag(
bag_pack(tostring(modifiedProperty["displayName"]),
bag_pack("oldValue", trim(@'[\"\s]+', tostring(modifiedProperty["oldValue"])),
"newValue", trim(@'[\"\s]+', tostring(modifiedProperty["newValue"])))))
)
| where not(tostring(modifiedProperties["Included Updated Properties"]["newValue"]) in ("LastDirSyncTime", ""))
| where not(tostring(modifiedProperties["Included Updated Properties"]["newValue"]) == "StrongAuthenticationPhoneAppDetail" and isnotempty(modifiedProperties["StrongAuthenticationPhoneAppDetail"]) and tostring(array_sort_asc(extract_all(@'\"Id\"\:\"([^\"]+)\"', tostring(modifiedProperties["StrongAuthenticationPhoneAppDetail"]["newValue"])))) == tostring(array_sort_asc(extract_all(@'\"Id\"\:\"([^\"]+)\"', tostring(modifiedProperties["StrongAuthenticationPhoneAppDetail"]["oldValue"])))))
| extend
Initiator = iif(isnotempty(InitiatedBy["app"]), tostring(InitiatedBy["app"]["displayName"]), tostring(InitiatedBy["user"]["userPrincipalName"])),
InitiatorId = iif(isnotempty(InitiatedBy["app"]), tostring(InitiatedBy["app"]["servicePrincipalId"]), tostring(InitiatedBy["user"]["id"])),
IPAddress = tostring(InitiatedBy[tostring(bag_keys(InitiatedBy)[0])]["ipAddress"])
) on AccountObjectId
| project TimeGenerated, Category, Identity, Initiator, IPAddress, OperationName, Result, AccountUPN, InitiatedBy, AdditionalDetails, TargetResources, AccountObjectId, InitiatorId, CorrelationId
| extend
InitiatorName = tostring(split(Initiator, "@")[0]),
InitiatorUPNSuffix = tostring(split(Initiator, "@")[1]),
AccountName = tostring(split(AccountUPN, "@")[0]),
AccountUPNSuffix = tostring(split(AccountUPN, "@")[1])
URL Added to Application from Unknown Domain
'Detects a URL being added to an application where the domain is not one that is associated with the tenant.
The query uses domains seen in sign in logs to determine if the domain is associated with the tenant.
Applications associated with URLs not controlled by the organization can pose a security risk.
Ref: https://learn.microsoft.com/en-gb/entra/architecture/security-operations-applications#application-configuration-changes'
Show query
let domains =
SigninLogs
| where ResultType == 0
| extend domain = split(UserPrincipalName, "@")[1]
| extend domain = tostring(split(UserPrincipalName, "@")[1])
| summarize by tolower(tostring(domain));
AuditLogs
| where Category =~ "ApplicationManagement"
| where Result =~ "success"
| where OperationName =~ 'Update Application'
| mv-expand TargetResources
| mv-expand TargetResources.modifiedProperties
| where TargetResources_modifiedProperties.displayName =~ "AppAddress"
| extend Key = tostring(TargetResources_modifiedProperties.displayName)
| extend NewValue = TargetResources_modifiedProperties.newValue
| extend OldValue = TargetResources_modifiedProperties.oldValue
| where isnotempty(Key) and isnotempty(NewValue)
| project-reorder Key, NewValue, OldValue
| extend NewUrls = extract_all('"Address":([^,]*)', tostring(NewValue))
| extend OldUrls = extract_all('"Address":([^,]*)', tostring(OldValue))
| extend AddedUrls = set_difference(NewUrls, OldUrls)
| where array_length(AddedUrls) > 0
| extend UserAgent = iif(tostring(AdditionalDetails[0].key) == "User-Agent", tostring(AdditionalDetails[0].value), "")
| extend InitiatingAppName = tostring(InitiatedBy.app.displayName)
| extend InitiatingAppServicePrincipalId = tostring(InitiatedBy.app.servicePrincipalId)
| extend InitiatingUserPrincipalName = tostring(InitiatedBy.user.userPrincipalName)
| extend InitiatingAadUserId = tostring(InitiatedBy.user.id)
| extend InitiatingIPAddress = tostring(InitiatedBy.user.ipAddress)
| extend InitiatedBy = tostring(iff(isnotempty(InitiatingUserPrincipalName),InitiatingUserPrincipalName, InitiatingAppName))
| extend AppDisplayName = tostring(TargetResources.displayName)
| where isnotempty(AddedUrls)
| mv-expand AddedUrls
| extend AddedUrls = trim(@'"', tostring(AddedUrls))
| extend Domain = extract("^(?:https?:\\/\\/)?(?:[^@\\/\\n]+@)?(?:www\\.)?([^:\\/?\\n]+)/", 1, replace_string(tolower(AddedUrls), '"', ""))
| where isnotempty(Domain)
| extend Domain = strcat(split(Domain, ".")[-2], ".", split(Domain, ".")[-1])
| where Domain !in (domains)
| project-reorder TimeGenerated, AppDisplayName, AddedUrls, InitiatedBy, UserAgent, InitiatingIPAddress
| extend InitiatingAccountName = tostring(split(InitiatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(InitiatingUserPrincipalName, "@")[1])
Unusual identity creation using exchange powershell
' The query below identifies creation of unusual identity by the Europium actor to mimic Microsoft Exchange Health Manager Service account using Exchange PowerShell commands
Reference: https://www.microsoft.com/security/blog/2022/09/08/microsoft-investigates-iranian-attacks-against-the-albanian-government/'
Show query
(union isfuzzy=true
(SecurityEvent
| where EventID==4688
| where CommandLine has_any ("New-Mailbox","Update-RoleGroupMember") and CommandLine has "HealthMailbox55x2yq"
| project TimeGenerated, DeviceName = Computer, AccountName = SubjectUserName, AccountDomain = SubjectDomainName, ProcessName, ProcessNameFullPath = NewProcessName, EventID, Activity, CommandLine, EventSourceName, Type
| extend InitiatingProcessAccount = strcat(AccountDomain, "\\", AccountName)
),
(DeviceProcessEvents
| where ProcessCommandLine has_any ("New-Mailbox","Update-RoleGroupMember") and ProcessCommandLine has "HealthMailbox55x2yq"
| extend timestamp = TimeGenerated, AccountDomain = InitiatingProcessAccountDomain, AccountName = InitiatingProcessAccountName
| extend InitiatingProcessAccount = strcat(AccountDomain, "\\", AccountName)
)
)
| extend HostName = tostring(split(DeviceName, ".")[0]), DomainIndex = toint(indexof(DeviceName, '.'))
| extend HostNameDomain = iff(DomainIndex != -1, substring(DeviceName, DomainIndex + 1), DeviceName)
User Account Created Using Incorrect Naming Format
'This query looks for accounts being created where the name does not match a defined pattern.
Attackers may attempt to add accounts as a means of establishing persistant access to an environment, looking for anomalies in created accounts may help identify illegitimately created accounts.
Created accounts should be investigated to ensure they were legitimated created.
The user_regex field in the query needs to be populated with the expected pattern for the environment before deployment.
R
Show query
// Add the environments expected username format regex below before deploying let user_regex = ""; AuditLogs | where OperationName =~ "Add user" | where Result =~ "success" | extend userAgent = tostring(AdditionalDetails[0].value) | extend InitiatingAppName = tostring(InitiatedBy.app.displayName) | extend InitiatingAppServicePrincipalId = tostring(InitiatedBy.app.servicePrincipalId) | extend InitiatingUserPrincipalName = tostring(InitiatedBy.user.userPrincipalName) | extend InitiatingAadUserId = tostring(InitiatedBy.user.id) | extend InitiatingIPAddress = tostring(InitiatedBy.user.ipAddress) | extend InitiatedBy = tostring(iff(isnotempty(InitiatingUserPrincipalName),InitiatingUserPrincipalName, InitiatingAppName)) | extend AddedUser = tostring(TargetResources[0].userPrincipalName) | where AddedUser matches regex user_regex | extend InitiatingAccountName = tostring(split(InitiatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(InitiatingUserPrincipalName, "@")[1]) | extend TargetAccountName = tostring(split(AddedUser, "@")[0]), TargetAccountUPNSuffix = tostring(split(AddedUser, "@")[1])
User account created without expected attributes defined
'This query looks for accounts being created that do not have attributes populated that are commonly populated in the tenant.
Attackers may attempt to add accounts as a means of establishing persistant access to an environment, looking for anomalies in created accounts may help identify illegitimately created accounts.
Created accounts should be investigated to ensure they were legitimated created.
Ref: https://docs.microsoft.com/azure/active-directory/fundamentals/security-operations-user
Show query
let threshold = 10;
let default_ad_attributes = dynamic(["LastDirSyncTime", "StsRefreshTokensValidFrom", "Included Updated Properties", "AccountEnabled", "Action Client Name", "SourceAnchor"]);
let addUsers = AuditLogs
| where OperationName =~ "Add user"
| where Result =~ "success"
| extend AccountProperties = TargetResources[0].modifiedProperties
| mv-expand AccountProperties
;
addUsers
| evaluate bag_unpack(AccountProperties) : (displayName:string, oldValue: string, newValue: string , TenantId : string, SourceSystem : string, TimeGenerated : datetime, ResourceId : string, OperationName : string, OperationVersion : string, Category : string, ResultType : string, ResultSignature : string, ResultDescription : string, DurationMs : long, CorrelationId : string, Resource : string, ResourceGroup : string, ResourceProvider : string, Identity : string, Level : string, Location : string, AdditionalDetails : dynamic, Id : string, InitiatedBy : dynamic, LoggedByService : string, Result : string, ResultReason : string, TargetResources : dynamic, AADTenantId : string, ActivityDisplayName : string, ActivityDateTime : datetime, AADOperationType : string, Type : string)
| extend displayName = column_ifexists("displayName", "Unknown Value")
| summarize count() by displayName, TenantId
| where displayName !in (default_ad_attributes)
| top threshold by count_ desc
| summarize make_set(displayName) by TenantId
| join kind=inner (
addUsers
| extend CreatingUserPrincipalName = tostring(parse_json(tostring(InitiatedBy.user)).userPrincipalName)
| extend CreatingAadUserId = tostring(InitiatedBy.user.id)
| extend CreatingUserIPAddress = tostring(InitiatedBy.user.ipAddress)
| extend CreatedUserPrincipalName = tostring(TargetResources[0].userPrincipalName)
| extend PropName = tostring(AccountProperties.displayName))
on TenantId
| summarize makeset(PropName) by TimeGenerated, CorrelationId, CreatedUserPrincipalName, CreatingUserPrincipalName, CreatingAadUserId, CreatingUserIPAddress, tostring(set_displayName)
| extend missing_props = set_difference(todynamic(set_displayName), set_PropName)
| where array_length(missing_props) > 0
| join kind=innerunique (
AuditLogs
| where Result =~ "success"
| where OperationName =~ "Add user"
| extend CreatedUserPrincipalName = tostring(TargetResources[0].userPrincipalName))
on CorrelationId, CreatedUserPrincipalName
| extend ExpectedProperties = set_displayName
| project-away set_displayName, set_PropName
| extend InitiatingAccountName = tostring(split(CreatingUserPrincipalName, "@")[0]), InitiatingAccountUPNSuffix = tostring(split(CreatingUserPrincipalName, "@")[1])
| extend TargetAccountName = tostring(split(CreatedUserPrincipalName, "@")[0]), TargetAccountUPNSuffix = tostring(split(CreatedUserPrincipalName, "@")[1])
User joining Zoom meeting from suspicious timezone
'The alert shows users that join a Zoom meeting from a time zone other than the one the meeting was created in.
You can also whitelist known good time zones in the tz_whitelist value using the tz database name format https://en.wikipedia.org/wiki/List_of_tz_database_time_zones'
Show query
let schedule_lookback = 14d; let join_lookback = 1d; // If you want to whitelist specific timezones include them in a list here let tz_whitelist = dynamic([]); let meetings = ( ZoomLogs | where TimeGenerated >= ago(schedule_lookback) | where Event =~ "meeting.created" | extend MeetingId = tostring(parse_json(MeetingEvents).MeetingId) | extend SchedTimezone = tostring(parse_json(MeetingEvents).Timezone)); ZoomLogs | where TimeGenerated >= ago(join_lookback) | where Event =~ "meeting.participant_joined" | extend JoinedTimeZone = tostring(parse_json(MeetingEvents).Timezone) | extend MeetingName = tostring(parse_json(MeetingEvents).MeetingName) | extend MeetingId = tostring(parse_json(MeetingEvents).MeetingId) | where JoinedTimeZone !in (tz_whitelist) | join (meetings) on MeetingId | where SchedTimezone != JoinedTimeZone | project TimeGenerated, MeetingName, JoiningUser=payload_object_participant_user_name_s, JoinedTimeZone, SchedTimezone, MeetingScheduler=User1 | extend AccountName = tostring(split(JoiningUser, "@")[0]), AccountUPNSuffix = tostring(split(JoiningUser, "@")[1])
User login from different countries within 3 hours (Uses Authentication Normalization)
'This query searches for successful user logins from different countries within 3 hours.
To use this analytics rule, make sure you have deployed the [ASIM normalization parsers](https://aka.ms/ASimAuthentication)'
Show query
let timeframe = ago(3h);
let threshold = 2;
imAuthentication
| where TimeGenerated > timeframe
| where EventType == 'Logon'
and EventResult == 'Success'
| where isnotempty(SrcGeoCountry)
| summarize
StartTime = min(TimeGenerated)
, EndTime = max(TimeGenerated)
, Vendors = make_set(EventVendor, 128)
, Products = make_set(EventProduct, 128)
, NumOfCountries = dcount(SrcGeoCountry)
, Countries = make_set(SrcGeoCountry, 128)
by TargetUserId, TargetUsername, TargetUserType
| where NumOfCountries >= threshold
| where TargetUserType !in ("Application", "Service", "System", "Other", "Machine", "ServicePrincipal")
| extend
Name = iif(
TargetUsername contains "@"
, tostring(split(TargetUsername, '@', 0)[0])
, TargetUsername
),
UPNSuffix = iif(
TargetUsername contains "@"
, tostring(split(TargetUsername, '@', 1)[0])
, ""
)
Vulnerable Machines related to OMIGOD CVE-2021-38647
'This query uses the Azure Defender Security Nested Recommendations data to find machines vulnerable to OMIGOD CVE-2021-38647.
OMI is the Linux equivalent of Windows WMI and helps users manage configurations across remote and local environments. The query aims to find machines that have this OMI vulnerability (CVE-2021-38647).
Security Nested Recommendations data is sent to Microsoft Sentinel using the continuous export feature of Azure Defender(refrence link below).
Reference: https://www.wiz
Show query
SecurityNestedRecommendation | where RemediationDescription has 'CVE-2021-38647' | parse ResourceDetails with * 'virtualMachines/' VirtualMAchine '"' * | summarize arg_min(TimeGenerated, *) by TenantId, RecommendationSubscriptionId, VirtualMAchine, RecommendationName,Description,RemediationDescription, tostring(AdditionalData),VulnerabilityId | extend HostName = tostring(split(VirtualMAchine, ".")[0]), DomainIndex = toint(indexof(VirtualMAchine, '.')) | extend HostNameDomain = iff(DomainIndex != -1, substring(VirtualMAchine, DomainIndex + 1), VirtualMAchine)
Windows host username encoded in base64 web request
'This detection will identify network requests in HTTP proxy data that contains Base64 encoded usernames from machines in the DeviceEvents table.
This technique was seen usee by POLONIUM in their RunningRAT tool.'
Show query
let accountLookback = 3d;
let requestLookback = 3d;
let extraction_regex = @"(?:\?|&)[a-zA-Z0-9\%]*=([a-zA-Z0-9\/\+\=]*)";
// Collect account names and base64 encode them
DeviceEvents
| where TimeGenerated > ago(accountLookback)
| summarize make_set(DeviceId), make_set(DeviceName) by InitiatingProcessAccountName
| where isnotempty(InitiatingProcessAccountName)
| extend base64_user = base64_encode_tostring(InitiatingProcessAccountName)
| join (
// Collect requests and extract base64 parameters
CommonSecurityLog
| where TimeGenerated > ago(requestLookback)
| where isnotempty(RequestURL)
// Summarize early on the RequestURL
| summarize FirstRequest=min(TimeGenerated), LastRequest=max(TimeGenerated), NumberOfRequests=count() by RequestURL
| extend base64_candidate = extract_all(extraction_regex, RequestURL)
| mv-expand base64_candidate to typeof(string)
) on $left.base64_user == $right.base64_candidate
| project FirstRequest, LastRequest, NumberOfRequests, RequestURL, DeviceIds=set_DeviceId, DeviceNames=set_DeviceName, UserName=InitiatingProcessAccountName
Workspace deletion activity from an infected device
'This query will alert on any sign-ins from devices infected with malware in correlation with workspace deletion activity.
Attackers may attempt to delete workspaces containing compute instances after successful compromise to cause service unavailability to regular business operation.'
Show query
SecurityAlert
| where TimeGenerated > ago(1d)
| where ProductName == "Azure Active Directory Identity Protection"
| where AlertName == "Sign-in from an infected device"
| mv-apply EntityAccount=todynamic(Entities) on
(
where EntityAccount.Type == "account"
| extend AadTenantId = tostring(EntityAccount.AadTenantId), AadUserId = tostring(EntityAccount.AadUserId)
)
| mv-apply EntityIp=todynamic(Entities) on
(
where EntityIp.Type == "ip"
| extend IpAddress = tostring(EntityIp.Address)
)
| join kind=inner (
IdentityInfo
| distinct AccountTenantId, AccountObjectId, AccountUPN, AccountDisplayName
| extend UserAccount = AccountUPN
| extend UserName = AccountDisplayName
| where isnotempty(AccountDisplayName) and isnotempty(UserAccount)
| project AccountTenantId, AccountObjectId, UserAccount, UserName
)
on
$left.AadTenantId == $right.AccountTenantId,
$left.AadUserId == $right.AccountObjectId
| extend CompromisedEntity = iff(CompromisedEntity == "N/A" or isempty(CompromisedEntity), UserAccount, CompromisedEntity)
| project AlertName, AlertSeverity, CompromisedEntity, UserAccount, IpAddress, TimeGenerated, UserName
| join kind=inner
(
AzureActivity
| where OperationNameValue has_any ("/workspaces/computes/delete", "workspaces/delete")
| where ActivityStatusValue has_any ("Succeeded", "Success")
| project TimeGenerated, ResourceProviderValue, _ResourceId, SubscriptionId, UserAccount=Caller, IpAddress=CallerIpAddress, CorrelationId, OperationId, ResourceGroup, TenantId
) on IpAddress, UserAccount
| extend AccountName = tostring(split(UserAccount, "@")[0]), AccountUPNSuffix = tostring(split(UserAccount, "@")[1])
3CX Supply Chain Attack Network Indicators
The following analytic identifies DNS queries to domains associated with the 3CX supply chain attack. It leverages the Network_Resolution datamodel to detect these suspicious domain indicators. This activity is significant because it can indicate a potential compromise stemming from the 3CX supply chain attack, which is known for distributing malicious software through trusted updates. If confirmed malicious, this activity could allow attackers to establish a foothold in the network, exfiltrate sensitive data, or further propagate malware, leading to extensive damage and data breaches.
Show query
| tstats `security_content_summariesonly`
count min(_time) as firstTime
max(_time) as lastTime
from datamodel=Network_Resolution where
DNS.query=*
NOT DNS.query IN ("-", "unknown")
by DNS.answer DNS.answer_count DNS.query
DNS.query_count DNS.reply_code_id DNS.src
DNS.vendor_product
| `drop_dm_object_name(DNS)`
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| lookup 3cx_ioc_domains domain as query OUTPUT Description isIOC
| search isIOC=true
| `3cx_supply_chain_attack_network_indicators_filter`
ASL AWS Create Access Key
The following analytic identifies the creation of AWS IAM access keys by a user for another user, which can indicate privilege escalation. It leverages AWS CloudTrail logs to detect instances where the user creating the access key is different from the user for whom the key is created. This activity is significant because unauthorized access key creation can allow attackers to establish persistence or exfiltrate data via AWS APIs. If confirmed malicious, this could lead to unauthorized access to AWS services, data exfiltration, and long-term persistence in the environment.
Show query
`amazon_security_lake` api.operation=CreateAccessKey
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_create_access_key_filter`ASL AWS Create Policy Version to allow all resources
The following analytic identifies the creation of a new AWS IAM policy version that allows access to all resources. It detects this activity by analyzing AWS CloudTrail logs for the CreatePolicyVersion event with a policy document that grants broad permissions. This behavior is significant because it violates the principle of least privilege, potentially exposing the environment to misuse or abuse. If confirmed malicious, an attacker could gain extensive access to AWS resources, leading to unauthorized actions, data exfiltration, or further compromise of the AWS environment.
Show query
`amazon_security_lake` api.operation=CreatePolicy
| spath input=api.request.data
| spath input=policyDocument
| regex Statement{}.Action="\*"
| regex Statement{}.Resource="\*"
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region api.request.data
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_create_policy_version_to_allow_all_resources_filter`ASL AWS Defense Evasion PutBucketLifecycle
The following analytic detects `PutBucketLifecycle` events in AWS CloudTrail logs where a user sets a lifecycle rule for an S3 bucket with an expiration period of fewer than three days. This detection leverages CloudTrail logs to identify suspicious lifecycle configurations. This activity is significant because attackers may use it to delete CloudTrail logs quickly, thereby evading detection and impairing forensic investigations. If confirmed malicious, this could allow attackers to cover their tracks, making it difficult to trace their actions and respond to the breach effectively.
Show query
`amazon_security_lake` api.operation=PutBucketLifecycle
| spath input=api.request.data path=LifecycleConfiguration.Rule.NoncurrentVersionExpiration.NoncurrentDays output=NoncurrentDays
| where NoncurrentDays < 3
| spath input=api.request.data
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region NoncurrentDays
bucketName
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_defense_evasion_putbucketlifecycle_filter`ASL AWS Detect Users creating keys with encrypt policy without MFA
The following analytic detects the creation of AWS KMS keys with an encryption policy accessible to everyone, including external entities. It leverages AWS CloudTrail logs from Amazon Security Lake to identify `CreateKey` or `PutKeyPolicy` events where the `kms:Encrypt` action is granted to all principals. This activity is significant as it may indicate a compromised account, allowing an attacker to misuse the encryption key to target other organizations. If confirmed malicious, this could lead to unauthorized data encryption, potentially disrupting operations and compromising sensitive information across multiple entities.
Show query
`amazon_security_lake` api.operation=PutKeyPolicy OR api.operation=CreateKey
| spath input=api.request.data path=policy output=policy
| spath input=policy
| rename Statement{}.Action as Action, Statement{}.Principal as Principal
| eval Statement=mvzip(Action,Principal,"
| ")
| mvexpand Statement
| eval action=mvindex(split(Statement, "
| "), 0)
| eval principal=mvindex(split(Statement, "
| "), 1)
| search action=kms*
| regex principal="\*"
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region api.request.data
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_detect_users_creating_keys_with_encrypt_policy_without_mfa_filter`ASL AWS IAM Successful Group Deletion
The following analytic detects the successful deletion of a group within AWS IAM, leveraging CloudTrail IAM events. This action, while not inherently malicious, can serve as a precursor to more sinister activities, such as unauthorized access or privilege escalation attempts. By monitoring for such deletions, the analytic aids in identifying potential preparatory steps towards an attack, allowing for early detection and mitigation. The identification of this behavior is crucial for a SOC to prevent the potential impact of an attack, which could include unauthorized access to sensitive resources or disruption of AWS environment operations.
Show query
`amazon_security_lake` api.operation=DeleteGroup status=Success
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_iam_successful_group_deletion_filter`ASL AWS SAML Update identity provider
The following analytic detects updates to the SAML provider in AWS. It leverages AWS CloudTrail logs to identify the `UpdateSAMLProvider` event, analyzing fields such as `sAMLProviderArn`, `sourceIPAddress`, and `userIdentity` details. Monitoring updates to the SAML provider is crucial as it may indicate a perimeter compromise of federated credentials or unauthorized backdoor access set by an attacker. If confirmed malicious, this activity could allow attackers to manipulate identity federation, potentially leading to unauthorized access to cloud resources and sensitive data.
Show query
`amazon_security_lake` api.operation=UpdateSAMLProvider
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_saml_update_identity_provider_filter`ASL AWS UpdateLoginProfile
The following analytic detects an AWS CloudTrail event where a user with permissions updates the login profile of another user. It leverages CloudTrail logs to identify instances where the user making the change is different from the user whose profile is being updated. This activity is significant because it can indicate privilege escalation attempts, where an attacker uses a compromised account to gain higher privileges. If confirmed malicious, this could allow the attacker to escalate their privileges, potentially leading to unauthorized access and control over sensitive resources within the AWS environment.
Show query
`amazon_security_lake` api.operation=UpdateLoginProfile
| fillnull
| stats count min(_time) as firstTime max(_time) as lastTime
BY actor.user.uid api.operation api.service.name
http_request.user_agent src_endpoint.ip actor.user.account.uid
cloud.provider cloud.region
| rename actor.user.uid as user api.operation as action api.service.name as dest http_request.user_agent as user_agent src_endpoint.ip as src actor.user.account.uid as vendor_account cloud.provider as vendor_product cloud.region as vendor_region
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `asl_aws_updateloginprofile_filter`AWS Bedrock Delete Knowledge Base
The following analytic identifies attempts to delete AWS Bedrock Knowledge Bases, which are resources that store and manage domain-specific information for AI models. It monitors AWS CloudTrail logs for DeleteKnowledgeBase API calls. This activity could indicate an adversary attempting to remove knowledge bases after compromising credentials, potentially to disrupt business operations or remove traces of data access. Deleting knowledge bases could impact model performance, remove critical business context, or be part of a larger attack to degrade AI capabilities. If confirmed malicious, this could represent a deliberate attempt to cause service disruption or data loss.
Show query
`cloudtrail` eventSource=bedrock.amazonaws.com eventName=DeleteKnowledgeBase | rename user_name as user | stats count min(_time) as firstTime max(_time) as lastTime values(requestParameters.knowledgeBaseId) as knowledgeBaseIds by src user user_agent vendor_account vendor_product dest signature vendor_region | `security_content_ctime(firstTime)` | `security_content_ctime(lastTime)` | `aws_bedrock_delete_knowledge_base_filter`
AWS Create Policy Version to allow all resources
The following analytic identifies the creation of a new AWS IAM policy version that allows access to all resources. It detects this activity by analyzing AWS CloudTrail logs for the CreatePolicyVersion event with a policy document that grants broad permissions. This behavior is significant because it violates the principle of least privilege, potentially exposing the environment to misuse or abuse. If confirmed malicious, an attacker could gain extensive access to AWS resources, leading to unauthorized actions, data exfiltration, or further compromise of the AWS environment.
Show query
`cloudtrail` eventName=CreatePolicyVersion eventSource = iam.amazonaws.com errorCode = success
| spath input=requestParameters.policyDocument output=key_policy_statements path=Statement{}
| mvexpand key_policy_statements
| spath input=key_policy_statements output=key_policy_action_1 path=Action
| where key_policy_action_1 = "*"
| rename user_name as user
| stats count min(_time) as firstTime max(_time) as lastTime values(key_policy_statements) as policy_added
BY signature dest user
user_agent src vendor_account
vendor_region vendor_product
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `aws_create_policy_version_to_allow_all_resources_filter`AWS CreateAccessKey
The following analytic identifies the creation of AWS IAM access keys by a user for another user, which can indicate privilege escalation. It leverages AWS CloudTrail logs to detect instances where the user creating the access key is different from the user for whom the key is created. This activity is significant because unauthorized access key creation can allow attackers to establish persistence or exfiltrate data via AWS APIs. If confirmed malicious, this could lead to unauthorized access to AWS services, data exfiltration, and long-term persistence in the environment.
Show query
`cloudtrail` eventName = CreateAccessKey userAgent !=console.amazonaws.com errorCode = success
| eval match=if(match(userIdentity.userName,requestParameters.userName),1,0)
| search match=0
| rename user_name as user
| stats count min(_time) as firstTime max(_time) as lastTime
BY signature dest user
user_agent src vendor_account
vendor_region vendor_product
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `aws_createaccesskey_filter`AWS CreateLoginProfile
The following analytic identifies the creation of a login profile for one AWS user by another, followed by a console login from the same source IP. It uses AWS CloudTrail logs to correlate the `CreateLoginProfile` and `ConsoleLogin` events based on the source IP and user identity. This activity is significant as it may indicate privilege escalation, where an attacker creates a new login profile to gain unauthorized access. If confirmed malicious, this could allow the attacker to escalate privileges and maintain persistent access to the AWS environment.
Show query
`cloudtrail` eventName = CreateLoginProfile
| rename requestParameters.userName as new_login_profile
| table src_ip eventName new_login_profile userIdentity.userName
| join new_login_profile src_ip [
| search `cloudtrail` eventName = ConsoleLogin
| rename userIdentity.userName as new_login_profile
| stats count values(eventName) min(_time) as firstTime max(_time) as lastTime
BY eventSource aws_account_id errorCode
user_agent eventID awsRegion
userIdentity.principalId user_arn new_login_profile
src_ip dest vendor_account
vendor_region vendor_product
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`]
| rename user_arn as user
| `aws_createloginprofile_filter`AWS Defense Evasion PutBucketLifecycle
The following analytic detects `PutBucketLifecycle` events in AWS CloudTrail logs where a user sets a lifecycle rule for an S3 bucket with an expiration period of fewer than three days. This detection leverages CloudTrail logs to identify suspicious lifecycle configurations. This activity is significant because attackers may use it to delete CloudTrail logs quickly, thereby evading detection and impairing forensic investigations. If confirmed malicious, this could allow attackers to cover their tracks, making it difficult to trace their actions and respond to the breach effectively.
Show query
`cloudtrail` eventName=PutBucketLifecycle user_type=IAMUser errorCode=success
| spath path=requestParameters{}.LifecycleConfiguration{}.Rule{}.Expiration{}.Days output=expiration_days
| spath path=requestParameters{}.bucketName output=bucket_name
| rename user_name as user
| stats count min(_time) as firstTime max(_time) as lastTime
BY signature dest user
user_agent src vendor_account
vendor_region vendor_product bucket_name
expiration_days
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `aws_defense_evasion_putbucketlifecycle_filter`AWS Detect Users creating keys with encrypt policy without MFA
The following analytic detects the creation of AWS KMS keys with an encryption policy accessible to everyone, including external entities. It leverages AWS CloudTrail logs to identify `CreateKey` or `PutKeyPolicy` events where the `kms:Encrypt` action is granted to all principals. This activity is significant as it may indicate a compromised account, allowing an attacker to misuse the encryption key to target other organizations. If confirmed malicious, this could lead to unauthorized data encryption, potentially disrupting operations and compromising sensitive information across multiple entities.
Show query
`cloudtrail` eventName=CreateKey OR eventName=PutKeyPolicy
| spath input=requestParameters.policy output=key_policy_statements path=Statement{}
| mvexpand key_policy_statements
| spath input=key_policy_statements output=key_policy_action_1 path=Action
| spath input=key_policy_statements output=key_policy_action_2 path=Action{}
| eval key_policy_action=mvappend(key_policy_action_1,key_policy_action_2)
| spath input=key_policy_statements output=key_policy_principal path=Principal.AWS
| search key_policy_action="kms:Encrypt" AND key_policy_principal="*"
| rename user_name as user
| stats count min(_time) as firstTime max(_time) as lastTime
BY signature dest user
user_agent src vendor_account
vendor_region vendor_product key_policy_action
key_policy_principal
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `aws_detect_users_creating_keys_with_encrypt_policy_without_mfa_filter`AWS Detect Users with KMS keys performing encryption S3
The following analytic identifies users with KMS keys performing encryption operations on S3 buckets. It leverages AWS CloudTrail logs to detect the `CopyObject` event where server-side encryption with AWS KMS is specified. This activity is significant as it may indicate unauthorized or suspicious encryption of data, potentially masking exfiltration or tampering efforts. If confirmed malicious, an attacker could be encrypting sensitive data to evade detection or preparing it for exfiltration, posing a significant risk to data integrity and confidentiality.
Show query
`cloudtrail` eventName=CopyObject requestParameters.x-amz-server-side-encryption="aws:kms"
| rename requestParameters.bucketName AS bucketName, requestParameters.x-amz-copy-source AS src_file, requestParameters.key AS dest_file
| rename user_name as user
| stats count min(_time) as firstTime max(_time) as lastTime
BY signature dest user
user_agent src vendor_account
vendor_region vendor_product bucketName
src_file dest_file
| `security_content_ctime(firstTime)`
| `security_content_ctime(lastTime)`
| `aws_detect_users_with_kms_keys_performing_encryption_s3_filter`Showing 101-150 of 930