A.8.16 Technological
Monitoring activities
Structured attributes from ISO/IEC 27002:2022 — control type · CIA properties · cybersecurity concept · operational capability · security domain. What do these mean?
Mapped NIST 800-53 r5 controls (21)
Our reading (mostly llm_verified) of how this ISO control and each NIST 800-53 control relate. Not an ISO or NIST product.
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- AU-12mostlycovers — A.8.16's monitoring activities to detect anomalies and incidents directly accounts for the bulk of AU-12's audit generation, selection, and content requirements (as the operational mechanism that consumes and acts on those records), but a real residual remains in AU-12's upstream definition and configuration aspects that sit in AU-2.
- AU-2mostlycovers — A.8.16's requirement to detect anomalous behaviour and incidents is substantially met by au-2's identification, coordination, and specification of auditable events that directly support anomaly and incident detection, but a residual of au-2's narrower audit-function focus (e.g., exact parameter selection mechanics) sits outside A.8.16's broader monitoring intent.
- AU-6mostlycovers — A.8.16's mandate to monitor for anomalous behaviour and potential incidents directly accounts for the core of AU-6's review/analysis/reporting of audit records for indications of inappropriate/unusual activity and risk-based adjustment; residual exists in AU-6's explicit law-enforcement-informed trigger and specific reporting recipients not required by A.8.16.
- CA-7mostlycovers — A.8.16's requirement to detect anomalous behaviour and potential incidents accounts for the bulk of CA-7's continuous monitoring implementation and metrics, but leaves a residual on the organization-level strategy development and formal assessment frequencies that sit outside pure detection activities.
- SI-4mostlycovers — A.8.16's broad mandate to monitor for anomalous behaviour and potential incidents accounts for the bulk of SI-4's detection objectives (attacks, unauthorized connections/use), but leaves a residual slice of specific techniques, methods, and invocation/deployment details uncovered.
Aligned NIST CSF 2.0 outcomes (19)
NIST CSF 2.0 outcomes this ISO control aligns with — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- DE.CM-01mostlycovers — The ISO control's emphasis on continuous, baseline-driven monitoring of network traffic, system activity, and security-tool logs directly fulfills the CSF outcome of monitoring networks and services to detect potentially adverse events.
- DE.CM-03mostlycovers — The ISO control's focus on tracking user access patterns, authentication attempts, and deviations from normal behavior aligns with the CSF outcome of monitoring personnel activity and technology usage.
- DE.CM-09mostlycovers — By requiring monitoring of hardware/software runtime behavior, resource utilization, and code integrity, the ISO control satisfies the CSF outcome of monitoring computing environments and their data for adverse events.
- DE.AE-02domain mandateimplements — A.8.16's monitoring activities give operational effect to adverse-event analysis within the detection domain that DE.AE-02 names, but the CSF outcome does not cite monitoring specifically
- DE.AE-06domain mandateimplements — A.8.16's monitoring activities directly operationalize the provision of adverse-event information to staff/tools that DE.AE-06 requires; the link is within the detection/analysis domain but the CSF outcome does not name monitoring explicitly
- PR.PS-04domain mandateimplements — A.8.16's monitoring activities give operational effect to generating and exposing logs specifically for continuous monitoring (PR.PS-04), as logging is a core technical means within the monitoring domain, though not the only one and not explicitly named by the control.
Related OWASP ASVS 5.0 requirements (13)
Application-security verification requirements (OWASP ASVS 5.0) this ISO control aligns with; links open the ASVS chapter. Our analysis (mostly llm_verified) — many ISO controls have no ASVS counterpart.
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Related weaknesses / CWE (2420)
Weakness classes this ISO control helps prevent or mitigate — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- CWE-1050mostlyfinds — Monitoring can alert on resource abuse but does not eliminate the coding flaw.
- CWE-400mostlyfinds — Resource-utilization monitoring and alerting on bottlenecks or overloads limits the impact of denial-of-service or resource-exhaustion attacks.
- CWE-406mostlyfinds — Monitoring activities can detect anomalous traffic volumes, yet detection alone does not enforce control.
- CWE-770mostlyfinds — Baseline comparison of CPU, memory and bandwidth usage helps surface uncontrolled resource allocations before they cause service degradation.
- CWE-117partialfinds — Monitoring activities rely on trustworthy logs but do not ensure log message integrity.
- CWE-202partialfinds — Monitoring can identify anomalous query patterns indicative of inference attacks.
- CWE-223partialfinds — Monitoring activities rely on the very information whose absence defines the weakness.
- CWE-284partialfinds — Detection of unauthorized access attempts and deviations from expected access patterns reduces the window in which improper access control can be exploited without notice.
- CWE-360partialfinds — Monitoring activities may detect anomalies in event data but do not guarantee the trustworthiness of the source.
- CWE-385partialfinds — Continuous monitoring may detect timing-based exfiltration but does not eliminate the channel itself.
- CWE-390partialfinds — Monitoring activities can detect errors, yet without defined response procedures the weakness persists.
- CWE-407partialfinds — Monitoring activities can identify anomalous resource consumption indicative of algorithmic complexity attacks.
- CWE-409partialfinds — Monitoring can detect anomalous resource usage but does not prevent the weakness.
- CWE-506partialfinds — Checks that executed code has not been tampered with and monitoring for malware-associated activity reduce the likelihood that hidden malicious code remains active.
- CWE-507partialfinds — Monitoring activities can identify Trojan Horse behavior but do not stop its introduction.
- CWE-509partialfinds — Monitoring activities enable early detection of replicating malicious code.
- CWE-511partialfinds — Monitoring activities may detect anomalous behavior triggered by a logic/time bomb.
- CWE-515partialfinds — Monitoring can detect covert storage but does not eliminate the underlying weakness.
- CWE-69partialfinds — Monitoring activities can flag anomalous ADS usage, providing partial mitigation through detection rather than prevention.
- CWE-74partialfinds — Monitoring activities can identify active injection attacks after they occur.
- CWE-755partialfinds — Monitoring can surface unhandled exceptions but does not enforce proper handling.
- CWE-807partialfinds — Monitoring can detect exploitation but does not stop the underlying weakness.
- CWE-91partialfinds — Monitoring can detect exploitation but provides no preventive control over XML handling.
- CWE-912partialfinds — Monitoring activities can detect anomalous behavior from hidden functions but do not eliminate the weakness.
Mitigated MITRE ATT&CK techniques (4373)
Adversary techniques (MITRE ATT&CK Enterprise) this ISO control helps mitigate; links open attack.mitre.org. Our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- T1498fulldetects — A.8.16 explicitly requires monitoring for anomalous behaviour including known attack characteristics (e.g. denial of service), unusual system behaviour, bottlenecks/overload (latency, jitter, queuing), and traffic patterns, with real-time/continuous tools, baselines, and alerts; this surfaces Network DoS in flight for most cases, with the bounded remainder being attacks below detection thresholds or outside the scoped monitoring.
- T1498.002fulldetects — A.8.16 explicitly enumerates DoS among the known attack characteristics to monitor for and lists bottlenecks/overloads (network queuing, latency, jitter) and inbound/outbound network traffic in its scope; reflection amplification produces massive, signature-rich inbound traffic from many reflector sources that network monitoring surfaces directly, consistent with the A.8.16 vs T1498 parent anchor at `full`.
- T1003mostlydetects — A.8.16 explicitly lists monitoring of process injection, anomalous system behaviour, event logs, security tool outputs, resource use, and known attack patterns; T1003 (credential dumping) produces observable artifacts in exactly these categories (e.g. LSASS access, memory reads, Mimikatz-like patterns) that fall inside the defined baseline and alerting scope, with only a bounded remainder (stealthiest in-memory techniques on unmonitored hosts) outside.
- T1003mostlyresponds — A.8.16 surfaces anomalous behaviour including process injection, unusual system behaviour, unauthorized access attempts, and deviations from baseline (all observable indicators of credential dumping in flight); alerts feed directly into 5.26 response procedures for containment and eradication once the technique is underway, with the named remainder being stealthy dumps that produce no detectable anomaly before completion.
- T1003.001mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, access to critical processes, security-tool logs, event logs, resource use, and real-time anomaly detection against baseline, all of which surface LSASS dumping or SSP tampering in flight or post-execution on monitored Windows hosts; the named remainder is execution outside the defined monitoring scope.
- T1003.002mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, security-tool logs (AV/IDS/IPS), event logs, resource use, and deviations from baseline (including unusual access and known attack patterns); these surface in-memory SAM extraction via tools such as Mimikatz that rely on process injection or registry access, though some purely file-based offline extraction after the fact sits outside real-time monitoring scope.
- T1003.006mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, access to critical systems/servers, security tool logs (including IDS/IPS), event logs, and deviations from baseline (including unusual user/system behaviour), which surfaces the observable artifacts of a DCSync (replication API calls, anomalous DC access, lsadump-like behaviour) in most cases once it occurs on the monitored estate; the bounded remainder is pre-compromise reconnaissance or fully stealthy implementations that evade all listed indicators.
- T1018mostlydetects — A.8.16 explicitly monitors network traffic, access, configuration files, event logs, resource use, anomalous behaviour (including unusual scanning and deviations from baseline), and known attack patterns, which directly surfaces T1018's active (ping/net view/esxcli) and passive (hosts/ARP) discovery methods on covered platforms; the remainder is discovery occurring entirely outside the monitored estate or before any baseline is established.
- T1020mostlydetects — A.8.16 explicitly monitors outbound network traffic, anomalous behaviour patterns, known malicious domains/IPs, resource use, and deviations such as unusual data movement that match automated exfiltration signatures or baselines, surfacing the technique in flight on the organization's estate.
- T1021mostlydetects — Detection of anomalous remote access patterns and unauthorized attempts to protected resources can reveal lateral movement via remote services.
- T1021.001mostlydetects — A.8.16 explicitly lists monitoring of inbound network/system traffic, access to systems/servers/critical apps, event logs, admin configs, resource use, and anomalous behaviour including unauthorized access attempts, unusual user/system behaviour, and process injection — all of which surface RDP logons (especially anomalous ones) once they occur.
- T1021.002mostlydetects — A.8.16 explicitly monitors access to systems/servers/critical apps, event logs, admin-level config files, resource use, anomalous behaviour (including unauthorized access attempts and deviations from baseline), and known attack patterns, which directly surfaces SMB/Windows Admin Shares abuse by valid admin accounts.
- T1021.004mostlydetects — A.8.16's scope explicitly covers inbound network traffic, access to systems, event logs, and successful access attempts to protected resources — SSH authentications and sessions are logged at multiple layers (sshd auth logs, PAM, network flow, bastion/jump-host logs) and behavioral baselines (time, location, frequency, destination) can flag anomalous use of valid accounts; the remainder is that valid-credential SSH sessions that mimic the legitimate user's normal pattern blend with baseline activity and evade behavioral detection.
- T1021.005mostlydetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic (RFB protocol on ports 5900+ is observable), access to systems (VNC session establishment and authentication events are loggable), and behavioural baselines of normal access time/location/frequency — all of which surface VNC usage; the bounded remainder is that in-session activity relayed through VNC appears as legitimate user behaviour and the clause sets scope by business requirements rather than mandating depth, so an implementation monitoring only application-layer logs without network or behavioural baselines would miss it.
- T1021.006mostlydetects — A.8.16's monitoring scope explicitly covers inbound/outbound network traffic (WinRM uses 5985/5986), access to systems, event logs, and security-tool logs, and its baseline guidance calls out unusual time/location/frequency of access — all telemetry surfaces where WinRM-based lateral movement is observable; the use of valid accounts means authentication events look legitimate, so detection depends on behavioral anomaly detection rather than authentication failure, leaving a bounded remainder.
- T1036.001mostlydetects — A.8.16 explicitly lists monitoring for code tampering (f), process injection and anomalous behaviour (d,e,i), security-tool logs, baselines of normal execution, and real-time signature/pattern recognition, which surfaces invalid-code-signature files that deviate from expected signed binaries on Windows/macOS.
- T1037.002mostlydetects — A.8.16 explicitly lists monitoring of critical/admin-level configuration files (item c), event logs (item e), and checking that executed code is authorized and untampered (item f) — all of which surface a login hook: the plist modification, the script execution at login, and the resulting log entries. Scope is set by organizational requirements, so detection is not automatic, but a competent implementation covering macOS estate would catch this technique through multiple listed mechanisms.
- T1041mostlydetects — A.8.16 explicitly includes outbound traffic monitoring, DLP tools, and baseline-based anomaly detection in its scope, all of which surface T1041: DLP inspects encoded payloads leaving the network, baseline monitoring catches unusual outbound volumes, and known-malicious C2 infrastructure is named in the clause's own examples. The technique is designed to blend with legitimate C2 traffic, so detection depends on depth of DLP content inspection and behavioural baselining rather than being automatic, leaving a bounded remainder.
- T1046mostlydetects — Detection of unauthorized scanning of business applications, systems and networks limits the adversary's ability to discover network services without triggering alerts.
- T1047mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour including process injection, unusual system behaviour, resource use, admin-level activity, event logs, security-tool logs and known attack characteristics; WMI abuse (local or remote) produces observable artifacts across those exact vectors on Windows, so the control surfaces the technique in flight.
- T1048mostlydetects — A.8.16's scope explicitly includes outbound network traffic, security tool logs (IDS/IPS, firewalls, DLP), and deviations in standard protocol use — all of which surface exfiltration over alternative protocols; the bounded remainder is encrypted/obfuscated channels and protocol blending that mimic legitimate traffic, which monitoring can be tuned to miss.
- T1048.002mostlydetects — A.8.16 explicitly covers outbound network traffic, resource utilization, baseline anomaly detection, and DLP/security tool logs — all of which surface T1048.002 through metadata (volume, destination, timing, resource spikes) even though the asymmetric encryption hides content; the bounded remainder is distinguishing encrypted exfiltration flows from legitimate encrypted traffic on content alone.
- T1048.003mostlydetects — A.8.16 explicitly mandates monitoring of outbound network traffic, traffic patterns, and deviations from baseline behaviour, all of which surface exfiltration over unencrypted protocols (HTTP/FTP/DNS) through volume anomalies, unusual destinations, and timing patterns; the bounded remainder is low-and-slow exfiltration that mimics normal traffic and obfuscation embedded in protocol headers that defeats content inspection.
- T1053.005mostlydetects — A.8.16 explicitly lists monitoring of system/network activity, event logs, admin configuration files, resource use, process anomalies, and known attack patterns including process injection and deviations; scheduled task creation/abuse produces observable artifacts (new tasks, registry changes, schtasks.exe execution, anomalous scheduled activity) that fall inside the defined baseline-monitoring and alerting scope on Windows.
- T1055mostlyresponds — A.8.16 requires real-time/periodic monitoring for anomalous behaviour (including explicit examples of process injection and related indicators), dedicated trained personnel, alert generation, and timely procedures to respond to positive indicators per 5.26, which directly enacts the containment/eradication act that `responds` names once the technique is underway.
- T1055.005mostlyresponds — A.8.16's real-time/periodic monitoring of system behaviour, process activity, resource use, known attack patterns and explicit anomalies such as process injection surfaces the technique once it is underway; dedicated trained personnel and procedures then respond per 5.26, containing/eradicated the actor's foothold while the remainder (impact already realised before detection) is left for recovery controls.
- T1057mostlydetects — A.8.16 explicitly lists monitoring of process injection, anomalous system behaviour, resource use, event logs, security-tool output and deviations from a normal baseline, all of which surface Process Discovery (T1057) when it occurs on monitored hosts; the remainder is the bounded slice of platforms or processes outside the organisation's chosen monitoring scope.
- T1059mostlydetects — Monitoring for anomalous command execution, process injection and deviations from standard protocol use can flag unauthorized use of command-line interpreters.
- T1068mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, unauthorized access attempts, known attack characteristics, resource anomalies, and deviations from baseline, all of which surface T1068 exploitation events in real time or near-real time on the monitored estate; the named remainder is exploitation that produces no observable deviation within the chosen monitoring scope.
- T1070.004mostlydetects — A.8.16 explicitly lists monitoring of event logs, system/network activity, resource use, process termination, unauthorized access, and anomalous behaviour (including process injection and malware-associated activity); file deletion via built-in commands or tools is observable in those logs and baselines, with a bounded remainder for stealthy or non-logged deletions outside monitored scope.
- T1071.001mostlydetects — A.8.16 explicitly requires monitoring of outbound/inbound network traffic, security-tool logs (firewalls/IDS/IPS), anomalous behaviour patterns, known malicious domains, deviations in standard protocols, and real-time alerting against a baseline; this directly surfaces T1071.001 C2 traffic that blends with or deviates from web-protocol norms, with the bounded remainder being fully stealthy implementations inside allowed traffic that match the baseline exactly.
- T1071.002mostlydetects — A.8.16 explicitly requires monitoring of outbound/inbound network traffic, security-tool logs (firewalls/IDS/IPS), anomalous behaviour patterns, known malicious domains/IPs, deviations in standard protocols, and real-time alerting against a baseline; this surfaces T1071.002's C2 traffic when it deviates from or matches known-bad signatures, with the bounded remainder being fully blended traffic on unmonitored segments or protocols outside the chosen scope.
- T1071.003mostlydetects — A.8.16 explicitly requires monitoring of outbound/inbound network traffic, event logs, security-tool logs, resource use, and anomalous behaviour patterns (including known malicious domains, protocol deviations, and C2-like activity); T1071.003's mail-protocol C2 produces observable network artefacts that fall inside this scope and would be surfaced by baseline deviation or signature-based detection, with only a bounded remainder (e.g., perfectly mimicked low-volume traffic or monitoring scopes that deliberately exclude certain mail flows) left unreached.
- T1074mostlydetects — A.8.16 explicitly lists monitoring of network traffic, access, configuration files, security-tool logs, event logs, resource use, baselines of normal behaviour, and specific anomalies including process injection, malware-associated activity, unauthorized access/scanning, and deviations from expected user/system behaviour — all of which surface T1074 staging activity (file copies, archiving, new cloud instances, central directories) in real time or near-real time; the remainder is activity that evades the chosen scope or produces no observable deviation.
- T1074.001mostlydetects — A.8.16 explicitly lists monitoring of file/system activity, resource use, process anomalies, baselines of normal behaviour, and real-time/periodic detection of deviations including unusual system behaviour; local data staging produces observable artifacts (new files, archive activity, registry changes, anomalous I/O or process execution) that fall inside the listed scope, with only a bounded remainder (e.g. perfectly stealthy in-memory staging on unmonitored hosts) left unreached.
- T1078mostlydetects — Continuous monitoring of access attempts against a baseline of normal user behaviour and locations directly reduces the stealth value of using valid accounts for unauthorized activity.
- T1087.002mostlydetects — Domain account enumeration via net user /domain, Get-ADUser, ldapsearch, etc. generates event logs (4662/4661 on DCs), LDAP network traffic, and process execution records — all explicitly within A.8.16's monitoring scope (event logs, access to systems, network traffic, unusual user/system behaviour). The bounded remainder is implementations whose scope excludes domain controller logs or internal LDAP traffic, which the clause permits by setting scope from business requirements rather than mandating depth.
- T1098mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour including unauthorized access/attempts, account-related activity patterns, admin/config changes, process anomalies, and deviations from baseline (e.g. unusual user behaviour, access times/locations), which surfaces most forms of account manipulation once it occurs on the monitored estate; the named remainder is pre-compromise manipulation on external identity providers or SaaS outside the organization's own monitored systems and logs.
- T1110mostlydetects — Baseline monitoring of access frequency and failed login attempts helps identify brute-force or password-spraying attempts against accounts.
- T1110.001mostlydetects — A.8.16 explicitly configures monitoring and alerting against a baseline for anomalous behaviour that squarely includes unsuccessful access attempts to protected resources, unusual login patterns, activity from known malicious IPs, and known attack characteristics; this surfaces password guessing in flight on the organization's estate (especially where it generates observable logs or traffic on covered services), with the bounded remainder being fully external/pre-authentication attempts that never reach monitored internal resources.
- T1110.001mostlyprevents — A.8.16 mandates real-time/periodic monitoring of network traffic, access attempts, auth logs, anomalous behaviour baselines, and explicit indicators such as unsuccessful access to protected resources, unusual login patterns, and known attack characteristics; this directly stops password-guessing success on covered vectors (e.g. via alerts triggering 5.26 response before compromise), with the bounded remainder being exempted identities, legacy protocols, or out-of-scope monitoring choices allowed by the clause.
- T1110.001mostlyresponds — A.8.16 requires real-time/periodic monitoring of network traffic, access attempts, logs, anomalous behaviour (including unauthorized access attempts and known attack patterns), baseline deviations, and alert generation with dedicated response personnel; this directly surfaces and enables timely response to password-guessing activity once underway, with the named remainder being stealthy attempts that evade the configured scope or baseline.
- T1110.003mostlydetects — A.8.16 explicitly monitors inbound/outbound traffic on the exact management ports and protocols (SSH, RDP, HTTP, LDAP, Kerberos, SMB, etc.) that password spraying uses, plus access attempts, event logs, anomalous login patterns, unusual user behaviour, and known attack characteristics, surfacing the technique in flight; the remainder is throttled or low-and-slow spraying that deliberately stays under the tuned baseline thresholds.
- T1110.004mostlydetects — A.8.16 explicitly covers monitoring of access attempts (successful and unsuccessful) to protected resources, authentication-related event logs, and unusual user/system behaviour — all of which surface credential stuffing through failed login bursts, account lockouts, and anomalous authentication patterns. The bounded remainder is scope: monitoring depth is set by business requirements, so an implementation that omits authentication service logs or that faces distributed low-and-slow stuffing from many sources may see less, mirroring the T1110.001 anchor.
- T1123mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, malware-associated activity, resource use, event logs, security-tool logs and deviations from baseline, all of which surface audio-capture techniques (API calls, new processes, unusual mic/webcam access, file writes) in real time or near-real time; the named remainder is capture performed entirely through legitimate, in-scope applications (e.g. voice-call services) that produce no detectable deviation.
- T1127mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, code execution integrity checks, resource anomalies, and known attack patterns (including those matching trusted-developer-utility abuse) within its monitoring scope and baseline; this surfaces T1127 when it runs on monitored Windows systems, with the bounded remainder being execution outside the defined monitoring scope or before baseline tuning.
- T1133mostlydetects — A.8.16's monitoring scope (inbound/outbound network traffic, access logs, authentication events, baseline-relative anomalies in time/location/frequency) directly covers external remote service usage — VPN/RDP/Citrix connections, gateway authentication, and unusual access patterns are all observable. The bounded remainder is use of valid credentials at expected times/locations, which baseline monitoring may not flag without additional UEBA-style analysis.
- T1134mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, unusual user/system behaviour, resource use deviations, and security-tool/event logs as monitoring targets; token manipulation produces observable anomalies in process tokens, parent-child relationships, privilege escalations to SYSTEM, and related API calls that fall inside the defined baseline-comparison scope.
- T1136mostlydetects — A.8.16 explicitly monitors for anomalous behaviour including unauthorized access/attempts, account-related events in logs, unusual user/system behaviour, admin config changes, and process/resource anomalies, which surfaces most instances of T1136 account creation across its broad platform scope once the baseline is established and alerts are tuned.
- T1136.001mostlydetects — Local account creation generates well-known event log entries (e.g., Windows Security log 4720, Linux auth.log) and process execution records that fall within A.8.16's explicit scope (item e: event logs relating to system and network activity; item d: security tool logs), so a conformant monitoring implementation detects this; the remainder is the clause's scope-setting nature — an organization that chooses not to include host event log monitoring is still conformant and would miss it.
- T1190mostlydetects — A.8.16 explicitly enumerates network/application traffic monitoring, IDS/IPS and web filter logs, known attack characteristics (including buffer overflows), and unauthorized access attempts — all squarely in the path of a public-facing application exploit. The technique is observable through the clause's stated scope; the bounded remainder is zero-day or novel exploits that evade signature-based detection and implementations that omit the recommended components.
- T1197mostlydetects — A.8.16's monitoring scope explicitly covers the artifacts BITS jobs leave — network traffic (a), security tool logs (d), event logs including BITS-Client operational logs (e), code execution authorization (f), and resource use (g) — and the clause specifically names monitoring for unusual system behaviour and deviations in standard protocols, which BITS abuse produces; however, BITS is designed to be stealthy (self-contained job database, no new files or registry modifications, firewall-friendly), so detection depends on implementation depth rather than being guaranteed by the clause's scope.
- T1210mostlydetects — A.8.16 explicitly configures monitoring to surface anomalous behaviour including known attack characteristics (DoS, buffer overflows), process injection, unauthorized access/attempts, unusual system behaviour, and traffic patterns from malicious sources, which directly catches the observable indicators and execution of remote service exploitation (T1210) on the monitored estate; the bounded remainder is pre-compromise discovery or exploitation that produces no detectable anomaly within the organization's own systems, networks or logs.
- T1212mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, known attack characteristics, malware-associated activity, unauthorized access attempts, and deviations from baseline (including resource use and executed code integrity), all of which surface T1212 exploitation in flight on the covered platforms; the named remainder is exploitation occurring entirely outside monitored estate or before a baseline exists.
- T1218.003mostlydetects — A.8.16 explicitly lists process injection, anomalous process termination, unusual system behaviour, signed-binary abuse patterns, resource anomalies, and real-time baseline monitoring of system activity, logs, and security tools, all of which surface CMSTP.exe proxy execution of malicious INF/DLL/SCT payloads on Windows.
- T1218.004mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, unsigned/tampered code execution, resource anomalies, and security-tool logs (including IDS/IPS) within its real-time baseline monitoring and alerting scope, which surfaces InstallUtil proxy execution when it deviates from the established baseline on a monitored Windows endpoint.
- T1218.005mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, unauthorized code execution, resource anomalies, security-tool logs, and real-time alerting against baselines, which directly surfaces mshta.exe abuse as a trusted binary proxying malicious HTA/VBS/JS payloads.
- T1218.007mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, unsigned/tampered code execution, resource anomalies, and security-tool/ event logs, all of which surface msiexec.exe proxying malicious payloads or DLLs as deviations from baseline.
- T1218.009mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, code execution/tampering checks, resource use, security-tool logs, and real-time baseline monitoring for malware-like activity, all of which surface Regsvcs/Regasm proxy execution; the named remainder is non-monitored processes or implementations that omit host telemetry.
- T1218.011mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, unauthorized code execution, resource anomalies, security-tool logs, and real-time baseline deviation monitoring, all of which surface rundll32.exe proxying of malicious DLLs/scripts as observable anomalies or events.
- T1219mostlydetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic, process execution, user behavior baselines, deviations in standard protocol use, and traffic to known malicious infrastructure — all of which surface remote access tool abuse through anomalous connections, unexpected process launches, and unusual session patterns; the bounded remainder is the case where a legitimately approved remote access tool is used by a compromised account in a manner indistinguishable from normal administrative activity.
- T1219.001mostlydetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic, process behavior, deviations in standard protocols, and resource use — all of which IDE tunneling produces (SSH/proprietary tunnel traffic, IDE process activity, bandwidth consumption). The technique is observable on the estate, but it is designed to blend with legitimate developer workflows, so detection depends on baselines of normal user/system behavior and proper scope configuration rather than being automatic.
- T1222mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour including process injection, unauthorized access attempts, changes to critical configuration files, resource anomalies, and known attack patterns; T1222 modifications of file/directory permissions (especially on config, binary, or ACL targets) are observable in event logs, access logs, and baseline deviations on the covered platforms.
- T1485mostlydetects — A.8.16's monitoring scope explicitly covers file system event logs, resource utilization (CPU/disk I/O spikes during overwriting), process execution logs, and 'activity typically associated with malware' — all of which surface mass file deletion, overwriting, and worm-like propagation; the bounded remainder is destruction performed with legitimate credentials at low volume that resembles normal activity and evades the baseline.
- T1489mostlydetects — A.8.16 explicitly enumerates 'unplanned termination of processes or applications' as anomalous behaviour to identify, and service stops are highly observable through system event logs, service state changes, and cloud API logs (e.g., CloudTrail for DisableAPIServiceAccess) — all within the clause's enumerated scope. `mostly` and not `full` because scope is set by business and security requirements: an implementation monitoring only network traffic is conformant and would miss service terminations, so the catch is bounded by what the implementer chose to instrument.
- T1491mostlydetects — A.8.16 explicitly lists monitoring for unauthorized access to systems and unusual system behaviour, and defacement requires modifying content on a system, which generates observable events (file changes, web server logs, unauthorized access) within the control's defined scope of outbound/inbound traffic, access to systems, and event logs.
- T1491.001mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour, known attack characteristics, unusual system behaviour (including process injection and deviations), resource/performance anomalies, unauthorized access, and real-time/continuous monitoring with alerts; internal defacement produces visible, logged changes to websites, login messages, desktops or files that deviate from baseline and are observable in the enumerated scopes (config files, event logs, system behaviour, resource use).
- T1496mostlydetects — A.8.16 explicitly names resource utilization monitoring (CPU, hard disks, memory, bandwidth) and unusual system behaviour against a baseline as in-scope items, and resource hijacking produces exactly those anomalies (sustained CPU spikes, bandwidth consumption, deviation from baseline). The bounded remainder is scope set by organizational requirements: an implementation that monitors only access logs and not resource use is conformant with A.8.16 and would miss this technique.
- T1496.001mostlydetects — A.8.16 explicitly lists resource utilization (CPU, memory, disk, bandwidth), performance, and unusual system behaviour in its monitoring scope — all of which are the primary signatures of compute hijacking — and the technique also generates observable network traffic to mining pools or C2; the bounded remainder is that scope is set by business requirements, so an implementation omitting resource monitoring would still conform and miss the technique's core signal.
- T1496.002mostlydetects — A.8.16 explicitly covers outbound traffic, bandwidth use, resource utilization and unusual system behaviour, all of which surface bandwidth hijacking through anomalous outbound volumes, connection patterns to known malicious infrastructure, and resource consumption spikes; the remainder is stealth variants such as low-volume proxyjacking and traffic that mimics legitimate patterns, which can stay below alerting thresholds.
- T1498mostlyresponds — A.8.16 explicitly requires real-time/periodic monitoring for known attack characteristics (including denial of service), anomalous traffic/behaviour, and generating tuned alerts so dedicated personnel can respond via defined procedures (cross-referenced to 5.26 incident response) once the Network DoS is underway, bounding impact; mostly because scope is set by organisational requirements rather than mandating universal coverage of every possible vector.
- T1499mostlydetects — A.8.16 explicitly names 'denial of service' as a known attack characteristic to monitor for, and its scope covers the relevant signals for endpoint DoS — resource use (CPU, memory, bandwidth), event logs (crashes), network traffic, and baseline deviations — but some application-layer or exploitation-based endpoint DoS can resemble legitimate traffic or lack clear signatures, so detection is not universal across all sub-techniques.
- T1499.001mostlydetects — A.8.16 explicitly lists denial of service and network bottlenecks as monitored anomalies, and OS exhaustion floods produce the traffic surges and resource exhaustion the control is designed to surface, with the bounded remainder being floods that saturate the link before reaching the monitoring system's vantage point.
- T1499.002mostlydetects — A.8.16 explicitly lists denial of service and network bottlenecks/overloads as anomalous behaviours to monitor for, and service exhaustion floods produce observable traffic volume and resource consumption patterns that fall within the monitoring scope.
- T1499.003mostlydetects — A.8.16 explicitly includes application traffic monitoring, resource use (CPU, memory, bandwidth) and performance monitoring, and detection of bottlenecks/overloads and unusual system behaviour — all of which an application exhaustion flood produces as observable signals; the remainder is the clause's scope-setting nature, where an implementation focused on network-layer telemetry rather than application-layer resource consumption could miss the technique despite being conformant.
- T1505.003mostlydetects — A.8.16 explicitly lists monitoring for process injection, unauthorized access, anomalous system behaviour, known attack characteristics, security-tool logs, event logs, resource use, and code-tampering checks; web-shell deployment and use produce observable artifacts (e.g. anomalous web-server processes, file writes, traffic patterns, or execution anomalies) that fall inside the defined baseline-and-alert scope, with only a bounded remainder (e.g. fully stealthy or out-of-scope network-device shells) left unreached.
- T1518.001mostlydetects — A.8.16 explicitly requires monitoring of security-tool logs (item d), event logs (e), system/network activity (e/g), resource/performance baselines, and anomalous behaviours including process injection and malware-associated activity; this surfaces the specific discovery commands, registry queries, and agent-enumeration behaviours named in T1518.001 once they execute on the monitored estate.
- T1529mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour including unplanned termination of processes, unusual system behaviour (explicitly naming process injection), known attack characteristics, resource use deviations, and successful/unsuccessful access attempts; shutdown/reboot is a visible, logged system event that deviates from the established baseline of normal behaviour and is surfaced by the listed monitoring items (event logs, resource use, system activity) in real time or near-real time.
- T1529mostlyresponds — A.8.16 surfaces shutdown/reboot via anomalous behaviour (e.g. unplanned termination, unusual system behaviour, resource spikes) in real time and feeds it to the 5.26 response procedure for containment/eradication once underway; mostly because the clause sets scope by business requirements rather than mandating universal coverage of every shutdown vector.
- T1543mostlydetects — A.8.16 explicitly lists monitoring of system/network config files, event logs, resource use, process termination, malware-associated activity, process injection, unauthorized access, and deviations from baseline — all of which surface creation or modification of system processes (services/daemons/agents) for persistence.
- T1543.002mostlydetects — A.8.16 explicitly lists monitoring of critical/admin configuration files (c), event logs for system/network activity (e), security tool logs (d), resource/performance anomalies (g), and specific anomalous behaviours including process injection and deviations (in its baseline/anomaly examples), which directly surfaces systemd service creation, modification, or generator activity on Linux when it deviates from the established baseline.
- T1543.003mostlydetects — A.8.16 explicitly lists monitoring of system/network config files, event logs, security-tool logs, resource use, process anomalies, unauthorized access attempts, and specifically names process injection and malware-associated behaviour; these directly surface the creation/modification of Windows services (registry changes, sc.exe, service start events, anomalous drivers) in the monitored estate, with the bounded remainder being hidden/masqueraded services that evade standard baselines.
- T1543.004mostlydetects — A.8.16 explicitly lists monitoring of system/network configuration files, event logs, resource use, process anomalies, and specifically names process injection and unauthorized access attempts; these directly surface Launch Daemon creation/modification and anomalous startup execution on macOS, with the named remainder being pre-compromise adversary activity on external infrastructure that the control cannot observe.
- T1547mostlydetects — A.8.16 explicitly lists monitoring for process injection, unauthorized access, anomalous system behaviour, config file changes, resource use, malware-associated activity, and deviations from a normal baseline; these directly surface the configuration changes, kernel extensions, and autostart mechanisms (e.g. registry run keys, special directories, login items) used by T1547 after they occur.
- T1547.012mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, unauthorized access attempts, event logs, security-tool logs, resource use, and baseline deviations; these observables surface T1547.012's print-processor DLL load and spoolsv.exe restart under SYSTEM.
- T1554mostlydetects — A.8.16 explicitly lists monitoring for process injection, unauthorized code execution, tampered binaries (via integrity checks on executed code), anomalous system behaviour, and deviations from baseline — all of which surface T1554's binary modification or patching in flight or on execution.
- T1557mostlydetects — A.8.16 explicitly enumerates network traffic monitoring, IDS/IPS logs, deviations in standard protocol use, and traffic from known malicious IPs — all of which surface AiTM artifacts (ARP/DNS/LLMNR anomalies, TLS downgrades, unusual routing). The remainder is the scope-setting nature of the clause and sophisticated AiTM over valid encrypted channels that evade signature and behavioural detection.
- T1557.001mostlydetects — A.8.16 explicitly requires monitoring of inbound/outbound network traffic, access attempts, security-tool logs (IDS/IPS/firewalls), event logs, anomalous behaviour patterns (including known attack characteristics and unusual system behaviour), and real-time/periodic anomaly detection against a baseline, which directly surfaces LLMNR/NBT-NS/mDNS spoofing, NTLM hash transmission, and relay activity on Windows networks as observable events.
- T1558.003mostlydetects — A.8.16's scope explicitly includes event logs (4769 TGS requests), network traffic to the DC, and access monitoring, and its guidance names 'deviations in use of standard protocols' and 'known attack characteristics' as things to detect — Kerberoasting fits both. The bounded remainder is implementation: detecting the pattern (many TGS requests, RC4 etype 23) requires the baseline and alerting to be tuned to Kerberos behaviour, which the clause permits but does not mandate.
- T1558.004mostlydetects — A.8.16's scope covers network traffic to the DC, event logs on domain controllers, and security-tool telemetry (IDS/IPS), all of which surface the on-estate portion of AS-REP Roasting: LDAP enumeration of accounts without preauthentication, the burst of AS-REQ messages, and the AS-REP responses. The bounded remainder is the offline password cracking, which produces no estate-side artifact.
- T1559.001mostlydetects — A.8.16 explicitly lists process injection, anomalous system behaviour, unauthorized code execution, resource anomalies, and malware-associated activity in its monitoring scope and baseline, which directly surfaces COM abuse for local execution on Windows; the remainder is COM objects used only for non-observable side effects such as pure privilege-escalation without process or network artefacts.
- T1561.001mostlydetects — A.8.16 explicitly lists monitoring of disk/CPU/memory use, anomalous system behaviour (including process injection and deviations), event logs, security-tool logs, and baselines for anomalies such as unplanned terminations or malware-like activity; disk-wiping directly produces observable deviations in resource use, I/O patterns, and system behaviour on the monitored estate, with only the bounded remainder of completely offline or non-monitored devices left uncovered.
- T1561.002mostlydetects — A.8.16 explicitly lists monitoring for anomalous behaviour including known attack characteristics (DoS, buffer overflows), unusual system behaviour (process injection, deviations in protocols), resource use anomalies, unauthorized access/scanning, and baseline deviations; disk structure wipe produces observable effects in these categories (e.g. unplanned process termination, boot failures, resource spikes, or MBR/partition anomalies) on the monitored estate, with the bounded remainder being pre-compromise acquisition or purely offline wipes that leave no runtime footprint.
- T1563.002mostlydetects — RDP hijacking via tscon.exe generates process creation events, session connect/disconnect records, and deviations from the established baseline of normal user behaviour (e.g., a session becoming active at unusual times or from unexpected sources), all of which fall within A.8.16's monitoring scope; the technique is not fully detected because it uses native Windows tools and hijacks an already-authenticated session, so detection depends on having process-level and session-level telemetry rather than only network-layer monitoring.
- T1567mostlydetects — A.8.16 explicitly requires monitoring of outbound network traffic, anomalous behaviour patterns, activity from known malicious domains/IPs, deviations in protocols, and real-time alerting on baseline deviations; T1567 is outbound exfiltration that routinely produces observable anomalies in traffic volume, destinations, or patterns even when disguised over legitimate web services.
- T1567.002mostlydetects — A.8.16's monitoring scope explicitly includes outbound network traffic, DLP tools, web filters, and anomalous behaviour detection — all of which surface uploads to cloud storage services. The technique is observable on the organization's own estate, but detection depends on scope set by business requirements: an implementation without DLP or without outbound-traffic monitoring is fully conformant yet sees little, and legitimate use of the same cloud services creates a baseline-noise problem that bounds what alerting catches.
- T1567.003mostlydetects — A.8.16's scope explicitly includes outbound network traffic, DLP, web filters, and IDS/IPS — all directly relevant to observing uploads to text storage sites — and the technique produces traffic that traverses the organization's perimeter; the bounded remainder is that monitoring scope is set by business requirements rather than mandated, and encrypted/paid pastebin features can evade signature- and DLP-based detection.
- T1569mostlydetects — A.8.16 explicitly lists monitoring for process injection, anomalous system behaviour, resource use, unauthorized access, event logs, security tool logs and deviations from baseline, all of which surface T1569 service abuse either in real time or near-real time; the named remainder is service creations or executions that produce no observable deviation within the scoped items.
- T1569.002mostlydetects — A.8.16 explicitly lists monitoring of system/network activity, event logs, resource use, process anomalies, unauthorized access attempts, and specifically calls out process injection and malware-associated behaviour; service execution via the service control manager (services.exe, sc.exe, PsExec) produces observable artifacts (new/modified services, anomalous process starts, resource spikes, event log entries) that fall inside the defined baseline-and-anomaly scope, with only the bounded remainder of completely stealthy or non-logged service creations outside monitored coverage.
- T1570mostlydetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic, access to systems, security tool logs (IDS/IPS/firewalls), event logs, and baseline-based anomaly detection — all of which surface lateral file transfers via SMB, RDP, SSH/scp/rsync/curl/sftp, or web-service sync. The bounded remainder is transfers that use legitimate protocols and credentials in patterns that fit the established baseline, where the content is encrypted and the activity resembles normal administration.
- T1571mostlydetects — A.8.16 explicitly includes monitoring of inbound/outbound network traffic and deviations in use of standard protocols as examples of unusual behaviour, and non-standard port usage is a textbook network-traffic anomaly visible to firewalls, IDS/IPS and traffic baselines; the bounded remainder is the clause's scope-setting language, which lets an implementation monitor only a subset of traffic and still be conformant.
- T1595.001mostlydetects — A.8.16's guidance explicitly names 'unauthorized scanning of business applications, systems and networks' as anomalous behaviour the monitoring system should identify, and scanning IP blocks produces inbound traffic visible at the perimeter in firewall, IDS/IPS and netflow logs; the bounded remainder is scope set by organisational requirements, so low-and-slow or distributed scans may evade a conformant but minimally-tuned implementation.
- T1595.002mostlydetects — A.8.16 explicitly names 'unauthorized scanning of business applications, systems and networks' as anomalous behaviour to detect, and inbound network traffic is in its monitoring scope; the bounded remainder is stealthy or low-and-slow scans that evade threshold-based detection, plus scope set by business requirements rather than mandated depth.
- T1599mostlydetects — A.8.16 explicitly lists 'networking equipment' and 'outbound and inbound network... traffic' within monitoring scope, and T1599's observable manifestations — unauthorized access to boundary devices, configuration changes to those devices, and anomalous cross-boundary traffic — fall within those named items; the bounded remainder is implementation depth, since the clause sets scope by business requirements and network device telemetry is often less mature than endpoint monitoring.
- T1599.001mostlydetects — A.8.16 explicitly includes 'networking equipment' and 'critical or admin level system and network configuration files' in its monitoring scope, and NAT configuration changes on a boundary device plus the resulting cross-boundary traffic flows fall squarely within those items; the bounded remainder is that scope is set by organizational requirements and the technique is specifically designed to obscure monitoring data transmissions.
- T1602.001mostlydetects — A.8.16's network traffic monitoring (item a), access logging to networking equipment (item b), IDS/IPS signatures (item d), and detection of known attack characteristics explicitly cover SNMP enumeration, which generates observable traffic and device logs; the bounded remainder is SNMPv3 encryption limiting payload inspection and scope decisions that may exclude internal SNMP traffic from the monitored baseline.
- T1621mostlydetects — A.8.16's scope covers event logs, access to systems, and anomalous behaviour detection — authentication logs showing repeated MFA request generation and login patterns are within that scope, and the technique's 'MFA fatigue' variant produces a recognisable volume anomaly. The bounded remainder is scope set by business requirements: an implementation that excludes identity-provider logs or has thresholds tuned above the attack's request rate is fully conformant and would miss it.
- T1654mostlydetects — A.8.16 explicitly requires monitoring of event logs, security-tool logs, access attempts, system/network activity, resource use, and anomalous behaviour (including process injection and known attack patterns); this directly surfaces log-enumeration activity by host tools or against a SIEM in real time or near-real time on the monitored estate.
- T1659mostlydetects — A.8.16 explicitly requires monitoring of inbound/outbound network traffic, application behaviour patterns, anomalous activity (including known attack characteristics and deviations from baseline), and real-time/continuous tools that surface indicators such as malicious traffic or injected content signatures, directly detecting T1659's network-channel manipulation and content injection on the monitored estate.
- T1667mostlydetects — Email bombing manifests as a massive spike in inbound SMTP traffic and mail server resource consumption, both explicitly within A.8.16's monitoring scope (inbound network traffic, resource use, event logs), and the attack occurs on the organization's own estate; the remainder is that scope is set by requirements, so an implementation that does not include email traffic or mail server telemetry in its monitored set would not detect it.
- T1685mostlydetects — A.8.16 explicitly lists monitoring of security-tool logs, event logs, system/network activity, configuration files, resource use, code tampering, and known attack patterns (including process injection and malware behaviour); these directly surface the disabling/tampering actions and their telemetry-blocking effects once they occur on the monitored estate, with the bounded remainder being pre-compromise or fully external tool tampering that leaves no observable artifact inside the organisation's scope.
- T1685.005mostlydetects — A.8.16 explicitly includes event logs, system/network activity, admin configuration changes, and anomalous behaviour (including unauthorized access or tampering) in its monitoring scope and baseline; clearing Windows Event Logs is a detectable anomaly against that baseline, though the control's scope is set by the organization rather than mandating universal coverage of every possible log-clearing vector.
- T1685.005mostlyresponds — A.8.16's real-time/periodic monitoring of event logs, anomalous behaviour (incl. unplanned terminations, unauthorized access, and deviations), automated alerts, and procedures to respond to positive indicators (explicitly linking to 5.26 incident response) directly engages the clearing of Windows Event Logs as a detectable adverse event that triggers timely response to minimize impact.
- T1686mostlydetects — A.8.16 explicitly lists monitoring for unauthorized access attempts, unusual system behaviour including deviations in protocol use, and checking that executed code is authorized, all of which directly surface the firewall tampering and rule changes described in T1686; the remainder is the bounded set of modifications performed through channels outside the monitoring scope the organization sets.
- T1686.003mostlydetects — A.8.16's scope explicitly enumerates firewall logs, event logs, and configuration files — all of which surface firewall rule changes, registry modifications, and the netsh/PowerShell invocations the technique uses. A conformant monitoring implementation catches the technique through multiple channels, but the clause sets scope by business requirements rather than mandating every listed source, so an implementation that omits firewall or process telemetry is still conformant and would miss it.
- T1001partialdetects — A.8.16's scope covers outbound/inbound network traffic, protocol-deviation detection, and baseline comparison, which are the right vantage points for T1001, but the technique is specifically engineered to make C2 traffic resemble normal traffic (steganography, junk data, protocol impersonation), so monitoring catches the cruder obfuscations and clear protocol deviations while sophisticated, well-crafted obfuscation that stays within baseline behavior evades detection.
- T1001.001partialdetects — A.8.16 monitors network traffic against a baseline and explicitly lists 'deviations in use of standard protocols' as an anomaly to detect, so junk-padded C2 traffic that alters packet sizes, timing or statistical patterns is observable; however, the technique is purpose-built to defeat trivial traffic analysis, and the clause sets scope by business requirements rather than mandating deep protocol-aware inspection, so what is caught is a slice of the technique's indicators rather than the bulk.
- T1001.002partialdetects — A.8.16's monitoring scope includes network traffic and anomaly detection, which can surface steganographic C2 through traffic pattern analysis, but the technique is designed to evade signature-based inspection and the control's scope is set by business requirements rather than mandating deep packet inspection, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1001.003partialdetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic and lists 'deviations in use of standard protocols' as an anomaly to detect, which is the surface T1001.003 operates on; however, the technique is specifically engineered to make C2 traffic blend in with legitimate protocols and services (fake SSL/TLS handshakes, mimicked HTTP headers and certificates), and the clause sets scope rather than mandating the depth of analysis (deep packet inspection, certificate chain validation, behavioural baselining of legitimate service patterns) needed to reliably surface such impersonation — so what is caught is a slice determined by implementation depth, not a bounded remainder of a mechanism that mostly works.
- T1003.001partialresponds — A.8.16 surfaces LSASS dumping or anomalous process/memory behaviour (explicitly names process injection and unusual system behaviour) in real time or near-real time, hands it to the 5.26 response procedure for containment/eradication once underway; the named remainder is post-exfiltration credential use that has already left the monitored estate.
- T1003.003partialdetects — NTDS.dit access, volume shadow copy creation, and execution of ntdsutil/secretsdump/Invoke-NinjaCopy are observable through host-level event logs, file-access auditing and process telemetry, all within A.8.16's scope — but the clause sets scope by business requirements rather than mandating host/file instrumentation depth, so an implementation monitoring only network traffic is fully conformant and sees nothing of this technique.
- T1003.004partialdetects — A.8.16's scope includes event logs and unusual system behaviour, and LSA secret access (registry reads of HKLM\SECURITY\Policy\Secrets or memory reads of lsass.exe) is observable through host telemetry such as Sysmon, registry auditing and process-access events; but the clause sets scope by business requirements rather than mandating the specific instrumentation depth needed, so an implementation monitoring only network traffic is fully conformant and sees nothing of this in-host credential access.
- T1003.005partialdetects — A.8.16's monitoring scope includes system and network activity logs and unusual system behaviour, which can surface credential dumping tools or anomalous access patterns, but the control determines scope by business requirements rather than mandating the host-level instrumentation depth needed to observe in-memory credential extraction, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1003.007partialdetects — A.8.16's scope includes file access auditing, event logs, and behavioral baselining against which /proc/<PID>/mem reads and credential-pattern searches would be anomalous, but the clause determines scope by business requirements rather than mandating the specific instrumentation (file access auditing on pseudo-filesystems, command-line/process telemetry) needed to reliably surface this technique; an implementation focused on network and application layers is fully conformant and sees nothing of in-process memory reads, and even with full scope the signal is buried in the noise of legitimate /proc access without behavioral baselining.
- T1003.008partialdetects — A.8.16's scope includes access monitoring, event logs, and detection of unauthorized access to protected resources — all of which can surface reads of /etc/shadow (root-only) and anomalous /etc/passwd access via auditd, process execution logs, or EDR. However, the clause sets scope by business and security requirements rather than mandating host file-access instrumentation, so a conformant implementation focused on network and application layers would miss this entirely, leaving a nameable slice rather than a bounded remainder.
- T1005partialdetects — A.8.16's monitoring scope includes access to systems, event logs, and unusual system/user behaviour against a baseline, which can surface local file access patterns and process activity associated with T1005; however, the clause sets scope by business requirements rather than mandating file-access auditing or DLP, and an adversary reading files with valid context produces activity that closely resembles normal behaviour, so detection depends on the implementation's depth rather than the clause's mechanism.
- T1006partialdetects — A.8.16's monitoring scope includes event logs, process execution, and behavioral baselines; the utilities used (vssadmin, esentutl, PowerShell with NinjaCopy) leave process-creation and command-line artifacts visible to event-log and security-tool monitoring. However, the technique is explicitly designed to bypass file system monitoring, and direct volume reads/writes against raw logical volumes produce no file-access events for the monitoring scope to surface, so the core mechanism evades detection.
- T1007partialdetects — A.8.16's scope includes event logs relating to system activity, which can capture process execution and command-line auditing that would surface `sc query`, `tasklist /svc`, `systemctl` and similar discovery commands; however, the clause sets scope by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this local enumeration.
- T1008partialdetects — A.8.16's scope explicitly includes outbound/inbound network traffic monitoring and detection of 'deviations in use of standard protocols' and 'unusual system behaviour,' which are the surfaces where fallback channels manifest; however, the technique is specifically designed to blend with legitimate traffic (often using standard protocols like HTTPS/DNS) and activates only intermittently, so detection depends on the baseline depth and behavioral analysis the organization implements rather than being guaranteed by the clause's scope.
- T1010partialdetects — A.8.16's scope is set by business and security requirements rather than mandating host-level process or API instrumentation, so a conformant implementation focused on network traffic, access logs and security-tool output sees nothing of a local window-enumeration call; an implementation that includes command-line auditing, process telemetry or API-call monitoring can surface it, but that is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1011partialdetects — A.8.16's monitoring scope is set by business requirements rather than mandating coverage of alternative network mediums; an implementation monitoring only the primary enterprise network is fully conformant and sees nothing of exfiltration over Bluetooth, cellular, or RF, while one that includes those mediums or watches resource-use anomalies can surface the activity.
- T1011.001partialdetects — Bluetooth exfiltration occurs on the organization's endpoints, so there is a vantage point within A.8.16's scope (system/application traffic, event logs, resource use, security tool logs), but the clause sets scope by business requirements rather than mandating Bluetooth-specific monitoring; an implementation focused on enterprise gateway traffic is fully conformant and sees nothing of a Bluetooth channel.
- T1012partialdetects — A.8.16's scope includes event logs and security-tool logs, which can capture registry access (e.g. via Sysmon, EDR, or Windows auditing), but the clause sets scope by business requirements rather than mandating host-level registry instrumentation; an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-host registry query.
- T1016partialdetects — T1016 runs on the estate via native utilities (ipconfig, arp, route, etc.), so process execution and command-line events fall within A.8.16's event-log and security-tool-log scope, but the clause sets monitoring scope by business requirements rather than mandating command-line auditing or behavioural baselining depth, and the technique blends with legitimate admin activity, leaving a nameable slice rather than a bounded remainder.
- T1016.001partialdetects — A.8.16's monitoring scope includes outbound network traffic and unusual system behaviour, which can surface the network requests used for connectivity checks, but the control determines scope by business requirements rather than mandating the depth of instrumentation needed to catch every method, so detection is a slice the implementer chooses.
- T1016.002partialdetects — A.8.16's scope includes event logs, system activity, and access to systems — host-based telemetry that can surface the netsh/wlanAPI calls or file reads this technique performs — but the clause sets monitoring scope by business requirements rather than mandating the specific instrumentation (command-line auditing, file access monitoring on Wi-Fi config paths) needed to catch it, so a fully conformant network-focused implementation sees nothing.
- T1020.001partialdetects — A.8.16 explicitly includes monitoring access to networking equipment (b), critical network configuration files (c), and event logs (e), which can surface unauthorized traffic-mirroring configuration changes on network devices or cloud environments; however, the technique is designed to be transparent, cloud-based mirroring (AWS/GCP/Azure) may sit outside on-premises monitoring scope, and the clause sets scope by organizational requirement rather than mandating depth of network-device configuration monitoring, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1021.003partialdetects — A.8.16's scope explicitly covers the artifacts DCOM produces — RPC network traffic, remote system access, event logs, and deviations in standard protocol use (clause d lists 'deviations in use of standard protocols' as a detection target) — but the clause sets scope rather than mandating instrumentation depth, so an implementation monitoring only network-layer traffic is fully conformant and may see the RPC calls without identifying them as DCOM-based lateral movement, while one with host telemetry and tuned Office-process baselines catches the technique; what is detected is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1021.007partialdetects — A.8.16's monitoring scope (access logs, event logs, behavioral baselines) can surface cloud service authentication and API activity, and anomalous usage patterns would be caught by the baseline-comparison mechanism the clause mandates; however, the technique uses valid/federated accounts so authentication events look legitimate, and A.8.16 sets scope by business requirements rather than mandating cloud-service log ingestion, so an implementation monitoring only on-premises traffic is fully conformant and sees nothing of the cloud-side activity.
- T1021.008partialdetects — A.8.16's monitoring scope includes access to systems and anomalous behaviour, which can surface direct cloud VM connections via unusual access patterns, but the control's scope is determined by business requirements rather than mandating cloud API telemetry, so detection depends on implementation choices.
- T1025partialdetects — A.8.16's scope includes event logs, system activity, and access monitoring, which can surface removable media file access and the process execution that accompanies it; however, the clause sets scope by business and security requirements rather than mandating file-access or USB-event instrumentation, so an implementation monitoring only network and application layers is fully conformant and would miss the on-host collection.
- T1027partialdetects — A.8.16's monitoring scope (security-tool logs, behavioural baselines, code-execution verification per item f, and known-attack-characteristic matching) can surface some obfuscation through signatures, behavioural anomalies and post-deobfuscation activity, but T1027 is specifically engineered to evade these layers, and the clause sets scope by business requirements rather than mandating depth, so what is caught is a slice rather than the bulk.
- T1027.001partialdetects — A.8.16's monitoring scope (AV/IDS logs, file integrity checks, execution authorization, behavioral baselines) can surface a padded binary when the underlying malware is caught by heuristic or behavioral detection, or when execution monitoring flags unauthorized code; however, binary padding is specifically designed to defeat the hash-based and static-signature detection that A.8.16 implementations commonly rely on, and a fully conformant monitoring system focused on file hashes and signatures will miss the technique entirely.
- T1027.002partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered, and its anomalous-behaviour list explicitly names process injection, but packed executables decompress in memory and evade signature-based detection, so the control surfaces the technique only where host instrumentation depth reaches memory-level execution rather than file-level signatures.
- T1027.003partialdetects — A.8.16's monitoring scope can surface surrounding indicators (the malware executing steganography, unusual file transfers, resource anomalies from embedding operations, deviations in protocol use), but the core technique — hiding data within legitimate-looking media so the carrier file appears normal — is specifically designed to evade the traffic, file and behavioural monitoring the clause mandates, and the clause does not require steganalysis tooling.
- T1027.004partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered, which can surface the compilation step or the resulting binary, but the control determines scope by business requirements rather than mandating host instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a local compile.
- T1027.005partialdetects — A.8.16's monitoring scope includes behavioral anomaly detection and code-integrity checks (item f: detecting tampered/recompiled code), which can surface a modified tool's execution or integrity violations, but the indicator-removal act itself occurs on adversary infrastructure and the technique is explicitly designed to evade signature-based detection, leaving a bounded remainder where the modified tool runs cleanly within the monitored baseline.
- T1027.006partialdetects — A.8.16's scope includes endpoint behavioural monitoring (unusual system behaviour, deviations in standard protocols, event logs) which can surface the JavaScript Blob construction, file drop and payload execution on the victim — but HTML Smuggling is specifically engineered to evade the network-level content filters and web filters that are also part of A.8.16's scope, and the clause's scope is set by business requirements rather than mandating endpoint instrumentation depth, so an implementation focused on network traffic and gateway controls is fully conformant and sees nothing of the smuggling itself.
- T1027.007partialdetects — Dynamic API resolution produces runtime behaviors (LoadLibrary/GetProcAddress calls, hash-based lookups, unusual API call patterns) that process-level telemetry can surface, and A.8.16's scope includes 'unusual system behaviour' and code-execution authorization checks; however, the clause sets scope by business requirements rather than mandating process-level instrumentation depth, so a conformant network- or application-layer implementation sees nothing of an in-process resolution pattern — the same shape as the T1055 anchor.
- T1027.008partialdetects — A.8.16's scope includes security tool logs (AV, IDS, IPS) and behavioral monitoring that can surface stripped payloads through file analysis or post-execution behavior, but the clause sets scope by business requirements rather than mandating the static-analysis instrumentation that reliably detects stripping itself; an implementation focused on network and access logs is fully conformant and sees nothing of a stripped binary's symbol/string removal.
- T1027.010partialdetects — A.8.16's monitoring scope includes event logs, system activity, and deviations from standard protocol use, which can surface some obfuscation (base64/URL-encoded payloads, known Invoke-Obfuscation signatures, unusual character patterns), but T1027.010 is specifically engineered to defeat signature and behavioural analysis, and A.8.16 sets scope by business requirements rather than mandating the depth needed to catch sophisticated obfuscation.
- T1027.011partialdetects — A.8.16's scope includes event logs, system activity, and anomalous behaviour monitoring — surfaces where fileless storage in registry, WMI, event logs, or shared memory can leave observable traces — but the technique is explicitly designed to evade file-based and standard endpoint detection, and A.8.16 sets scope by business requirement rather than mandating instrumentation depth, so a conformant network-and-application-layer implementation sees nothing of it.
- T1027.012partialdetects — A.8.16's monitoring scope includes network traffic and anomalous behaviour, which can surface the outbound download triggered by the LNK, but the technique's obfuscation is designed to evade content filters and the control's scope is set by business requirements, not mandated instrumentation depth, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1027.013partialdetects — A.8.16's monitoring scope includes security tool logs (AV/IDS/IPS), behavioural baselines, and code-execution authorisation checks that can surface the deobfuscation step and runtime deviations, but T1027.013 is explicitly engineered to defeat static signature detection, and the encrypted/encoded file artifact itself is designed to look benign to the very tools A.8.16 enumerates.
- T1027.014partialdetects — A.8.16 includes both signature-based tools (AV, IDS, IPS — explicitly listed in item d) and behavioural/anomaly-based monitoring against baselines (items f, g, and the 'unusual system behaviour' examples). Polymorphic code is specifically designed to defeat the signature-based half by mutating its footprint each execution, but the behavioural half — deviations in protocol use, unusual system behaviour, code-authorization checks — can still surface some instances. The technique's explicit purpose is to evade the very monitoring A.8.16 mandates, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1027.015partialdetects — A.8.16's monitoring scope includes security tool logs (AV, IDS, IPS, web filters, DLP) that scan compressed archives, network traffic monitoring that surfaces file transfers, and event logs that record compression activity, so the mechanisms to detect compressed obfuscated payloads are inside the clause; however, the technique is specifically designed to evade detection, compression itself is not anomalous behaviour, and the clause sets scope by business requirements rather than mandating content-inspection depth, so an implementation monitoring only network flows without DLP/email inspection is fully conformant and misses the payload inside the archive.
- T1027.017partialdetects — A.8.16's monitoring scope (network traffic, security tool logs, endpoint behaviour) can surface downstream effects of SVG smuggling — payload downloads, C2 connections, script execution from image files, and anomalous outbound traffic — but the technique is specifically engineered to evade content filters, and the clause sets scope by business requirements rather than mandating deep content inspection of file types, so catching the smuggling itself depends on the implementation's chosen instrumentation depth.
- T1027.018partialdetects — A.8.16's monitoring scope includes execution authorization checks, known attack characteristics, and unusual system behaviour, which can surface the runtime reconstruction of hidden payloads (e.g. unexpected eval/Proxy trap activity), but the core of T1027.018 — invisible Unicode characters embedded in files and scripts — is a content-level obfuscation that requires static or deep content inspection, which the clause does not mandate; scope is set by organizational requirements, so an implementation monitoring only network traffic, access logs and resource use is fully conformant and sees nothing of the hidden characters themselves.
- T1029partialdetects — A.8.16 monitors outbound traffic and can surface scheduled exfiltration when the transfer pattern deviates from the established baseline, but the technique's deliberate blending with normal activity and timing means detection depends on the monitoring scope and baseline quality the organization chooses, not on a mandated mechanism.
- T1030partialdetects — A.8.16's scope includes outbound traffic monitoring, DLP tools, baseline-based anomaly detection, and pattern recognition — all of which can surface chunked exfiltration that evades simple volume thresholds — but the technique is specifically designed to avoid triggering threshold alerts, and the clause sets monitoring scope rather than mandating detection depth, so what is caught depends on implementation choices (DLP content inspection, statistical baselining) rather than being guaranteed by the mechanism.
- T1033partialdetects — T1033 runs commands like `whoami`, `w`, and `who` on the organization's own systems, which fall within A.8.16's monitoring scope (event logs, process execution, baseline anomaly detection), but the commands are legitimate utilities commonly used by users, so detection depends on the organization's specific configuration of process auditing, command-line logging, and baseline thresholds rather than being guaranteed by the control itself.
- T1036partialdetects — A.8.16 includes mechanisms that surface masquerading — item (f) explicitly requires checking that executed code is authorized and untampered, baselines of normal behaviour cover expected file names/locations, and 'unusual system behaviour' detection catches masqueraded binaries acting anomalously — but the clause sets scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of file/metadata masquerading, leaving a nameable slice undetected.
- T1036.002partialdetects — A.8.16's monitoring scope can include file system events and 'unusual system behaviour', and RTLO characters in file names are detectable if the implementation inspects file names for Unicode anomalies — but the clause sets scope by business requirements rather than mandating this specific check, so a fully conformant monitoring programme that watches network traffic, access logs and resource use sees nothing of a renamed file sitting on disk waiting for a user to click it.
- T1036.003partialdetects — A.8.16's scope includes event logs, configuration file monitoring, and behavioural baselines that can surface renamed utilities — particularly the alternative case where a utility is copied to a non-standard path, which leaves a file-system artifact and a deviation from baseline. But the primary case (renaming in place) is specifically designed to evade these mechanisms: the renamed binary behaves identically to the original, and A.8.16 does not mandate the hash-comparison or path-whitelisting controls that would catch it reliably. An implementation monitoring only network and application layers is fully conformant and sees nothing of this technique, mirroring the T1055 anchor's reasoning.
- T1036.004partialdetects — A.8.16's baseline-comparison mechanism and event-log monitoring can surface new or modified tasks/services that deviate from the established norm, but the technique is specifically designed to evade this by mimicking legitimate names, and the clause sets scope by business requirements rather than mandating task/service inventory baselines — so detection is achievable but not guaranteed.
- T1036.005partialdetects — A.8.16's monitoring scope (file integrity, behavioural baselines, code-authorization checks, deviation from standard protocols) can surface some masquerading — e.g. duplicate svchost.exe, unsigned binaries in trusted paths, or a 'legitimate' resource behaving outside baseline — but T1036.005 is engineered to look normal, so a conformant monitoring implementation catches a slice rather than the bulk.
- T1036.006partialdetects — A.8.16's monitoring scope (process execution, file system events, security tool logs) can surface the unexpected binary launch and potentially the trailing-space filename anomaly, but the clause sets scope by business requirements rather than mandating file-integrity or filename inspection, so a conformant implementation focused on network and access monitoring would miss the masquerading itself.
- T1036.007partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and not tampered with, and its anomaly list explicitly covers unusual system behaviour, but double-extension files that reach a user's inbox and are opened rely on endpoint instrumentation depth that the clause makes subject to business requirements rather than mandating, so detection is a slice the implementer chooses.
- T1036.008partialdetects — A.8.16's monitoring scope explicitly includes security tool logs (antivirus, IDS, IPS, DLP) and checks for code execution authorization, which can surface file-type mismatches and known masquerade signatures; however, the technique is specifically engineered to bypass file validation checks, and monitoring implementations vary in depth, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1036.009partialdetects — A.8.16's scope includes event logs, unusual system behaviour, and code-authorization checks that can surface process-tree anomalies (e.g. unexpected fork patterns, processes adopted by init without a legitimate parent), but the technique is specifically engineered to make the resulting process tree look normal and on Unix this behaviour is common for legitimate daemons, so detection depends on the implementation's instrumentation depth rather than being guaranteed by the clause.
- T1036.010partialdetects — A.8.16's scope includes event logs and access monitoring, which would capture account creation events, and the clause explicitly calls for monitoring 'unusual user and system behaviour in relation to expected behaviour' against a baseline; however, detecting that a new account name is a masquerade (rather than a legitimate account) requires specific baseline comparison and similarity analysis that the clause does not mandate, and the technique is designed to blend in with expected names, so what is caught is a slice dependent on implementation depth rather than a bounded remainder of a mechanism that mostly works.
- T1036.011partialdetects — A.8.16's scope explicitly includes monitoring for 'unusual system behaviour' and 'deviations in use of standard protocols', and process argument spoofing is observable in principle through host telemetry (e.g., comparing a process's claimed identity in /proc/<PID>/cmdline against its actual behaviour, or memory-integrity checks); however, the clause sets scope by business and security requirements rather than mandating the host-level instrumentation depth needed, so a baseline-conformant implementation monitoring only network and application layers sees nothing of an in-memory argv overwrite.
- T1036.012partialdetects — A.8.16's scope includes outbound/inbound application traffic and the clause calls for baselines of normal behaviour and detection of anomalies, so a monitoring implementation can surface some indicators of browser fingerprint spoofing (User-Agent inconsistencies with actual client behaviour, known malicious UA patterns, deviations from standard protocol use). However, the technique is specifically designed to make traffic look legitimate, sophisticated spoofing that perfectly mimics a real browser evades this kind of analysis, and the clause sets scope rather than mandating the deep header/attribute correlation needed to catch all instances.
- T1037partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but boot or logon initialization scripts are not named, and the clause determines scope rather than mandating instrumentation depth, so detection depends on whether the implementation includes host-level telemetry that surfaces script execution at startup.
- T1037.001partialdetects — A.8.16's monitoring scope includes registry changes and logon events, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a registry-based logon script.
- T1037.003partialdetects — A.8.16's scope includes process execution monitoring, event logs, access events, and configuration file monitoring — all of which can surface a network logon script being added to a GPO, modified, or executing at logon. But the clause sets scope by business and security requirements rather than mandating instrumentation depth: an implementation monitoring only network traffic and access logs is fully conformant and would miss the script's creation, modification, or in-process execution. What is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1037.004partialdetects — A.8.16's scope explicitly includes monitoring critical/admin configuration files (item c) and checking that executed code has not been tampered with (item f), both of which can surface RC script modifications via file integrity monitoring; however, the clause sets scope per business and security requirements rather than mandating FIM, so an implementation monitoring only network and access layers is fully conformant and sees nothing of an on-disk file change.
- T1037.005partialdetects — A.8.16's scope includes configuration files, event logs, and unusual system behaviour, so monitoring file-system changes to /Library/StartupItems or boot-time process execution as root would surface this technique; but the clause sets scope by business and security requirements rather than mandating file-integrity or boot-process instrumentation, so an implementation focused only on network/application layers is fully conformant and sees nothing of a persistence write to a system directory.
- T1039partialdetects — A.8.16's monitoring scope includes access to systems and event logs, which can surface enumeration of network shares, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1040partialdetects — A.8.16's monitoring scope includes network traffic, system behaviour, and configuration files, which can surface some indicators of network sniffing (promiscuous mode on hosts, span port or traffic-mirroring configuration changes, unusual packet-capture tools), but passive sniffing is designed to be invisible — the traffic itself is unmodified and the capture leaves no signature in the data stream — so the core activity is not observable and the clause's scope is set by organisational requirements rather than mandated to include all detection points.
- T1046partialresponds — A.8.16's real-time/periodic monitoring, anomaly detection (including unauthorized scanning and known attack patterns), alerting, and procedures to respond to positive indicators directly engage T1046 once the scan is underway on the monitored estate, enabling timely containment; partial because scope is set by business requirements (not all scanning surfaces as observable anomalies, especially in unmonitored cloud/on-prem segments or non-baseline Bonjour/mDNS activity).
- T1047partialresponds — A.8.16 surfaces anomalous WMI usage (e.g. via process injection signatures, unusual system behaviour, admin-level access, or resource anomalies) once underway and feeds it to 5.26 response procedures, but the control itself performs none of the core respond actions (contain/eradicate) and many T1047 executions (especially remote or non-signatured) fall outside its scoped monitoring.
- T1048partialresponds — A.8.16's real-time/periodic monitoring of network traffic, anomalous behaviour, known attack patterns and alerts (with dedicated response personnel and procedures feeding 5.26) surfaces and enables response to exfiltration events once underway, but only for the subset observable in monitored traffic or baselines (e.g. unusual protocols or volumes) while many alternate-channel, encrypted, or console/API exfils fall outside its scope.
- T1048.001partialdetects — A.8.16's monitoring scope includes outbound traffic and anomalous behaviour, which can surface encrypted exfiltration via volume, destination, or protocol anomalies, but the control's scope is set by business requirements rather than mandating the deep packet inspection or decryption needed to reliably detect exfiltration within an encrypted tunnel, leaving a significant remainder.
- T1048.001partialresponds — A.8.16 surfaces anomalous exfiltration (e.g. unusual traffic patterns, unexpected protocols, or deviations from baseline) in real time or near-real time and feeds it to the 5.26 response procedure for containment/eradication; the technique has already run and produced outbound data, so this is genuine response once underway, but only a slice (network-layer observables inside the chosen monitoring scope) while many symmetric non-C2 exfil shapes (e.g. covert channels inside permitted HTTPS) remain unseen.
- T1048.003partialresponds — A.8.16's real-time/periodic monitoring, anomaly detection (including unusual outbound traffic, known attack patterns, and deviations from baseline), alerting, and procedures to respond to positive indicators directly engage an exfiltration event once underway for containment and eradication, but only for the observable slice (network traffic, logs, resource use) while leaving non-monitored protocols, obfuscated embedding, or non-network exfil untouched.
- T1049partialdetects — A.8.16's scope is set by business and security requirements rather than mandating host-level process execution telemetry; an implementation monitoring only network and application layers is fully conformant and sees nothing of a local netstat/lsof invocation, while one with process creation logging and behavioural baselining can surface it — the same shape the T1055 anchor teaches.
- T1052partialdetects — A.8.16's enumerated scope includes event logs (e), DLP tools (d), access to systems (b), and resource use (g) — all of which can surface USB device connections, file-copy operations, and removable-media transfers on the organization's own estate. However, the clause sets scope by business and security requirements rather than mandating USB-specific depth, so a network-focused implementation is fully conformant and sees nothing of physical exfiltration; what is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1052.001partialdetects — A.8.16's scope includes system event logs, resource use, and access to systems — all of which can surface USB device connections and file copies to removable media — but the clause sets scope by business and security requirements rather than mandating endpoint instrumentation, so a fully conformant network-focused implementation sees nothing of a physical-media exfiltration.
- T1053partialdetects — A.8.16's scope explicitly includes event logs (e) and system configuration files (c), which are the primary artifacts T1053 produces (Windows event 4698/4702, scheduled task definitions), and the anomalous-behavior list covers unusual user/system behavior (i) and malware-associated activity (b). However, the clause uses 'should be considered for inclusion' rather than mandating these specific items, so an implementation monitoring only network and access layers is fully conformant and would miss scheduled task creation — the same scope-not-depth argument that placed A.8.16 vs T1055 at `partial`.
- T1053.002partialdetects — A.8.16's scope includes event logs, security-tool logs, and unusual system/user behaviour, and the `at` utility leaves observable artifacts (process creation, command-line, scheduled-task records, at-spool files on Linux/macOS) that fall inside that scope where host telemetry is instrumented; the grade is partial rather than mostly because the clause sets monitoring scope by business and security requirements rather than mandating the process-execution and command-line depth needed to catch every instance, so a fully conformant implementation that monitors only network and application layers sees nothing of an on-host scheduling action.
- T1053.003partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but its scope is set by business requirements rather than mandating host-level instrumentation, so cron-based persistence is observable only where the implementer has chosen to monitor crontab changes or scheduled process execution.
- T1053.006partialdetects — A.8.16's scope explicitly covers admin-level configuration files (c), event logs (e), and unusual system behaviour (i) — all of which can surface a new or modified systemd timer unit, its journal entries, and the service it triggers — but the clause sets scope by business requirement rather than mandating host-based/file-integrity depth, so an implementation monitoring only network layers is fully conformant and sees nothing of this Linux persistence mechanism.
- T1053.007partialdetects — A.8.16's monitoring scope (API access logs, event logs, security tool logs, resource use, code-authorization checks) can surface a malicious Kubernetes CronJob through audit-logged API calls, unexpected workload creation, anomalous container resource use, and runtime behaviour — but the clause sets scope by business and security requirements rather than mandating container-orchestration instrumentation, so an implementation monitoring only traditional infrastructure is fully conformant and sees nothing of this technique, mirroring the A.8.16 vs T1055 partial pattern.
- T1055partialdetects — Monitoring for process injection and other unusual system behaviour provides visibility into stealthy code execution techniques.
- T1055partialprevents — A.8.16 explicitly lists process injection as an anomalous behaviour to baseline and alert on (and requires monitoring of code execution integrity), which can stop some injections from completing or succeeding when the monitoring system is in scope, but the clause only sets requirements for scope and does not mandate the instrumentation depth that would catch all platform-specific or in-process variants.
- T1055.001partialdetects — A.8.16 explicitly names process injection as an example of unusual system behaviour to monitor, and DLL injection is observable where host telemetry (process, memory, API call monitoring) is in scope; `partial` because the clause sets scope by business requirements rather than mandating instrumentation depth, so a conformant implementation monitoring only network/application layers sees nothing of an in-process injection.
- T1055.002partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor for, and PE injection leaves host-level artifacts (VirtualAllocEx/WriteProcessMemory/CreateRemoteThread calls, anomalous thread creation in target processes) that monitoring can surface; however, the clause sets scope by business and security requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process injection — what is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1055.003partialdetects — A.8.16 explicitly names 'process injection' as an anomalous behaviour to monitor for and includes code-authorization checks and system event logs in scope, but the clause sets scope rather than mandating instrumentation depth — an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-memory thread hijack, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1055.004partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor, and APC injection is observable where host-level telemetry (API calls like OpenThread/QueueUserAPC, thread queue state, process behaviour) is in scope; the grade is partial and not mostly because the clause sets monitoring scope by business and security requirements rather than mandating the host-instrumentation depth needed to see an in-process injection, so a conformant implementation monitoring only network and application layers sees nothing of it.
- T1055.005partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor for, and security tools in scope (AV, IDS, IPS) can surface TLS callback manipulation through behavioural analysis; however, the clause sets monitoring SCOPE rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process injection, and TLS callbacks are specifically designed to evade process-based defences.
- T1055.008partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor for, and ptrace attachments are observable via host-level telemetry (auditd, eBPF, syscall tracing), but the clause sets scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process ptrace injection.
- T1055.009partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor for, and proc memory injection leaves observable artefacts (writes to /proc/[pid]/mem via dd, syscall traces, parent-child anomalies, resource spikes) that host telemetry inside the monitoring scope can surface; but the clause sets scope by business and security requirements rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process injection — the same scope-driven remainder that places the parent T1055 at `partial`.
- T1055.009partialresponds — A.8.16 surfaces anomalous behaviour including process injection (explicitly listed) via real-time/periodic monitoring of system activity, memory use, and logs, feeding into 5.26 response procedures for containment/eradication once underway; partial because scope is set by organisational requirements rather than mandating universal host-level visibility of /proc-based injection on all Linux systems.
- T1055.011partialdetects — A.8.16 explicitly names process injection as an example of unusual system behaviour to monitor for, and EWM injection is a sub-technique of process injection observable through host telemetry (process memory operations, window class registration, EWM allocation). However, the clause sets scope rather than mandating instrumentation depth: an implementation monitoring only network and application layers is fully conformant with A.8.16 and sees nothing of an in-process injection, so what is caught is a slice the implementer chose.
- T1055.012partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor for, and process hollowing is an in-process injection technique observable via host telemetry (process creation in suspended state, memory unmapping, cross-process writes); however, the clause sets scope by business and security requirements rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process hollowing — what is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1055.013partialdetects — A.8.16's monitoring scope explicitly names process injection as an example of unusual system behaviour to detect, and items (e) event logs, (f) code execution authorization, and (g) resource use can surface TxF transactions, process creation from tainted memory sections, and rollback artifacts; however, Process Doppelgänging is specifically engineered to evade the highly-monitored APIs (NtUnmapViewOfSection, VirtualProtectEx, SetThreadContext) that standard process-injection detection relies on, and A.8.16 sets scope by organizational requirement rather than mandating the host-level telemetry depth needed to catch this evasive variant.
- T1055.014partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour the monitoring system should surface, and VDSO hijacking is a sub-technique of T1055; however, the clause sets scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process vdso hijack, and the technique is specifically designed to mask execution under a legitimate process.
- T1055.015partialdetects — A.8.16 explicitly names process injection as a behaviour to monitor, but ListPlanting is specifically engineered to evade the standard detection surface — it uses PostMessage/SendMessage window messages and 2-byte payload copies to avoid the highly monitored WriteProcessMemory path, and the clause sets monitoring scope by business requirement rather than mandating the deep process-level telemetry that would catch it; a conformant A.8.16 implementation focused on network and application layers sees nothing of this in-process injection variant.
- T1056partialdetects — A.8.16 explicitly names keystroke logging as an example of unusual system behaviour to monitor for, and T1056 spans both host-based sub-techniques (keylogging, GUI input capture, API hooking) and a network-based sub-technique (Web Portal Capture) that would surface in traffic monitoring; the bounded remainder is an implementation that monitors only one layer and misses the other, but the explicit naming of keystroke logging and the technique's network footprint push this above the `partial` given to T1055.
- T1056.001partialdetects — A.8.16 explicitly lists keystroke logging as anomalous behaviour to monitor for, but its scope is set by business and security requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a kernel-level keylogger.
- T1056.001partialresponds — A.8.16 surfaces keystroke logging as anomalous behaviour (explicitly listed) and hands it to the 5.26 response procedure for containment/eradication once underway; partial because the clause sets monitoring scope by business requirements rather than mandating universal coverage of all keylogging variants (e.g. raw hardware buffer reads or network-device hooks may fall outside chosen telemetry).
- T1056.002partialdetects — A.8.16's scope includes process behaviour, script execution (AppleScript/PowerShell/shell), and deviations in standard protocol use, so monitoring can surface the spawning process, the scripting interpreter invoking the dialog, and any exfiltration of captured credentials; however, the core mechanism — a fake GUI dialog that mimics legitimate OS prompts — is a UI-level event that monitoring does not directly observe, and a conformant implementation focused on network and application layers would miss it entirely, leaving a nameable remainder rather than a bounded one.
- T1056.003partialdetects — A.8.16's monitoring scope includes web portal traffic and anomalous behaviour such as unauthorized code execution, but detecting injected credential-capture code on a legitimate login page requires application-layer instrumentation that the clause permits rather than mandates, so what is caught is a slice the implementer chose.
- T1056.004partialdetects — A.8.16's scope includes host-level signals relevant to API hooking (process behaviour, code-authorization checks, security-tool logs) and explicitly names 'process injection' as an example of unusual system behaviour to monitor, but the clause sets scope rather than mandating instrumentation depth — an implementation monitoring only network and access layers is fully conformant and sees nothing of an in-process hook, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1059.002partialdetects — A.8.16's scope includes event logs and unusual system behaviour, which can surface AppleScript execution (osascript process spawning, AppleEvent traffic, script file writes), but the clause sets scope by business requirement rather than mandating endpoint telemetry depth — an implementation monitoring only network and application layers is fully conformant and sees nothing of in-process AppleScript abuse.
- T1059.003partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but its scope is set by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a command shell executing locally.
- T1059.004partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but shell execution is a legitimate administrative activity and the control's scope is set by business requirements, so detection depends on whether the implementation monitors command-level telemetry and baselines normal shell usage.
- T1059.005partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but its scope is set by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a VB payload executing in-process.
- T1059.006partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but Python execution is a legitimate administrative activity whose malicious use is detectable only when it deviates from the established baseline, and the clause determines scope rather than mandating the depth of instrumentation that would surface every Python-based technique.
- T1059.007partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but its scope is set by business requirements rather than mandating host-level instrumentation depth, so JavaScript execution outside monitored contexts (e.g., obfuscated in-memory JXA via OSAKit) may evade detection.
- T1059.008partialdetects — A.8.16 explicitly includes networking equipment in its monitoring scope (item b) and covers network traffic, configuration files, event logs and resource use, so SSH/telnet sessions to devices, config changes and anomalous command patterns are observable. However, the technique notes adversaries can disable security features or logging on the device itself, local console access generates no network traffic to monitor, and many CLI actions resemble legitimate admin activity, so what is detected is a slice rather than the bulk.
- T1059.009partialdetects — Cloud API calls generate network traffic and event logs that fall within A.8.16's monitoring scope (items a, b, e), and behavioral baselines can flag anomalous API usage patterns, but the technique uses legitimate APIs with valid credentials, making detection depend on the organization's chosen scope and baseline depth rather than being guaranteed by the clause.
- T1059.010partialdetects — A.8.16 enumerates monitoring of security-tool logs (antivirus, IDS/IPS), code-authorization checks, and unusual system behaviour — all of which can surface AutoHotKey/AutoIT script execution — but the clause sets scope per organizational requirements rather than mandating instrumentation depth, so a conformant implementation focused only on network-layer monitoring would miss this technique.
- T1059.011partialdetects — A.8.16's monitoring scope is set by organizational requirements rather than mandating host-level process-execution telemetry, so Lua execution is observable where the implementation includes process creation, command-line, or script-execution logging (and the clause's item (f) on authorized code execution supports this), but invisible where monitoring is limited to network or application layers — the same scope-determined remainder that grounds the T1055 anchor.
- T1059.012partialdetects — A.8.16's scope is set by organizational requirements rather than mandating hypervisor-specific instrumentation; ESXi hypervisors generate event logs and resource-use telemetry that fall within the clause's monitoring items, but an implementation focused on network/application layers is fully conformant and would miss hypervisor CLI abuse, and legitimate admin use of esxcli/vim-cmd is hard to distinguish from malicious use without behavioral baselines tuned to the hypervisor.
- T1059.013partialdetects — Container CLI/API abuse generates multiple observable signals within A.8.16's scope — process execution of docker/kubectl binaries, network calls to the docker daemon and Kubernetes API server, resource usage changes from container creation, and audit/event logs — and the clause explicitly lists process authorization checks (f), resource monitoring (g), and known attack characteristics (c) as in-scope; the bounded remainder is the scope-setting nature of the clause and the difficulty of distinguishing malicious from legitimate use of these admin tools without a tuned baseline.
- T1069partialdetects — A.8.16's scope includes event logs, access logs, and unusual user/system behaviour against a baseline — telemetry that can surface permission enumeration commands (process creation, directory service queries, etc.). But the clause sets scope by business requirements rather than mandating instrumentation depth, so a conformant implementation monitoring only network traffic or application layers would miss local enumeration, and built-in discovery commands often look legitimate without specific baselining.
- T1069.001partialdetects — A.8.16's scope includes event logs and unusual system behaviour, so local group enumeration via `net localgroup` or `groups` is observable where process execution and command-line auditing fall within the monitoring scope; but the clause sets scope by business requirements rather than mandating instrumentation depth, so an implementation monitoring only network and access layers is fully conformant and sees nothing of a local discovery command.
- T1069.002partialdetects — Domain group enumeration generates observable artifacts (LDAP queries to domain controllers, process execution events, directory service access logs like 4662) that fall within A.8.16's scope of access monitoring, event logs, and unusual user behaviour against baseline, but the clause sets scope by business requirements rather than mandating instrumentation depth, so a conformant implementation focused on network perimeter or application layers can miss internal enumeration, and the technique is a low-noise discovery activity that often blends with legitimate admin behaviour unless specifically baselined.
- T1069.003partialdetects — Cloud group enumeration generates API calls that produce cloud audit logs (CloudTrail, Azure Activity Log, GCP Audit Logs) and network traffic to cloud provider endpoints, all within A.8.16's scope of access monitoring and event logs; however, the technique uses legitimate authenticated API calls that look like normal administrative activity, so detection depends on whether cloud audit logs are ingested into the monitoring system and whether behavioral baselines catch enumeration patterns — neither of which A.8.16 mandates, leaving an implementation focused on traditional on-premises monitoring blind to it.
- T1070partialdetects — Logging and real-time monitoring of system and network activity make it harder for adversaries to clear traces of their presence without detection.
- T1070.003partialdetects — A.8.16's monitoring of event logs, file operations, and anomalous behavior against baseline can surface the clearing commands themselves (history -c, Clear-History, rm .bash_history), the file deletions, and the resulting absence of expected history; the bounded remainder is that adversaries with sufficient access can also tamper with the very logs that would record the clearing, and in-place edits to history files are less visible than full deletions.
- T1070.005partialdetects — T1070.005's `net use /delete` execution and the resulting share disconnection fall within A.8.16's scope (event logs, system activity, process execution), but the clause sets scope by business requirements rather than mandating command-line or process-execution auditing depth, so an implementation focused on network traffic or access logs alone is fully conformant and sees nothing of the cleanup command.
- T1070.006partialdetects — A.8.16's scope includes event logs, security tool logs, baseline comparison of normal behaviour, and detection of unusual system behaviour — all of which can surface timestomping through file integrity monitoring, $SI/$FN divergence checks, or API-call telemetry — but the technique is explicitly designed to evade forensic and file-analysis detection by mimicking legitimate timestamps, and a conformant implementation focused on network and access monitoring (as the clause permits) would see nothing of file metadata manipulation.
- T1070.007partialdetects — A.8.16's monitoring scope includes system and network event logs and configuration files, so deletion of connection history artifacts is observable where those logs are within the implemented scope, but the clause determines scope by business requirements rather than mandating universal instrumentation, leaving a slice of implementations that would not capture this activity.
- T1070.008partialdetects — A.8.16's monitoring scope includes event logs and application activity, which can surface the deletion of mailbox data, but the control determines scope by business requirements rather than mandating mailbox-level instrumentation, so detection is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1070.009partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and clearing persistence artifacts generates observable system changes (file deletions, registry modifications, service removals) that fall within its scope, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of these host-level artifacts.
- T1070.010partialdetects — A.8.16's monitoring scope can include file system events, process execution logs, and security-tool telemetry (antivirus, IDS) that would surface copy/rename operations and known malware signatures at new locations, but the technique is explicitly designed to evade these by renaming payloads to match legitimate resources and moving them into expected paths; detection depends on whether file-integrity and process telemetry are within the scope the clause leaves to organizational requirements, and the renamed payload may not trigger signature-based tools.
- T1071.005partialdetects — A.8.16 covers outbound/inbound network traffic and IDS/IPS logs, and its baseline-of-normal-behaviour framing can surface unusual pub/sub use, but the clause sets scope by business requirements rather than mandating protocol-level inspection depth; an implementation monitoring only basic traffic flows is fully conformant and would miss commands embedded inside legitimate-looking MQTT/XMPP/AMQP/STOMP traffic, and the technique is explicitly designed to blend in.
- T1072partialdetects — A.8.16's monitoring scope explicitly covers outbound/inbound network traffic, code execution authorization checks (item f), event logs, and unusual system behaviour against a baseline — all of which surface T1072's abuse of deployment tools through anomalous management-server activity, mass endpoint execution, and resource spikes; the bounded remainder is that the technique is designed to mimic legitimate administrative action, so detection depends on the monitoring being configured to baseline the management tool's normal behaviour rather than being automatic from the clause alone.
- T1074.002partialdetects — A.8.16's monitoring scope includes network traffic, system access, and event logs, which can surface the lateral data movement and centralization characteristic of staging, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of local file copies.
- T1078partialresponds — A.8.16 surfaces anomalous use of valid accounts (e.g. unusual login times/locations, unauthorized access attempts, deviations from baseline) once the technique is underway and feeds those indicators to 5.26 response procedures, but the control itself performs none of the core respond actions (contain, eradicate) and many T1078 vectors (inactive accounts, legitimate-looking pivots without malware) produce no observable anomaly at all.
- T1078.001partialdetects — A.8.16 monitoring captures authentication events, access logs, and unauthorized access attempts that surface default account abuse, but it does not inherently know which accounts are 'default' — detection depends on the organization configuring the baseline to flag known default accounts or treating their use as anomalous, and legitimate-looking use of an enabled default account may not deviate from baseline.
- T1078.002partialdetects — A.8.16 explicitly enumerates behavioral baselines (usual time, location, frequency of access per user/group) and 'unusual user and system behaviour in relation to expected behaviour' as monitoring targets, which is the mechanism for surfacing valid domain account abuse; however, the clause sets scope by business requirements rather than mandating UEBA depth, and adversaries using valid domain accounts can mimic the legitimate user's pattern, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1078.003partialdetects — A.8.16 explicitly covers access monitoring, event logs, and behavioural baselines (time, location, frequency of access), which surface anomalous use of local accounts even when credentials are valid; the remainder is implementations that omit host-level authentication logging or fail to establish behavioural baselines, both within the clause's scope-setting language.
- T1078.004partialdetects — A.8.16's behavioural baselining and anomaly detection (unusual user/system behaviour, time/location/frequency deviations, unauthorized access) can surface anomalous use of valid cloud accounts where cloud control-plane logs and identity-provider telemetry are inside the monitoring scope, but the clause sets scope by organisational requirements rather than mandating cloud-specific instrumentation, so an implementation focused on traditional network/system layers is fully conformant and sees nothing of cloud account abuse.
- T1080partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but tainted shared content executes on the remote system when a user opens a file, and whether that execution is visible depends on whether the implementation's monitoring scope reaches the endpoint's process and file-access telemetry rather than only network and application layers.
- T1082partialdetects — A.8.16's scope includes event logs and unusual system/user behaviour against a baseline, which can surface T1082 via process execution telemetry (e.g., `systeminfo`, `uname -a`, `show version`) when command-line capture and per-user baselines are in place; however, the activity is inherently low-noise and indistinguishable from routine administration, so detection depends on the implementation's depth (process-level logging, command-line argument capture, behavioural baselining) rather than being inherent to the clause, and many conformant implementations focused on network/access telemetry would miss it.
- T1083partialdetects — A.8.16's scope includes event logs, system activity, and unusual system behaviour, which can surface process execution and command-line activity associated with T1083 enumeration commands; however, the clause sets scope by business and security requirements rather than mandating host-based process/command-line instrumentation, so a network-only or traffic-focused implementation is fully conformant and sees nothing of an on-host enumeration.
- T1087partialdetects — A.8.16's scope includes event logs, security tool logs, and behavioural baselines against which account enumeration creates anomalies, but the clause sets scope by business requirements rather than mandating the specific telemetry (process creation auditing, command-line logging, directory-service access logs, cloud audit logs) needed to surface T1087; an implementation monitoring only network-layer traffic is fully conformant and would miss local or API-based account discovery.
- T1087.001partialdetects — A.8.16's monitoring scope includes system event logs and unusual system behaviour, which can surface local account enumeration commands, but the clause determines scope rather than mandating specific command-level logging, so detection depends on implementation choices and is not guaranteed.
- T1087.003partialdetects — T1087.003 runs on the estate (PowerShell cmdlets against Exchange, authenticated directory queries) and is observable where A.8.16's scope includes event logs, access to critical applications, and security-tool logs — but the clause sets scope by business requirements rather than mandating PowerShell/Exchange audit depth, so an implementation monitoring only network traffic is fully conformant and sees nothing of the enumeration.
- T1087.004partialdetects — A.8.16's scope includes event logs, access logs, and network traffic, which can surface cloud API enumeration calls (CloudTrail, Azure Activity Log, GCP audit logs) and the CLI/PowerShell invocations that drive them; however, the clause sets scope by business requirements rather than mandating cloud control-plane log ingestion, and the technique uses legitimate authenticated calls that look like normal admin activity, so detection depends on whether cloud audit telemetry is in scope and whether baselines for enumeration frequency are established.
- T1090partialdetects — A.8.16's monitoring scope includes network traffic and anomalous behaviour patterns, which can surface proxy usage, but the clause determines scope by business requirements rather than mandating the deep packet inspection or TLS decryption needed to reliably identify proxied C2, so detection is a slice the implementer chooses.
- T1090partialresponds — A.8.16 surfaces anomalous proxying patterns (e.g. unexpected traffic, known C2 domains, process injection, or deviations from baseline) once underway and feeds them to 5.26 response procedures, satisfying the core of `responds`; it is only partial because the clause's scope is set by organizational requirements rather than mandating coverage of every proxy variant or platform, and many stealthy or in-band proxy uses produce no observable anomaly at all.
- T1090.001partialdetects — A.8.16's monitoring scope includes outbound/inbound traffic and unusual system behaviour, which can surface internal proxy traffic patterns, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an internal proxy.
- T1090.002partialdetects — A.8.16's scope explicitly includes outbound/inbound network traffic, IDS/IPS logs, and behavioural anomalies, and external proxy usage manifests as observable network traffic to non-corporate destinations — but the technique is specifically designed to evade detection by using clean infrastructure and mimicking legitimate HTTPS, so detection depends on threat intelligence coverage and baseline tuning that the clause leaves to organisational requirements rather than mandating.
- T1090.003partialdetects — A.8.16 monitors network traffic and can surface the last-hop proxy connection, but the multi-hop chain beyond the organization's perimeter is outside its monitoring scope, so detection is limited to the final link.
- T1090.004partialdetects — A.8.16's scope includes outbound network traffic and explicitly lists 'deviations in use of standard protocols' as an anomaly to detect, and an SNI/Host header mismatch is exactly such a deviation — but the clause sets scope by business requirements rather than mandating TLS deep-packet inspection, and domain fronting is engineered to look like legitimate HTTPS traffic to a trusted CDN, so detection depends on implementation choices the clause does not require.
- T1091partialdetects — A.8.16's event-log and system-activity monitoring can surface removable-media insertion, Autorun execution and file writes to USB devices, but the technique is specifically designed for air-gapped or disconnected systems where monitoring coverage is typically absent or reduced, and firmware-level media manipulation falls outside the clause's scope.
- T1092partialdetects — A.8.16's scope includes system event logs and 'unusual system behaviour' monitoring, which can surface removable media insertion and file transfer activity, but the clause's primary scope is network traffic and the technique is specifically designed to operate on disconnected networks where network monitoring is blind; an implementation focused on network layers is fully conformant and sees nothing of the removable-media C2, so detection depends on host-level instrumentation the clause does not mandate.
- T1095partialdetects — A.8.16's scope includes outbound/inbound network traffic and deviations in use of standard protocols, so where monitoring is configured to cover non-application layer protocols (ICMP, UDP, VMCI) anomalous C2 can be surfaced; however, the clause sets scope by business requirements rather than mandating depth, and the technique description explicitly notes these protocols are 'not as commonly monitored' and that VMCI traffic is invisible to external monitoring tools, leaving a substantial slice undetected.
- T1098.001partialdetects — Cloud credential additions (CreateKeyPair, CreateAccessKey, service principal credential adds, app password creation) generate audit log entries in CloudTrail, Azure Activity Log, GCP Audit Logs, and Entra ID audit logs, which fall within A.8.16's scope of security tool logs and event logs; however, the clause sets scope by business and security requirements rather than mandating cloud control plane instrumentation, so an implementation monitoring only network and host layers is fully conformant and sees nothing of this technique — the same shape as the T1055 anchor.
- T1098.002partialdetects — A.8.16's scope is set by business requirements rather than mandating specific instrumentation, so detection depends on whether the implementation ingests PowerShell execution logs (Add-MailboxPermission cmdlet), Office 365 unified audit logs, and Google Workspace admin audit logs; an implementation monitoring only network traffic or basic system logs is fully conformant yet sees nothing of cloud-mailbox permission grants made via web consoles.
- T1098.003partialdetects — A.8.16's scope includes event logs relating to system and network activity (clause e) and access to critical applications (clause b), which can encompass cloud control plane logs (CloudTrail, Azure AD audit, Office 365 unified audit) that record role/permission additions; however, the clause sets scope by business and security requirements rather than mandating cloud audit log ingestion, so a fully conformant implementation focused on on-premises telemetry would miss this technique entirely, and the variant where roles are added to adversary-controlled accounts outside the victim tenant produces no event on the monitored estate at all.
- T1098.004partialdetects — A.8.16 explicitly covers monitoring of critical configuration files (authorized_keys, sshd_config), access events, and system event logs — all primary indicators for this technique. File integrity monitoring, auth.log analysis, and cloud API audit logs would surface the modification. The grade is `mostly` rather than `full` because the clause sets scope by business requirements rather than mandating instrumentation depth: an implementation focused on network traffic alone would miss the file-level modification, and cloud-API variants require cloud audit logs to be in scope.
- T1098.005partialdetects — A.8.16's scope includes event logs, access monitoring, and detection of unusual user/system behaviour, which can surface IdP device registration events where identity-provider logs are within the monitored estate; however, the clause sets scope by business and security requirements rather than mandating IdP-level instrumentation, so a fully conformant implementation focused on network and host telemetry would not see this technique.
- T1098.006partialdetects — A.8.16's scope includes event logs, configuration file monitoring, and access to critical applications — all of which can surface Kubernetes RBAC/ABAC modifications if the monitoring implementation extends to the container orchestration control plane — but the clause sets scope by business requirements rather than mandating Kubernetes audit-log instrumentation, so a conformant implementation focused on traditional infrastructure would miss this technique entirely.
- T1098.007partialdetects — A.8.16 explicitly lists unauthorized access attempts and unusual user behaviour as monitoring targets, and adding a user to a privileged group is observable in event logs and access records, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a local group change.
- T1102partialdetects — A.8.16's monitoring scope covers outbound network traffic, baseline anomaly detection, and deviations in standard protocol use, which can surface some T1102 activity through beaconing analysis, unusual data volumes to web services, and anomalous API usage patterns; however, the technique is specifically designed to blend with legitimate traffic to common services (Google, Microsoft, Twitter) using standard HTTPS, so a substantial portion of T1102 C2 is indistinguishable from normal baseline behavior and falls outside what monitoring can reliably flag.
- T1102.001partialdetects — A.8.16's scope covers outbound network traffic, IDS/IPS logs, and behavioural baselines against which anomalous patterns can be surfaced, and the victim does contact the estate's network when reaching the legitimate web service — so there is a vantage point, unlike T1583.001 where the `none` anchor sits. But the technique is purpose-built to blend in: traffic goes to popular legitimate services (Google, Twitter) that hosts already contact, and SSL/TLS encryption hides the embedded resolver content from inspection. Detection depends on specific implementation choices — behavioural baselining of access frequency/timing to particular services, IDS signatures for known dead-drop patterns, or analysis of post-redirect follow-on connections — that go beyond a conformant baseline monitoring implementation. A network-and-application-layer scope that is fully conformant with A.8.16 can still miss most instances.
- T1102.002partialdetects — A.8.16's scope includes outbound network traffic, DLP logs, and behavioural baselines against which 'unusual user and system behaviour' is checked, so beaconing patterns, atypical access frequency/timing to a web service, or unusual data volumes posted to a legitimate service are observable in principle; however, T1102.002 is specifically designed to blend with traffic to sites the estate already contacts and the channel is TLS-encrypted, so signature and content inspection largely fail and detection depends on the behavioural and DLP layers being implemented and tuned — a slice, not the bulk.
- T1102.003partialdetects — A.8.16's scope includes outbound network traffic, IDS/IPS logs, baseline-of-normal-behaviour comparison, and detection of deviations in standard protocols — all of which can surface beaconing patterns, anomalous volumes/frequencies, or known malicious indicators in one-way C2 traffic. But T1102.003 is specifically engineered to blend with legitimate encrypted traffic to popular services, and A.8.16 sets monitoring scope by business requirements rather than mandating instrumentation depth (TLS fingerprinting, deep packet inspection, behavioural beaconing analytics), so what is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1104partialdetects — A.8.16's monitoring scope includes network traffic and anomalous behaviour patterns, which can surface the loader's callback to a first-stage server, but the technique's deliberate separation of infrastructure and use of fallback channels means the second stage and the full C2 chain are not necessarily observable from the same vantage point.
- T1105partialdetects — Monitoring inbound and outbound traffic for signatures associated with malware or known malicious domains can detect and alert on unauthorized tool transfers.
- T1106partialdetects — A.8.16's scope is set by business and security requirements rather than mandating syscall-level or EDR instrumentation; an implementation monitoring only network traffic, access logs, and standard security-tool logs is fully conformant and sees little of the native API calls themselves, though downstream effects (process creation, file writes, network connections) and behavioural baselines can surface a slice of abuse.
- T1110partialresponds — A.8.16 surfaces brute-force indicators (failed logins, anomalous access patterns, unusual login times/locations, traffic to/from malicious IPs) in real time or near-real time and feeds them to the 5.26 response process for containment and eradication once the technique is underway.
- T1111partialdetects — A.8.16 explicitly names keystroke logging and process injection as examples of unusual system behaviour the monitoring system should identify, so the keylogger-based vectors of T1111 (smart card password capture, hardware token PIN capture) are within scope; however, the SMS-service-provider compromise and algorithm/seed theft vectors occur entirely off the organisation's estate and have no observable artefact for A.8.16 to surface.
- T1112partialdetects — A.8.16's scope includes 'critical or admin level system and network configuration files' and 'event logs relating to system and network activity,' both of which can surface registry modifications (e.g., Sysmon Event ID 13, Windows Security logs), but the clause sets scope by business and security requirements rather than mandating endpoint configuration or registry-hive monitoring, so an implementation focused on network and application layers is fully conformant and sees nothing of an in-host registry write.
- T1113partialdetects — A.8.16 can surface screen capture through host-level telemetry (process behaviour, resource use, security-tool logs catching screen-capture API calls or RAT activity), but the clause sets scope rather than mandating host instrumentation depth — a network- and application-focused implementation is fully conformant and sees nothing of an in-process screen grab, and legitimate apps use the same APIs.
- T1114partialdetects — A.8.16 monitors network traffic, access to systems, and event logs, which can surface anomalous email access patterns or forwarding rules, but the clause determines scope rather than mandating specific email-server instrumentation, so detection depends on implementation choices.
- T1114.001partialdetects — A.8.16's scope includes access to systems and event logs, which can surface local email file access (.ost/.pst reads) and anomalous process behavior, but the clause sets monitoring scope by business and security requirements rather than mandating file-access or endpoint telemetry depth — an implementation focused on network traffic is fully conformant and sees nothing of this local technique.
- T1114.002partialdetects — A.8.16 can surface Remote Email Collection through access logs and unusual user behaviour when email systems fall inside its discretionary monitoring scope, but it does not guarantee coverage of cloud/application logs and may not distinguish valid-credential collection from normal mailbox access.
- T1114.003partialdetects — A.8.16's monitoring scope is set by business and security requirements rather than mandating email-system instrumentation; rule creation, MAPI modifications and forwarding activity are observable in principle through Exchange/IIS/EWS logs and outbound mail flow, but an organization monitoring only network and application layers (fully conformant with A.8.16) sees nothing of an in-mailbox rule change, and hidden MAPI rules are specifically designed to evade standard visibility.
- T1115partialdetects — A.8.16 can surface clipboard access where the monitoring scope includes process execution logs (capturing clip.exe, Get-Clipboard, pbpaste invocations) and host-level telemetry, but the clause sets scope by business and security requirements rather than mandating clipboard-specific instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process clipboard read.
- T1119partialdetects — A.8.16's monitoring scope includes system and network activity logs, resource usage, and anomalous behaviour patterns, which can surface automated collection tools running scripts or moving data, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of in-process collection.
- T1120partialdetects — A.8.16's scope is set by business and security requirements rather than mandating process-execution or command-line telemetry, so peripheral enumeration via legitimate tools (lsusb, wmic, system_profiler) is observable only where that scope includes it; an implementation monitoring only network and application layers is fully conformant and sees nothing of this on-host discovery.
- T1124partialdetects — A.8.16's scope includes event logs relating to system activity (item e) and unusual user/system behaviour against a baseline, so process execution logging with command-line capture would surface `net time`, `w32tm /tz`, `systemsetup -gettimezone` and similar commands; however, the clause sets scope by business requirements rather than mandating command-line auditing depth, and these commands are inherently low-noise and indistinguishable from normal system administration, so detection depends on the implementation's chosen scope and baseline granularity.
- T1125partialdetects — A.8.16's scope (process execution, file system writes, resource use, outbound traffic, security tool logs, 'unusual system behaviour') covers the telemetry that surfaces video capture — the malware/script invoking the camera API, the resulting image/video files written to disk, and any exfiltration — but the clause sets scope by business and security requirements rather than mandating instrumentation depth, so a fully conformant implementation monitoring only network and access layers sees nothing of an on-host capture that stores locally before any later transfer.
- T1127.001partialdetects — A.8.16 explicitly lists checking that executed code is authorized and has not been tampered with, and MSBuild abuse (unsigned inline tasks compiled by a signed binary) is observable through process execution telemetry, command-line logging, and network monitoring — but the clause sets scope by business and security requirements rather than mandating process-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1127.002partialdetects — A.8.16's scope includes process creation events, command-line auditing, security tool logs, and network traffic — all of which surface ClickOnce abuse (DFSVC.EXE spawning children, rundll32.exe dfshim.dll, anomalous downloads). However, the clause sets scope by business requirements rather than mandating host-level telemetry depth, so a conformant implementation monitoring only network or application layers can miss the process-level artifacts, matching the same scope-determined remainder that graded A.8.16 vs T1055 at partial.
- T1127.003partialdetects — A.8.16's monitoring scope includes process execution, event logs, and item (f) on checking that executed code is authorized and untampered — all directly relevant to JamPlus proxying a malicious script through a legitimate build tool — but the clause sets scope by business requirements rather than mandating instrumentation depth, so a conformant implementation monitoring only network and application layers sees nothing of the in-process script execution through the trusted developer tool.
- T1129partialdetects — A.8.16's scope is set by business and security requirements rather than mandating instrumentation depth; the clause explicitly lists process-level telemetry sources (security tool logs, code-authorization checks, process injection as an example anomaly) that would surface shared module loads, but an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process DLL/.so load.
- T1132partialdetects — A.8.16 monitoring can surface some encoded C2 traffic via IDS/IPS signatures, baseline-deviation analysis, and detection of unusual protocol use, but T1132 is explicitly designed to evade detection by adhering to standard protocol specifications (Base64, MIME, etc.) that look like legitimate traffic, and A.8.16's scope is set by organizational requirements rather than mandating deep inspection of all encoded payloads.
- T1132.001partialdetects — A.8.16's monitoring scope covers outbound/inbound network traffic, IDS/IPS logs, and deviations in use of standard protocols, so the data sources are in scope; however, T1132.001 is specifically designed to make C2 traffic look like normal protocol traffic (Base64, ASCII, gzip in HTTP/DNS are routine), so the encoding itself is not a signature — detection depends on behavioral baselines, known-malicious destinations, or protocol deviations, which catch a slice rather than the bulk.
- T1132.002partialdetects — A.8.16 monitors outbound network traffic and includes detection of 'deviations in use of standard protocols' and behavioural anomalies, so non-standard encoding in C2 traffic does pass through a vantage point where it can be surfaced; however, the technique is specifically designed to evade signature and content inspection, and the clause sets monitoring scope by business requirements rather than mandating deep protocol-content analysis, so what is caught is a slice rather than the bulk.
- T1134.001partialdetects — Token impersonation/theft is observable through Windows security event logs (e.g., 4624/4672/4648), Sysmon/EDR telemetry on token API calls, and behavioural baselines of privilege usage — all within A.8.16's scope of system event logs and unusual system behaviour — but the technique operates at the process/API level, so an implementation monitoring only network and application layers is fully conformant with A.8.16 and sees nothing of it, mirroring the partial grade on T1055.
- T1134.002partialdetects — A.8.16's scope is set by business and security requirements rather than mandating specific instrumentation depth; detecting T1134.002 requires host-level telemetry (process creation events with token information, specific audit policies) that an organization monitoring only network and application layers — fully conformant with A.8.16 — would not see, mirroring the A.8.16 vs T1055 anchor.
- T1134.003partialdetects — A.8.16's scope includes event logs and unusual user/system behaviour, which can surface token creation/impersonation through Windows security events (4624, 4648, 4672) and anomalous authentication patterns; however, the clause sets scope by business and security requirements rather than mandating host-level audit policy, so a conformant implementation focused on network and application layers would miss this technique, paralleling the T1055 anchor.
- T1134.004partialdetects — A.8.16's scope includes system and process monitoring (explicitly citing 'process injection' and 'unusual system behaviour' as examples), and PPID spoofing is observable where host telemetry captures parent-child process relationships and detects inconsistencies between claimed and actual call chains; however, the clause sets scope by business and security requirements rather than mandating this instrumentation depth, so a conformant implementation monitoring only network or application layers sees nothing of an in-process PPID manipulation.
- T1134.005partialdetects — A.8.16's scope includes event logs and unusual user/system behaviour, which can surface SID-History changes on domain controllers (e.g., event 4782/5136), but the clause sets scope by business and security requirements rather than mandating AD attribute monitoring, so a conformant implementation focused on network and application layers would see nothing of this technique.
- T1135partialdetects — T1135 generates observable artifacts within A.8.16's scope — SMB session requests, authentication events, command execution logs, and access attempts to remote systems — but the clause sets scope by business and security requirements rather than mandating instrumentation depth, so a conformant implementation monitoring only perimeter traffic or application-layer events may miss host-level discovery commands like `net view`.
- T1136.002partialdetects — A.8.16's monitoring scope includes access to systems and event logs, which can surface domain account creation events, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a domain account creation.
- T1136.003partialdetects — A.8.16 can detect cloud account creation when cloud admin/audit logs are included in its monitoring scope, but the control leaves that scope to business requirements and the technique can be tuned to reduce detection.
- T1137partialdetects — A.8.16's scope includes event logs, code-execution authorization checks, and unusual system behaviour — all of which can surface Office startup persistence (template macros, add-ins, Outlook rules) through host telemetry. But the clause sets scope by business and security requirements rather than mandating the host-based instrumentation depth needed to catch every persistence mechanism, so an implementation focused on network traffic alone is fully conformant and sees nothing of this technique.
- T1137.001partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and template macro modification leaves file-system and registry artefacts within the monitoring scope, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a local template change.
- T1137.002partialdetects — A.8.16 explicitly includes monitoring of critical system configuration files (item c) and event logs (item e), and the Office Test Registry key modification plus the resulting DLL load are observable through these channels; however, the clause sets scope by business and security requirements rather than mandating Registry or process-load instrumentation depth, so a conformant implementation monitoring only network and application layers would miss the in-host persistence mechanism.
- T1137.003partialdetects — A.8.16's scope includes security tool logs (AV/EDR), event logs, code execution authorization checks, and unusual system behaviour — all of which can surface Outlook form persistence (form template additions to the mailbox, code execution within Outlook's process, spawned child processes). However, the clause sets monitoring scope by business and security requirements rather than mandating host-level instrumentation depth, so an implementation focused on network and application-layer monitoring is fully conformant and sees nothing of this in-application persistence technique.
- T1137.004partialdetects — A.8.16's monitoring scope can surface T1137.004 through configuration-file monitoring (Outlook Home Page registry settings), code-authorization checks (item f), and network traffic if an external URL is fetched, but the clause sets scope by business requirements rather than mandating the specific instrumentation needed; an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process Outlook HTML load.
- T1137.005partialdetects — A.8.16's scope includes process, event, and configuration monitoring, which can surface Outlook rule abuse via mailbox audit logs, Outlook-spawned child processes, and Exchange server-side rule creation events — but the clause sets scope by business requirement rather than mandating instrumentation depth, so a network-only or access-only implementation is fully conformant and sees nothing of the rule creation or its dormant trigger.
- T1137.006partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and add-in loading is observable through file-system, registry and process-start events, but the clause determines scope by business requirements rather than mandating host instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an Office add-in persistence mechanism.
- T1140partialdetects — T1140 runs on the estate and produces observable artifacts (process execution of certutil/copy/type with decode arguments, new files written, command-line telemetry) that fall within A.8.16's scope of system/application activity logs and security-tool logs; however, the clause sets monitoring scope by business requirements rather than mandating endpoint process or command-line instrumentation, so a fully conformant implementation focused on network traffic alone would miss most of the technique.
- T1176partialdetects — A.8.16's scope is set by organizational requirements and includes code-execution verification (item f), security-tool logs (item d), and behavioural baselines that can surface extension installations, host-app activity, and C2/tunnelling traffic — but the technique's own description notes detection is challenging because extensions inherit host-app trust and blend into normal workflows, and a baseline implementation focused on network layers is fully conformant yet blind to in-application extension behaviour, leaving a nameable slice rather than a bounded remainder.
- T1176.001partialdetects — A.8.16's monitoring scope covers network traffic, configuration file changes, process behaviour and resource use — all of which a malicious browser extension touches (C2/exfil traffic, Preferences/Secure Preferences modifications, CPU/memory anomalies). However, the technique is specifically designed to operate within the trusted browser process and inherit its permissions, so the malicious activity blends with legitimate browser traffic and in-process execution; an A.8.16 implementation monitoring only network and application layers — fully conformant with the clause — sees little of the in-browser abuse, leaving a bounded but real remainder.
- T1176.002partialdetects — A.8.16 enumerates multiple monitoring dimensions directly relevant to a malicious IDE extension — code execution authorization checks (item f), process behavior anomalies including injection-like patterns (item d), resource use (item g), and network traffic (item a) — and the technique's downstream effects (arbitrary code execution, backdoors, crypto mining, exfiltration) produce observable signals across these dimensions; however, the extension runs within a legitimate IDE process, scope is set by business and security requirements rather than mandated, and distinguishing malicious extension activity from normal IDE behavior requires specific tuning, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1185partialdetects — A.8.16 explicitly names 'process injection' and unusual system behaviour as items to monitor for, and the proxy/pivoting traffic produces observable network patterns, but the clause sets scope by business requirements rather than mandating instrumentation depth — an implementation monitoring only network flows sees the proxy traffic but not the browser injection, and one monitoring only host telemetry sees the injection but may miss the intranet pivoting, so coverage is a slice determined by implementation choices rather than a bounded remainder of a mechanism that mostly works.
- T1185partialresponds — A.8.16 surfaces anomalous behaviour (process injection into browser, unusual network patterns, resource deviations) once the hijacking is underway and feeds it to 5.26 response procedures for containment/eradication; partial because the clause sets scope by organisational requirements rather than mandating universal coverage of all browser-process or proxy-pivot artefacts, and some realisations (e.g. silent certificate inheritance without observable deviation) stay outside monitored baselines.
- T1187partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and forced authentication via SMB/WebDAV generates observable network traffic to external hosts, but the control's scope is set by business requirements rather than mandating the depth of host telemetry needed to catch every variant, so what is caught is a slice the implementer chose.
- T1189partialdetects — A.8.16's monitoring scope explicitly includes outbound/inbound network traffic, web filters, IDS/IPS, and detection of traffic to known malicious domains and known attack characteristics — all of which surface drive-by compromise via malicious infrastructure or signature matches, plus post-exploitation behavior (process injection, resource anomalies). However, scope is set by business requirements rather than mandated, so an implementation monitoring only network perimeter sees watering-hole exploits on legitimate sites and zero-day browser exploits that lack signatures, leaving a meaningful slice undetected.
- T1195.001partialdetects — A.8.16 can detect runtime indicators of a compromised dependency executing on the estate (anomalous network calls, process behavior, resource use, unauthorized code execution per item f), but the supply chain compromise itself — the poisoning of the package on a registry or CI/CD component — happens on external infrastructure outside A.8.16's monitoring scope, and well-crafted malicious payloads are designed to blend with legitimate dependency behavior.
- T1199partialdetects — A.8.16's monitoring scope covers access events, authentication logs, and baseline-based anomaly detection (including time, location, and frequency of access per user/group), so third-party access events are visible and deviations from the trusted party's normal pattern can surface. However, T1199 is specifically designed to abuse existing trusted relationships that 'receive less scrutiny,' and the adversary's activity uses valid accounts performing actions that look legitimate, so distinguishing a compromised third party from a normally-behaving one is a slice rather than a bounded remainder of the mechanism.
- T1200partialdetects — A.8.16's monitoring scope covers network traffic, system access logs, and event logs, so some hardware additions are observable — a new wireless AP or networked implant generates traffic (a), USB enumeration can appear in system logs (e), and resource monitoring (g) may catch a device drawing power or bandwidth. But the technique's core mechanism is physical introduction of a device, and many variants — passive network taps, keystroke injection devices, DMA attacks — are designed to be invisible to software-based monitoring, so what is caught is a slice rather than the bulk.
- T1201partialdetects — A.8.16's scope includes event logs and unusual user/system behaviour against a baseline, so password-policy-discovery commands (net accounts, Get-ADDefaultDomainPasswordPolicy, chage -l, etc.) are observable in principle via process-execution and command-line telemetry; however, the clause sets scope by business and security requirements rather than mandating the specific instrumentation, and these commands resemble legitimate administrative activity, so an implementation focused on network traffic or security-tool logs alone would miss the technique entirely.
- T1202partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and indirect command execution via utilities like forfiles or pcalua.exe produces observable process creation and command-line events, but the control's scope is set by business requirements rather than mandating the depth of endpoint telemetry needed to surface every indirect execution path.
- T1203partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but exploitation for client execution often manifests as a brief, legitimate-looking process launch that may not trigger the configured baselines or signatures unless the monitoring scope is tuned to detect the specific exploit technique.
- T1204partialdetects — A.8.16's monitoring scope includes system and application event logs, network traffic, and unusual user behaviour, which can surface the follow-on execution of a malicious file or link, but the control's scope is determined by business requirements rather than mandating universal endpoint instrumentation, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1204.001partialdetects — A.8.16's monitoring scope includes network traffic and event logs, which can surface the outbound request and subsequent download or execution chain triggered by a malicious link, but the control's depth is determined by business requirements rather than mandating full endpoint instrumentation, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1204.002partialdetects — A.8.16 monitors for anomalous behaviour including unusual system behaviour and unauthorized code execution, but the technique's execution depends on a user opening a file, which is a user action that monitoring may surface indirectly through follow-on system events rather than directly observing the user's click.
- T1204.003partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered, and its anomalous-behaviour list explicitly names process injection and unusual system behaviour, so a backdoored image executing unexpected processes or resource use is observable where host-level telemetry is deployed; the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of the in-instance execution.
- T1204.004partialdetects — A.8.16 can detect the command execution (event logs, process creation, item f's code-authorization check) and downstream network behavior when command-interpreter and host telemetry are inside the monitoring scope, but the user action of being deceived into copying/pasting is not observable, and the clause sets scope by business requirements rather than mandating instrumentation depth.
- T1204.005partialdetects — A.8.16's enumerated scope (network traffic, security-tool logs, event logs, code-authorization checks, resource use) catches the malicious library's post-install behavior -- C2 traffic, anomalous process activity, unauthorized code execution, resource spikes -- but the initial user-driven installation event (e.g. `npm install`) is a user action outside the monitoring scope the clause sets, and an implementation monitoring only network/application layers is fully conformant yet sees nothing of the library's in-process execution, mirroring the T1055 anchor's reasoning.
- T1205partialdetects — T1205 is fundamentally network-based — the signaling IS the network traffic (port-knock sequences to closed ports, magic packets, crafted packets to network devices) — and A.8.16 explicitly requires monitoring inbound/outbound network traffic, IDS/IPS logs, and detection of known attack characteristics and deviations in standard protocol use, all of which surface traffic signaling; the remainder is sophisticated signaling that mimics legitimate traffic or rides encrypted channels where pattern matching fails.
- T1205partialresponds — A.8.16's real-time/periodic monitoring of network traffic, anomalous patterns, known attack characteristics, and alerts feeds directly into 5.26 response procedures once the signaling packets or triggered behavior are observed on the estate; partial because the control's scope is set by organizational requirements (not all signaling is guaranteed to match the baseline or be in monitored layers) and the core response (containment/eradication) lives in the referenced 5.26 clause rather than in A.8.16 itself.
- T1205.001partialdetects — A.8.16's scope explicitly covers inbound network traffic, firewall/IDS logs, and unauthorized access attempts — port knocking produces a sequence of connection attempts to closed ports that fall squarely within that scope and are detectable by IDS signatures and traffic-pattern analysis. However, the technique is specifically designed to look like ordinary failed connections, and detection depends on having signatures or anomaly rules configured for the knock sequence; a conformant A.8.16 implementation focused on basic logging without port-knocking signatures would miss it, leaving a nameable remainder.
- T1205.002partialdetects — A.8.16's network traffic monitoring can surface the crafted trigger packet and the resulting backdoor/C2 activation, and resource-use and event-log monitoring may catch secondary artifacts, but the technique is explicitly designed for low visibility — the dormant filter itself produces minimal CPU/memory footprint and limited raw-socket telemetry, so the core mechanism sits below what standard monitoring reliably sees.
- T1207partialdetects — A.8.16's broad scope (AD schema changes, replication traffic, access to domain controllers, unusual system behaviour) can surface some indicators of rogue DC registration, but the technique is explicitly designed to bypass the SIEM and system-logging sensors A.8.16 relies on, and the bypass is a named characteristic of T1207.
- T1210partialresponds — A.8.16 surfaces anomalous behaviour (e.g. process injection, unauthorized access, known attack patterns, unusual system behaviour) that can occur during or after T1210 exploitation and hands it to the 5.26 response procedure, but the core of `responds` (containment/eradication of an already-underway event) is performed by the separate incident-response control, not by monitoring itself.
- T1211partialdetects — A.8.16 monitors configuration files, security tool integrity, and unusual system behaviour, so some T1211 attempts that tamper with monitoring infrastructure or leave anomalous traces are surfaced; however, the technique is specifically designed to evade detection by suppressing logging, operating in unmonitored components, or exploiting blind spots, and A.8.16's scope is set by the organization rather than mandating depth across all evasion vectors.
- T1213partialdetects — A.8.16's scope includes access to critical applications and event logs, and its behavioral baseline (usual time, location, frequency of access per user) is well-suited to spotting anomalous repository mining, but the clause sets scope by business requirements rather than mandating repository coverage — a fully conformant implementation focused on network and infrastructure layers would see little of T1213, which often runs through authorized application sessions.
- T1213.001partialdetects — A.8.16's scope includes access to critical applications and event logs, and its baseline guidance explicitly covers usual time, location and frequency of access per user/group — all of which can surface anomalous Confluence access patterns (bulk reads, off-hours access, unusual repositories). However, the technique is fundamentally authorized-looking access to read content, so detection depends on the monitoring being configured with Confluence-specific behavioral baselines and application-layer instrumentation; a network- or host-focused implementation that is fully conformant with A.8.16 may see the access but not flag the data-mining behavior itself.
- T1213.002partialdetects — SharePoint access generates audit logs, event logs, and network traffic that fall within A.8.16's scope (access to systems/applications, event logs, behavioral baselines), but the technique is designed to look like legitimate repository use by a credentialed user reading documents they may be authorized to see, so detection depends on whether SharePoint is in the monitored scope and whether behavioral baselines are sophisticated enough to flag unusual access patterns.
- T1213.003partialdetects — A.8.16 can detect anomalous access patterns to code repositories (unusual times, locations, volumes, or users) via outbound traffic monitoring, access logs, and behavioural baselines, but cannot detect the technique when an adversary uses valid credentials and mimics normal developer activity — the scope is set by business requirements rather than mandating depth, and what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1213.004partialdetects — A.8.16 can detect some indicators of CRM data mining (unusual login times/locations, anomalous access patterns to the CRM application, large data exports traversing the network), but the core technique—querying and reading customer records within the SaaS CRM—occurs on the provider's infrastructure where the organization's monitoring has no native visibility, so detection depends on whatever access telemetry the SaaS platform exposes rather than on A.8.16's own instrumentation.
- T1213.005partialdetects — A.8.16 can surface some indicators of messaging-app data mining through network traffic to SaaS platforms, authentication/access logs, and endpoint telemetry, but the core activity — reading, searching and collecting messages within the SaaS application — happens on the provider's infrastructure outside the organization's monitoring scope, so what is caught is a slice rather than the technique.
- T1213.006partialdetects — A.8.16's scope explicitly includes access to systems and servers (databases qualify) and event logs relating to system activity, so database access logs and query patterns fall within standard monitoring; anomalous data-mining queries (volume, timing, source) deviate from the access baseline the clause requires. The bounded remainder is legitimate-credential access with normal-looking queries and implementations that monitor connections but not query-level detail, which is still conformant with A.8.16.
- T1216partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and not tampered with, and its anomalous-behaviour list explicitly names process injection, but system script proxy execution uses trusted signed scripts to run malicious files, which can appear authorized and evade signature-based checks unless the monitoring baseline is tuned to detect the specific proxy pattern.
- T1216.001partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and PubPrn abuse leaves observable traces in command-line logs and network traffic, but the control's scope is set by business requirements rather than mandating the depth of instrumentation needed to surface every signed-script proxy execution.
- T1216.002partialdetects — A.8.16's scope includes event logs and system activity monitoring that would surface the wscript.exe→SyncAppvPublishingServer.vbs→powershell.exe chain and the resulting PowerShell execution, but the clause sets scope rather than mandating instrumentation depth, so an implementation focused on network/application layers is fully conformant and sees nothing of this host-resident LOLBin technique.
- T1217partialdetects — A.8.16's scope includes access to systems and event logs, so file reads of browser databases are observable where endpoint file-access telemetry or EDR is in scope; but the clause sets scope by business requirements rather than mandating that depth, and an implementation focused on network/perimeter monitoring is fully conformant yet sees nothing of local browser data enumeration.
- T1218partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and not tampered with, and its anomalous-behaviour list explicitly names process injection, but system binary proxy execution uses signed, trusted binaries that blend into normal behaviour, so detection depends on the depth of behavioural baselining the implementer chooses rather than being guaranteed by the clause's mandatory scope.
- T1218.001partialdetects — A.8.16's monitoring scope covers process execution (hh.exe spawning and loading IE components), network traffic from embedded payloads, and security tool logs (AV signatures for known malicious CHMs), all of which can surface this technique; however, the clause sets scope rather than mandating instrumentation depth, and CHM's documented ability to bypass application control on unpatched systems means a conformant implementation focused on narrower layers may miss it.
- T1218.002partialdetects — A.8.16's monitoring scope covers process execution events, security tool logs (AV/IDS), and code-authorization checks, all of which can surface control.exe proxying malicious .cpl files; however, the clause determines scope by business and security requirements rather than mandating the specific endpoint telemetry depth needed, so an implementation monitoring only network layers is fully conformant and sees nothing of this in-process execution technique.
- T1218.005partialresponds — A.8.16 surfaces anomalous behaviour (e.g. process injection, unusual system behaviour, malware-associated activity) once the mshta.exe proxy execution is underway and hands it to 5.26 response procedures; the core of `responds` (containment/eradication) is engaged only where the chosen monitoring scope includes the relevant host telemetry, which the clause itself determines rather than mandates.
- T1218.008partialdetects — A.8.16's scope is set by business and security requirements rather than mandating specific instrumentation depth; odbcconf.exe abuse is observable where host-based process execution, command-line auditing, and DLL-load telemetry are in scope (the clause explicitly mentions 'event logs relating to system and network activity' and 'checking that the code being executed is authorized to run'), but an implementation monitoring only network traffic is fully conformant and sees nothing of this local LOLBin execution, so detection is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1218.010partialdetects — A.8.16's scope includes event logs, network traffic, security tool logs, and code-execution authorization checks — all of which can surface regsvr32.exe launching with suspicious arguments (e.g., /s /n /u /i:http) or generating outbound URL fetches — but the clause sets scope rather than mandating instrumentation depth, and the technique is explicitly designed to evade monitoring through allowlists of the signed Microsoft binary, so a fully conformant implementation that does not specifically tune for this pattern can miss it.
- T1218.010partialresponds — A.8.16 surfaces anomalous behaviour (process injection, unusual system behaviour, malware-associated activity, unauthorized execution) in real time or near-real time and feeds it to the 5.26 response procedure; this engages the core of `responds` (containment/eradication once underway) for the in-process execution slice of T1218.010 but leaves the pre-execution proxying, COM scriptlet loading from URL, and persistence-via-hijacking aspects outside monitored observables or response scope.
- T1218.011partialresponds — A.8.16's real-time/periodic monitoring, anomaly detection (incl. process injection, unusual behaviour, malware patterns), alerting and procedures to respond to positive indicators directly engage an in-flight T1218.011 instance on monitored Windows systems, enabling containment/eradication once underway, but scope is set by org requirements so coverage of all possible rundll32 abuse is a chosen slice rather than bounded remainder.
- T1218.012partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but verclsid.exe is a signed native binary whose abuse may blend with legitimate shell-extension verification, so detection depends on whether the implementation's baseline and behavioural rules are tuned to catch it.
- T1218.013partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour the monitoring system should identify, and item (f) calls for checking that executed code is authorized — both directly relevant to Mavinject's DLL/import-descriptor injection. However, the clause sets scope by business and security requirements rather than mandating host-level instrumentation depth, so a conformant implementation monitoring only network and application layers sees nothing of an in-process injection, leaving a nameable remainder.
- T1218.013partialresponds — A.8.16 surfaces process injection (explicitly listed as anomalous behaviour) and other signs of mavinject.exe abuse via real-time/periodic monitoring of system behaviour, resource use, logs and security-tool output, then feeds alerts to dedicated trained personnel and procedures that respond per 5.26; partial because the clause sets scope by organisational requirements rather than mandating universal coverage of every possible mavinject variant or unmonitored processes.
- T1218.014partialdetects — A.8.16's monitoring scope includes process execution logs, security tool logs (EDR/AV), and event logs that can surface mmc.exe invocations with suspicious command-line arguments, .msc file loads, and CLSID registry modifications, but the clause sets scope by organizational requirements rather than mandating endpoint telemetry depth, so an implementation focused on network layers alone would miss this host-based proxy-execution technique.
- T1218.015partialdetects — A.8.16's monitoring scope includes process creation events, network traffic, resource use, and 'unusual system behaviour' (with process injection named as an example), so an implementation with process-level telemetry can surface Electron abuse such as chrome.exe spawning cmd.exe or anomalous child processes from legitimate apps; however, scope is set by business and security requirements rather than mandated, so a conformant implementation monitoring only network and application layers would miss the in-process abuse while still catching any C2 traffic, leaving a nameable slice undetected.
- T1219.002partialdetects — A.8.16's monitoring scope (network traffic, process execution, user/system behavior baselines) can surface remote desktop software activity and flag anomalous patterns such as unusual access times or unauthorized installations, but the technique specifically uses legitimate, often allow-listed tools, so distinguishing malicious C2 from authorized support use depends on behavioral baselining and scope choices the clause sets rather than mandates — an implementation monitoring only network flows without process or user-behavior telemetry is fully conformant and would miss much of the technique.
- T1219.003partialdetects — A.8.16's network traffic and access monitoring can surface KVM-over-IP traffic and connection patterns to remote access hardware, but the technique explicitly leverages legitimate tools allowed by peripheral device policy and can bypass software security solutions, so distinguishing malicious from authorized use against an established baseline is a slice rather than a bounded remainder of a working mechanism.
- T1220partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and XSL script processing via msxsl.exe or wmic leaves observable command-line and process-creation artifacts, but the control's scope is set by business requirements rather than mandating the host-level instrumentation depth needed to catch every invocation, so detection is a slice the implementer chooses.
- T1221partialdetects — A.8.16's network and system monitoring scope covers the outbound HTTP/SMB request to fetch the remote template and the subsequent payload execution, both of which produce observable events (anomalous outbound traffic, process activity, potential forced-authentication attempts); the bounded remainder is template URLs pointing to compromised-but-trusted domains and payloads that execute entirely within legitimate process behavior.
- T1222.001partialdetects — A.8.16's scope includes event logs, process execution, and unusual system behaviour, which can surface DACL modifications via icacls/cacls/takeown/attrib and the resulting Windows Security Event Log entries (4663, 4670) — but the clause sets scope by business requirements rather than mandating the specific audit policies and command-line logging needed, so a conformant network-only or application-layer implementation sees nothing of this host-resident technique.
- T1222.002partialdetects — A.8.16's scope includes event logs, security tool logs, and configuration file monitoring, which can surface chmod/chown calls via auditd, EDR, or similar host telemetry, but the clause sets scope by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network traffic is fully conformant and sees nothing of local permission changes.
- T1480partialdetects — A.8.16's monitoring scope (network/system/application traffic, event logs, configuration files, resource use) can capture some artifacts of guardrail checks — unusual file accesses, registry queries, network share enumerations, or API call sequences that deviate from baseline — but execution guardrails are specifically designed to mimic normal system behavior (checking AD domains, network shares, user-agents are routine operations), so distinguishing malicious guardrail logic from legitimate system queries requires depth of instrumentation the clause does not mandate, only enables through its scope-setting language.
- T1480.001partialdetects — A.8.16 monitoring can observe the post-execution behavior of the payload if environmental keying succeeds (anomalous process activity, resource use, deviations in standard protocols), and may surface some pre-execution artifacts (system queries for target-specific values, cryptographic operations), but the technique is specifically designed to evade detection by making checks look like normal system activity and avoiding transmission of decryption material over monitored channels, so the keying mechanism itself is largely invisible to the monitoring scope.
- T1480.002partialdetects — A.8.16's scope is set by business and security requirements rather than mandating host-level process instrumentation, so an implementation monitoring only network/application layers is fully conformant and sees nothing of mutex creation; where the scope includes security-tool logs (AV/EDR signatures for known malware mutexes) and event logs covering process/handle activity, the technique is detectable, but the clause does not guarantee that scope.
- T1482partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but domain trust discovery via API calls or LDAP queries is not a named anomalous behaviour and its detection depends on whether the organization's baseline and scope include host-level telemetry for such reconnaissance.
- T1484partialdetects — A.8.16 explicitly lists monitoring for unauthorized access to systems and unusual system behaviour, which covers the permission changes and configuration modifications of T1484, but the technique's guidance notes adversaries can revert changes to remove indicators, leaving a window where detection depends on logging granularity and retention rather than being assured.
- T1484.001partialdetects — A.8.16's scope includes critical/admin configuration files (GPOs are configuration files in SYSVOL), event logs (Windows event IDs 5136/5137 capture directory service changes), and security tool logs that could surface GPO modifications, but the clause sets scope by business requirement rather than mandating AD/file-integrity instrumentation, so a fully conformant implementation monitoring only network and application layers would miss this technique.
- T1484.002partialdetects — A.8.16 explicitly lists monitoring for unauthorized access to protected resources and unusual system behaviour, which can surface trust modifications, but the control's scope is determined by business requirements rather than mandating the depth of directory-object auditing needed to catch every trust change, so what is detected is a slice the implementer chose.
- T1485.001partialdetects — The PutBucketLifecycle API call generates cloud audit log entries (CloudTrail, Azure Activity Log) that fall under A.8.16(e) event logs, so a monitoring implementation that includes cloud control-plane telemetry can detect the lifecycle policy modification; however, A.8.16 sets scope by business and security requirements rather than mandating cloud audit log ingestion, and an implementation monitoring only traditional network/host telemetry is fully conformant yet sees nothing of this technique.
- T1486partialdetects — A.8.16 explicitly lists monitoring for activity associated with malware, unusual system behaviour, resource utilization spikes, and known attack characteristics — all of which ransomware encryption produces in abundance (mass file modifications, CPU/disk spikes, C2 traffic for key exchange, known signatures). The scope-setting nature of the clause prevents `full`: an implementation monitoring only network traffic is fully conformant and would see C2 and resource anomalies but miss the file-level encryption itself, mirroring the partial-grade logic on the T1055 anchor.
- T1491.002partialdetects — T1491.002 is defacement of systems external to the organization; A.8.16's monitoring scope is the organization's own estate (its traffic, systems, applications, logs), so there is no vantage point from which to observe the defacement event itself — analogous to the T1583.001 anchor where the technique happens somewhere the control cannot look.
- T1495partialdetects — A.8.16's scope includes code-integrity checking (item f) and unusual system behaviour, and observable precursors exist (firmware-flashing utilities, network traffic fetching firmware images, post-corruption device failures), but the actual write to flash memory typically occurs below the OS where standard monitoring has limited visibility, so detection covers a slice rather than the core event.
- T1496partialresponds — A.8.16 surfaces anomalous resource consumption (CPU, memory, bandwidth, bottlenecks, overloads) and unusual behaviour such as process injection or deviations from baseline, enabling the incident response procedure (5.26) to contain/eradicate an in-progress hijacking; partial because the clause sets scope by business requirements rather than mandating universal coverage of every hijacking vector (e.g. SMS spam or proxy selling may fall outside monitored systems).
- T1496.003partialdetects — A.8.16 monitoring can surface SMS pumping through volume anomalies in outbound API calls to the SMS provider, unusual web-form submission patterns against the established baseline, and resource/cost spikes — all within the clause's scope (traffic, event logs, resource use, unusual behavior). However, the actual SMS traffic occurs at the third-party SaaS provider, sophisticated attackers mimic legitimate OTP/verification patterns to stay under thresholds, and the clause sets scope by business requirements rather than mandating the specific instrumentation needed to catch this technique reliably.
- T1496.004partialdetects — A.8.16's scope covers outbound traffic, resource use (CPU, bandwidth, quotas), and event logs, which can surface symptoms of SaaS abuse such as anomalous API call volumes, unusual authentication patterns, and abnormal consumption — but the actual resource-intensive work runs on the SaaS provider's infrastructure, outside the organization's monitoring vantage point, so only a slice of the technique is observable.
- T1497partialdetects — A.8.16's scope includes security tool logs, event logs, and resource usage that can surface some indicators of VM-evasion behavior (sleep-timer resource patterns, process/registry checks for artifacts), but the technique is specifically designed to evade exactly these monitoring systems by checking for security tools and disengaging, and A.8.16 sets scope by business requirements rather than mandating the endpoint telemetry depth needed to comprehensively detect sandbox-evasion checks.
- T1497.001partialdetects — A.8.16 monitoring can detect some indicators of VM/sandbox evasion checks (WMI queries, registry queries, PowerShell execution, file enumeration) through event logs and security tool logs, and the clause explicitly lists 'unusual system behaviour' and 'activity typically associated with malware' as in-scope; however, the technique is specifically designed to blend in with normal system behavior, the checks are individually indistinguishable from legitimate administrative activity, and the clause sets scope rather than mandating the specific telemetry needed to correlate the pattern of system-information queries that characterizes the technique.
- T1497.002partialdetects — A.8.16's monitoring scope (set by business requirements) can surface some user-activity-check behaviors — unusual file/registry enumeration, deviations from baseline user behavior, idle/wait patterns — but the checks are designed to be stealthy passive reads of existing artifacts that blend with normal system activity, and a conformant implementation focused on network or application layers sees nothing of them.
- T1497.003partialdetects — T1497.003 runs on the organization's estate (API calls within a process), so it is in principle observable by A.8.16's monitoring scope, but the technique is subtle — time-based API calls leave no network or obvious system event traces unless endpoint/API-level instrumentation is in place, which A.8.16 does not mandate (scope is set by business requirements, as with the T1055 anchor). An implementation focused on network and event logs is fully conformant and sees nothing of these in-process calls.
- T1498partialprevents — A.8.16 mandates real-time/periodic monitoring of network traffic, known attack characteristics (explicitly including denial of service), anomalous behaviour, and baselines that can trigger alerts to stop or deter a Network DoS before full exhaustion occurs, but the control only sets scope by requirements rather than mandating universal detection or blocking mechanisms, leaving many volumetric DDoS cases (e.g., external saturation before internal monitoring) outside its reach.
- T1498.002partialresponds — A.8.16 surfaces anomalous traffic patterns, known attack characteristics (DoS, buffer overflows), and unusual network behaviour in real time, feeding alerts to the 5.26 response procedure for containment; partial because the control's scope is set by organisational requirements and many reflection amplifiers lie outside the monitored estate, so the event on the target may not be visible until impact is already realised.
- T1499partialresponds — A.8.16 surfaces anomalous resource consumption, process crashes, overloads, and known attack patterns (including DoS and process injection) in real time and feeds them to 5.26 response procedures, but its core mechanism is detection/monitoring rather than containment or eradication once the technique is underway, and several T1499 vectors (e.g. pure resource exhaustion without detectable signature) sit outside its configured baseline.
- T1499.001partialresponds — A.8.16 surfaces the anomalous resource exhaustion, overloads, and known attack patterns (DoS, protocol deviations) in real time via monitoring and alerting, feeding directly into 5.26 response procedures for containment/eradication once the flood is underway; partial because the clause sets scope by business needs rather than mandating universal coverage of all OS-exhaustion vectors or platforms, and core response actions (contain/eradicate) sit in the referenced 5.26 rather than in monitoring itself.
- T1499.002partialresponds — A.8.16 surfaces anomalous resource consumption, overloads, bottlenecks, and known attack patterns (including DoS floods) in real time and feeds them to 5.26 response procedures, but its core mechanism is detection/monitoring rather than containment or eradication once the flood is underway.
- T1499.003partialresponds — A.8.16 surfaces anomalous resource consumption, overloads, and application behaviour matching the flood (via baselines, real-time monitoring, and alerts on CPU/memory/bandwidth bottlenecks), handing the event to 5.26 response procedures for containment/eradication once underway; partial because the clause sets scope by organisational requirements rather than mandating universal coverage of all application-exhaustion vectors or guaranteeing eradication of every instance.
- T1499.004partialdetects — A.8.16 explicitly lists monitoring for known attack characteristics (DoS, buffer overflows), unplanned terminations, unusual system behaviour (including process injection), resource bottlenecks/overload, and deviations from baseline, which surfaces many application/system exploitation attempts in real time; however the clause's scope is set by organisational requirements rather than mandating coverage of every possible exploit vector or zero-day, leaving a genuine slice unseen.
- T1499.004partialresponds — A.8.16 surfaces anomalous behaviour (crashes, unplanned terminations, resource overloads, known attack patterns) once the exploitation is underway and feeds it to 5.26 response procedures, but its core mechanism is detection and alerting rather than containment/eradication actions that `responds` names on the event lane.
- T1505partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but its scope is set by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a server software component installation.
- T1505.001partialdetects — A.8.16's scope covers event logs, security tool logs, process creation, file integrity and access to critical applications — all of which surface SQL stored procedure abuse (xp_cmdshell process spawns, CLR assembly file changes, SQL Server audit logs, anomalous database traffic). However, the clause sets scope by business requirements rather than mandating database-specific monitoring, so an implementation monitoring only network and host layers is fully conformant and would miss the database-level creation/modification of stored procedures themselves.
- T1505.002partialdetects — Transport agent registration is observable through Exchange event logs, configuration-file monitoring, and process-execution telemetry — all within A.8.16's scope — but the clause sets scope by business requirements rather than mandating Exchange-specific instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an on-server persistence mechanism designed to sit dormant until adversary-defined criteria are met.
- T1505.004partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and IIS component installation leaves observable artifacts in web server configuration and module loading, but the control's scope is set by business requirements rather than mandating the host-level instrumentation depth needed to catch every persistence mechanism.
- T1505.005partialdetects — A.8.16 explicitly lists monitoring configuration files (covering the ServiceDll registry value) and checking that executed code has not been tampered with (directly relevant to termsrv.dll modification), so a conformant implementation with file-integrity and registry monitoring will detect this; however, the clause sets scope by business requirements rather than mandating those specific instrumentation depths, so an implementation monitoring only network traffic is fully conformant and sees nothing of the DLL swap.
- T1505.006partialdetects — A.8.16's scope is set by business and security requirements rather than mandating ESXi hypervisor monitoring, so an implementation that excludes hypervisors is fully conformant and sees nothing of a VIB installation; when ESXi is in scope, the technique is observable through multiple channels (VIB install events, configuration file changes, new firewall rules, new listeners, startup tasks), but the clause does not guarantee that scope.
- T1518partialdetects — A.8.16's scope includes event logs and unusual system behaviour, which can surface software-discovery commands (process creation, command-line auditing, registry/file enumeration) against a baseline of normal user activity; however, the clause sets scope by business and security requirements rather than mandating process-execution or command-line telemetry depth, so an implementation monitoring only network traffic is fully conformant and sees nothing of on-host enumeration, and cloud-API discovery (T1518.001) is reachable only if the organization has chosen to monitor those control-plane logs.
- T1518.001partialresponds — A.8.16's real-time/continuous monitoring, anomaly detection (incl. unauthorized access/scanning, process injection indicators, security tool logs), alerting, and procedures to respond to positive indicators (see 5.26) enable containment/eradication once discovery behavior is observed in flight; partial because the clause sets scope by org requirements (not mandating coverage of all discovery commands, cloud APIs, or non-monitored platforms) and detection itself is not response.
- T1518.002partialdetects — Backup software discovery via reg query, dir, tasklist, etc. generates process execution, command-line, and registry access events that fall within A.8.16's scope of 'event logs relating to system and network activity' and 'unusual system behaviour,' but the clause sets scope by organizational requirement rather than mandating process-level telemetry, so a conformant implementation focused on network and access monitoring would miss this local enumeration.
- T1525partialdetects — A.8.16's monitoring scope is set by business requirements and can include registry access logs, configuration file changes, and code execution authorization checks (clause f), which can surface image tampering or unauthorized registry access; however, the specific activity of implanting code into a container or cloud image is not in the mandated scope, so detection depends on implementation choices rather than the clause's mechanism.
- T1526partialdetects — A.8.16's monitoring scope includes access to systems and applications and unusual system behaviour, which can surface cloud service discovery API calls where host or network telemetry is in place, but the clause determines scope rather than mandating cloud API instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of the enumeration.
- T1528partialdetects — A.8.16's behavioral baselines and API access monitoring can surface anomalous token USE (unusual time, location, frequency, or resource access patterns), but the theft itself — reading a Kubernetes service account token from a container's filesystem, an IMDS request, or OAuth grant to a malicious app — often occurs without triggering monitored events, and cloud/SaaS API calls may bypass traditional network monitoring scope.
- T1530partialdetects — A.8.16's scope is set by business requirements and can include outbound API traffic to cloud storage, cloud-provider logs (CloudTrail, Azure Activity), and authentication events, so active access from corporate systems is observable; however, the clause does not mandate cloud-environment instrumentation, and access using stolen credentials from outside the corporate network, or exploitation of misconfigured public buckets, occurs outside the vantage point an on-premises-focused implementation would have.
- T1531partialdetects — Account modifications generate well-known event log entries (e.g., Windows 4720/4724/4725/4738, Group Policy changes, PowerShell 4104) that fall within A.8.16's explicit scope of 'event logs relating to system and network activity' and 'unusual user and system behaviour', but the clause sets scope by business and security requirements rather than mandating instrumentation depth, so an implementation monitoring only network traffic or application layers is fully conformant and sees nothing of in-account changes.
- T1534partialdetects — A.8.16's monitoring scope (network traffic, security tool logs, event logs, behavioral baselines) can surface indicators of internal spearphishing — unusual sending volume or recipient patterns from a compromised account, malicious links caught by web filters, attachments flagged by AV/IPS, deviations from per-user baselines — but the clause sets scope by organizational requirements rather than mandating email-content inspection or per-user communication baselining, and the technique is specifically designed to look legitimate because it originates from a trusted internal account on normal channels.
- T1535partialdetects — A.8.16's monitoring scope is determined by business and security requirements, and an organization that does not use a cloud region may not extend its monitoring baseline there, so the anomalous instance creation is observable only where the implementer has chosen to instrument that region.
- T1537partialdetects — A.8.16's scope includes access logs, event logs, resource utilization and unusual user/system behaviour, which can surface cloud API calls, anomalous data volumes and access patterns associated with T1537; however, the technique is specifically designed to evade traditional network monitoring by using internal cloud provider infrastructure and APIs, and the clause's scope is set by business requirements rather than mandating cloud-API telemetry, so a fully conformant implementation focused on network traffic would miss it.
- T1538partialdetects — A.8.16 monitors access to systems and applications, which can surface dashboard logins from unusual locations or times, but the control's scope is determined by business requirements and an implementation monitoring only network and application layers is fully conformant without instrumenting cloud dashboard GUI activity.
- T1539partialdetects — A.8.16's enumerated scope covers process behaviour (it explicitly names process injection as an example of unusual system behaviour), file access to critical applications, and network traffic deviations — all of which can surface cookie theft via browser memory access or AiTM proxying. However, the clause sets scope by business requirements rather than mandating instrumentation depth, and a conformant implementation monitoring only network and application layers sees nothing of a local cookie-store read by malware running with user privileges, leaving a real remainder.
- T1542partialdetects — A.8.16's host-based monitoring scope (event logs, security tool logs, OS-level resource use) largely cannot see firmware/BIOS/UEFI modification, which the technique description explicitly notes evades host software-based defenses; however, some peripheral manifestations are observable through the clause's network traffic monitoring (rogue PXE/DHCP activity), resource-use anomalies during boot, and configuration-file monitoring, and the clause's scope is set by business requirements so firmware integrity monitoring is admissible though not mandated.
- T1542.001partialdetects — A.8.16's scope includes code-integrity checking ('checking that the code being executed is authorized to run... and has not been tampered with') and behavioural baselines that could surface firmware compromise, but the clause's primary monitoring scope is OS-level and above (network traffic, system/application logs, access, resource use); firmware operates below the OS where most of these mechanisms have limited or no visibility, so detection is achievable only where an implementation extends integrity monitoring into firmware, leaving a substantial remainder.
- T1542.002partialdetects — A.8.16's monitoring scope (network traffic, system/application logs, security tool logs, resource use) can surface downstream indicators of compromised component firmware — anomalous C2 traffic, resource anomalies, configuration drift — and item (f) on code-authorization checks could in principle extend to firmware integrity, but the core technique executes below the OS where host-based telemetry is blind, so what is caught is a slice of secondary indicators rather than the modification itself.
- T1542.003partialdetects — A.8.16's scope includes monitoring critical system configuration files and checking that executing code has not been tampered with, which can surface boot-sector modifications (MBR/VBR/ESP) via file-integrity monitoring; however, the bootkit's actual execution happens below the OS before any OS-level monitoring is active, so the core runtime behaviour is outside the clause's vantage point regardless of implementation depth.
- T1542.004partialdetects — A.8.16's scope includes network equipment, configuration files, network traffic and resource usage, so the ROMMON image loading (e.g. TFTP transfer to a device, config changes, CPU/memory spikes during flash write) is within reach of a conformant monitoring implementation; however, the persistent firmware compromise itself operates below the OS and is explicitly designed to be difficult to detect, and standard monitoring tools listed in the clause (IDS, IPS, antivirus, firewalls) do not cover firmware integrity of network device ROMMON, so what is caught is a slice of the loading indicators rather than the technique's core effect.
- T1542.005partialdetects — A.8.16's scope includes networking equipment access (b), network and system configuration files (c), and inbound/outbound network traffic (a), so TFTP boot traffic to an unexpected server and configuration changes specifying a malicious TFTP server are observable artifacts within the clause's stated scope; however, detection depends on the organization having included network device configuration and TFTP traffic in its monitoring scope and baselined normal boot behaviour, which the clause requires but does not mandate at a specific depth.
- T1543.001partialdetects — A.8.16's scope includes configuration files (c), security tool logs (d), event logs (e), and code execution authorization (f), all of which can surface launch agent creation/modification on macOS — but the clause sets scope by business requirements rather than mandating file-integrity or EDR instrumentation, so a conformant implementation focused only on network traffic would see nothing of a .plist being written to ~/Library/LaunchAgents.
- T1543.005partialdetects — A.8.16's monitoring scope includes system and network activity, and its guidance explicitly lists 'process injection' and 'unusual system behaviour' as anomalies to identify, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a container service modification.
- T1546partialdetects — A.8.16's monitoring scope includes system and application event logs and unusual system behaviour, which can surface event-triggered execution, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a registry or scheduled-task modification.
- T1546.001partialdetects — A.8.16's scope includes configuration files, event logs, and security-tool logs that can surface registry modifications and the resulting process executions, but the clause sets scope by business and security requirements rather than mandating host-level registry or file-integrity monitoring; an implementation focused on network and application layers is fully conformant and would miss the file-association change itself.
- T1546.002partialdetects — A.8.16's scope includes configuration file changes (item c), event logs (item e), security tool logs (item d), and code execution authorization checks (item f), all of which can surface screensaver persistence via registry modifications to HKCU\Control Panel\Desktop, .scr file creation, and unauthorized process execution; however, the clause sets scope by business and security requirements rather than mandating instrumentation depth, so an implementation monitoring only network traffic is fully conformant and sees nothing of this user-space persistence technique.
- T1546.003partialdetects — WMI event subscription creation leaves observable artifacts (mofcomp.exe / PowerShell WMI cmdlet process creation, .mof file creation, WMI repository changes, WmiPrvSe.exe child processes, and WMI-Activity event log entries) that fall within A.8.16's enumerated monitoring items (process execution, event logs, configuration files, security tool logs). However, the clause sets scope rather than mandating instrumentation depth: an implementation monitoring only network and application layers is fully conformant with A.8.16 and sees none of these host-level artifacts, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1546.004partialdetects — Shell config file modifications fall within A.8.16's explicit scope (item c: critical/admin-level system configuration files; item f: code tampering checks; item d: unusual system behaviour), so monitoring can surface both the file write and the unexpected process execution at shell launch — but the clause sets scope by business requirements rather than mandating file-integrity monitoring on /etc/profile, ~/.bashrc and the rest, so a conformant implementation that omits that scope misses the technique entirely.
- T1546.005partialdetects — A.8.16 can detect T1546.005 through file integrity monitoring of shell profiles (.bashrc, .profile), process execution when the trap-triggered payload runs, and command-line auditing — but the trap registration itself is a shell builtin that does not create a separate process event, and detection of the profile modifications depends on whether file integrity monitoring falls within the scope set by business requirements, mirroring the A.8.16 vs T1055 `partial` reasoning.
- T1546.007partialdetects — A.8.16's monitoring scope includes checking that executing code is authorized and not tampered with, and its anomalous-behaviour list explicitly names process injection, but Netsh Helper DLL persistence loads a legitimate-signed Microsoft binary that executes a registered DLL during normal administrative or automated operations, so the technique blends into expected system behaviour and is observable only where the implementation's baseline and instrumentation depth reach registry and DLL-load telemetry, which the clause leaves to the implementer's discretion.
- T1546.008partialdetects — A.8.16's listed items (file/config integrity, security-tool logs, event logs, code-execution authorization) can surface binary replacement of sethc.exe/utilman.exe, registry modifications to accessibility feature pointers, and IFEO debugger value changes, but the clause sets scope by business and security requirements rather than mandating file-integrity or registry-change instrumentation, so what is caught is a slice the implementer chose.
- T1546.009partialdetects — A.8.16 explicitly lists process injection as an example of unusual system behaviour to monitor for, and AppCert DLLs are a form of DLL injection into processes, but the control's scope is determined by business requirements rather than mandating the host-level instrumentation depth needed to reliably catch every instance.
- T1546.010partialdetects — A.8.16's scope includes security tool logs, event logs, configuration file monitoring, and code execution authorization checks — all of which can surface AppInit_DLLs registry modifications and the resulting DLL loads — but the clause sets scope by organizational requirement rather than mandating host-level instrumentation depth, so a conformant network-focused implementation sees nothing of this in-process persistence mechanism.
- T1546.011partialdetects — A.8.16 explicitly lists process injection as an example of unusual system behaviour to monitor for, and application shimming is a form of process injection that is observable where host-level telemetry is in scope, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process shim.
- T1546.012partialdetects — A.8.16's scope explicitly includes monitoring of critical system configuration files (item c, which covers IFEO registry modifications) and unusual system behaviour including process injection (item d, directly analogous to IFEO injection), so the mechanism is observable in principle; however, the clause sets scope by business and security requirements rather than mandating instrumentation depth, and an implementation focused on network or application-layer monitoring is fully conformant yet sees nothing of registry-based IFEO changes, mirroring the T1055 anchor's reasoning.
- T1546.013partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and a modified PowerShell profile executing on launch is observable where host-level telemetry is in scope, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a profile script running inside a user session.
- T1546.014partialdetects — A.8.16's scope (configuration files, event logs, security tool logs, authorized-code checks) can in principle surface emond abuse via file integrity monitoring of /etc/emond.d/rules/, process execution telemetry from the emond daemon, and event log analysis, but the clause sets scope by business and security requirements rather than mandating these specific instrumentation points — an implementation monitoring only network and application layers is fully conformant and sees nothing of this macOS-specific persistence mechanism.
- T1546.015partialdetects — A.8.16's enumerated scope includes configuration files (c), event logs (e), security tool logs (d), and code execution authorization checks (f), all of which can surface COM Hijacking via Registry modifications and unexpected code execution; however, the clause sets scope by business requirements rather than mandating instrumentation depth, so a fully conformant implementation focused on network monitoring would miss this Registry-resident technique entirely.
- T1546.016partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but installer script execution is a legitimate system activity that blends with normal software installation, so detection depends on whether the monitoring scope includes host-level telemetry and behavioural baselines that can distinguish malicious post-install scripts from benign ones.
- T1546.017partialdetects — A.8.16's scope explicitly includes monitoring critical system configuration files (item c) and checking that executed code is authorized and untampered (item f), both of which directly cover udev rule file modifications; however, the clause sets scope by business and security requirements rather than mandating file-integrity monitoring on /etc/udev/rules.d/, so a conformant implementation focused on network and application layers would miss this technique entirely.
- T1546.018partialdetects — A.8.16's scope is set by business requirements and includes file/configuration monitoring and code execution authorization checks (item f), so Python startup hook persistence is observable in principle via file integrity monitoring on site-packages or process telemetry at interpreter startup; however, an implementation focused on network and application layers is fully conformant and would miss the file writes and module loads entirely, leaving the technique detectable only where the implementer chose to instrument Python paths.
- T1547.001partialdetects — A.8.16's scope explicitly includes configuration files, security tool logs (antivirus/IDS/IPS/EDR), event logs, and authorization-of-executed-code checks — all of which can surface registry run key and startup folder modifications (e.g., Sysmon Event ID 13, EDR alerts, file integrity monitoring). However, the clause sets scope by business and security requirements rather than mandating endpoint/registry instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this persistence mechanism — the same shape the T1055 anchor teaches.
- T1547.002partialdetects — A.8.16 explicitly lists monitoring critical configuration files (item c, which covers registry), checking that executed code is authorized and untampered (item f), and detecting unusual system behaviour such as process injection (item d) — all of which surface authentication package DLL loading. However, the clause sets scope by business and security requirements rather than mandating host-based instrumentation depth, so a conformant implementation focused on network or application layers would not see this boot-time LSA activity, leaving a nameable remainder.
- T1547.003partialdetects — A.8.16 explicitly covers monitoring of configuration files (c), code execution authorization (f), event logs (e), and network traffic (a) — all of which surface T1547.003's artifacts: new subkeys under W32Time\TimeProviders, DLL loads by the time service at boot, and W32Time event log entries. The grade is 'mostly' rather than 'full' because A.8.16 sets scope by business requirement rather than mandating specific instrumentation, so a conformant implementation focused on network-layer monitoring could miss the registry and code-execution signals.
- T1547.004partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and Winlogon helper DLL modifications produce observable registry changes and process loads, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1547.005partialdetects — A.8.16 explicitly includes monitoring of 'critical or admin level system and network configuration files' (item c) and event logs (item e), which covers the LSA Security Packages registry modifications central to T1547.005; however, the clause sets scope rather than mandating instrumentation depth, so a conformant implementation monitoring only network traffic or application layers would miss the registry change and the DLL load into LSA, leaving the technique observable only when file-integrity or registry auditing is actually deployed.
- T1547.006partialdetects — A.8.16's scope explicitly covers event logs, configuration files, and unusual system behaviour, which can surface kernel module loading (kernel log entries, new files in module directories, system call traces), but the technique is specifically designed to evade detection — LKM rootkits run at Ring 0, tamper with logs, and hide their presence from user-space tooling — and the clause sets scope by business requirements rather than mandating kernel-level instrumentation depth, so a fully conformant implementation monitoring only network and application layers sees nothing of this technique.
- T1547.007partialdetects — A.8.16 covers monitoring of configuration files and unusual system behaviour, which can surface the plist modification and the unexpected application launch at login, but the clause sets scope by business and security requirements rather than mandating file-integrity monitoring of user-level preference directories, so a conformant implementation focused on critical/admin configuration files would miss this technique.
- T1547.008partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and an LSASS driver load is observable where host instrumentation is in scope, but the clause determines scope by business requirements rather than mandating kernel-level telemetry, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1547.009partialdetects — A.8.16's scope includes event logs, configuration file changes, and unusual system behaviour, which can surface shortcut creation/modification in startup folders via file integrity monitoring or endpoint telemetry, but the clause sets scope by business requirements rather than mandating the specific instrumentation needed; an implementation monitoring only network and access layers is fully conformant and sees nothing of this file-system persistence technique.
- T1547.010partialdetects — A.8.16's scope includes configuration file monitoring (item c), event logs (item e), code execution authorization checks (item f), and baseline-deviation detection (item i), all of which can surface a port monitor DLL registration or load — but the clause sets scope by business requirements rather than mandating registry or DLL-load instrumentation, so a conformant implementation focused on network traffic alone sees nothing of this on-host persistence technique.
- T1547.013partialdetects — A.8.16 explicitly covers monitoring of configuration files (item c) and checking that executed code is authorized and untampered (item f), both directly relevant to detecting modified .desktop autostart entries; however, the clause sets scope by organizational requirement rather than mandating file-integrity or login-process instrumentation, so an implementation focused on network and Windows-endpoint telemetry is fully conformant and sees nothing of this Linux persistence technique.
- T1547.014partialdetects — A.8.16 explicitly enumerates monitoring of critical/admin configuration files and event logs, both of which surface Active Setup registry modifications and StubPath execution; however, scope is set by business and security requirements rather than mandated, so an implementation monitoring only network traffic is conformant and blind to this technique.
- T1547.015partialdetects — Login items leave observable artifacts (the backgrounditems.btm file, launchd entries, process executions at login, and event log entries) that fall within A.8.16's enumerated monitoring scope — configuration files (c), event logs (e), code execution authorization checks (f), and unusual system behaviour (i) — but the clause sets scope by business and security requirements rather than mandating the specific host-level instrumentation (file integrity monitoring on macOS login item locations, launchd telemetry, login-time process execution logging) needed to reliably surface this technique, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1548partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but T1548's abuse of built-in mechanisms often occurs within legitimate processes and authorized contexts, making it detectable only where the organization's chosen monitoring scope includes deep host-level telemetry and behavioural baselines.
- T1548.001partialdetects — A.8.16's scope includes file/configuration monitoring (c), code execution authorization checks (f), and unusual system behaviour (d), which can surface setuid bit changes, new setuid binaries, and execution of unexpected privileged processes — but the clause sets scope by business requirements rather than mandating host-level file-integrity or privilege-context instrumentation, so a conformant implementation focused on network and access logs sees nothing of this local privilege-escalation technique.
- T1548.002partialdetects — A.8.16 explicitly lists process injection and known attack characteristics as examples of what monitoring should surface, and UAC bypasses frequently involve process injection into trusted auto-elevated binaries; however, the clause sets scope by business requirements rather than mandating host-level instrumentation depth, so a network-and-application-only implementation is fully conformant and sees nothing of an in-process UAC bypass.
- T1548.003partialdetects — A.8.16 can surface sudo caching and sudoers abuse through host event logs, critical configuration file changes, and anomalous privileged behavior when those are included in the monitoring scope, but the clause leaves that scope to business and security requirements, so it does not guarantee coverage.
- T1548.005partialdetects — A.8.16's scope includes access events and system/network activity logs, which can encompass cloud audit logs (CloudTrail, Azure Activity Log, GCP Audit Logs) showing JIT requests, role assumption, and PassRole usage; however, the clause sets scope by organizational requirement rather than mandating cloud telemetry, and the technique is explicitly noted as stealthy when 'logs do not clarify when role impersonation is taking place,' so detection is a slice the implementer chooses rather than a bounded remainder of a working mechanism.
- T1548.006partialdetects — A.8.16's scope is set by business and security requirements rather than mandating specific instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of TCC database manipulation; a comprehensive implementation that includes file integrity monitoring on system databases, process execution telemetry, and macOS event logs can surface the technique, but the clause does not require that scope.
- T1550partialdetects — A.8.16 explicitly lists monitoring for unauthorized access attempts to protected resources and unusual system behaviour, which can surface the use of stolen tokens or tickets, but the control's scope is set by business requirements rather than mandating the deep session-level instrumentation needed to catch all pass-the-hash or token replay events.
- T1550.001partialdetects — A.8.16's scope includes monitoring access to systems/applications, event logs, and deviations from baseline user behavior, which can surface anomalous API call patterns, unusual token usage, or access from unexpected locations — but the technique is explicitly designed to look legitimate (the token IS authorized), A.8.16 sets scope by business requirements rather than mandating cloud/SaaS API token monitoring depth, and the technique's own description notes API-channel abuse is difficult to detect even from the provider end, leaving a substantial slice unobserved.
- T1550.002partialdetects — A.8.16's monitoring scope includes network traffic and system event logs, which can surface lateral movement using NTLM hashes, but the technique's core authentication step occurs within the local security authority and is not directly observable without deep host instrumentation that the clause does not mandate.
- T1550.003partialdetects — A.8.16's monitoring of event logs, network traffic, and baseline-based anomaly detection can surface Kerberos ticket anomalies (unusual source/time, deviations in protocol use, unusual user behaviour), but detection depends on the implementation's scope — set by business and security requirements rather than mandated by the clause — and valid Kerberos tickets can blend with normal authentication traffic without Kerberos-specific monitoring rules configured.
- T1550.004partialdetects — A.8.16's network and access monitoring sees the session on the estate and its baseline-based anomaly detection (unusual time, location, frequency, user-agent) can flag some cookie-replay accesses, but the technique is specifically designed to present as a legitimate authenticated session, so detection depends on the depth of behavioural analytics configured rather than being inherent to the clause's mechanism.
- T1552partialdetects — A.8.16's monitoring scope (file access, event logs, security tool logs, anomalous behaviour) can surface credential dumping tool signatures, access to credential stores, and exfiltration of harvested credentials, but the technique's core mechanism is reading existing files — a normal operation that requires specific file-access auditing or credential-store monitoring to detect, and A.8.16 sets scope by business requirements rather than mandating instrumentation of credential repositories.
- T1552.001partialdetects — A.8.16's scope can include file access logs, process execution monitoring, and anomaly detection against baselines that could surface credential-file searching (e.g., unusual file enumeration, known tools, deviations from normal access patterns), but the clause sets scope by business requirements rather than mandating the file-access or command-auditing depth needed, and the technique often resembles legitimate file reads by an authenticated user, so detection is a slice of well-instrumented implementations rather than a bounded remainder of a mechanism that mostly works.
- T1552.002partialdetects — A.8.16's scope includes event logs and access to systems, so host-based telemetry (process creation with command line, registry auditing) can surface `reg query` activity, but the clause sets scope by business requirement rather than mandating instrumentation depth — a fully conformant implementation focused on network traffic or application-layer monitoring sees nothing of an on-host registry query, and the technique blends with legitimate admin activity without behavioral baselining.
- T1552.003partialdetects — A.8.16's scope includes event logs and access monitoring, which can capture file access events if auditd/file access auditing is configured and collected — but the clause does not mandate this, and an implementation monitoring only network traffic and standard system logs (fully conformant with A.8.16) sees nothing of a local file read of shell history; detection depends on scope choices the clause leaves to the implementer.
- T1552.004partialdetects — A.8.16's monitoring scope includes access to systems, event logs, and DLP/security tool logs that can surface file access to key directories, DLP alerts on credential file access, and network device CLI command logging (e.g., `crypto pki export`), but the clause sets scope by business requirements rather than mandating file-system or DLP instrumentation depth, so a conformant implementation focused on network and application layers would miss an adversary simply reading key files on a compromised host.
- T1552.005partialdetects — A.8.16's network traffic monitoring and security tool logs (IDS/IPS, web filters) can surface access to 169.254.169.254, including SSRF patterns from web servers, but the clause sets scope by organizational requirement rather than mandating metadata-API-specific monitoring, and legitimate applications query this endpoint routinely, so the signal is buried without dedicated rules.
- T1552.006partialdetects — A.8.16's monitoring scope includes access to systems and servers (b), event logs (e), and unusual user behaviour against a baseline (i), all of which could surface SYSVOL enumeration, GPP XML file reads, or known decryption-tool execution on a domain controller. However, the technique is stealthy by design — any domain user has legitimate read access to SYSVOL, so detection depends on the monitoring implementation specifically covering file-access auditing on domain controllers and tuning baselines to flag this pattern, which A.8.16 sets as scope rather than mandates.
- T1552.007partialdetects — Container API calls (Docker daemon, Kubernetes API server) generate event logs and access records that fall within A.8.16's scope of monitoring access to systems and event logs, but the clause sets scope by business and security requirements rather than mandating container-infrastructure telemetry, so an implementation focused on traditional systems can be fully conformant and still miss credential retrieval via these APIs.
- T1552.008partialdetects — A.8.16 can surface some manifestations — DLP tools (item d) can flag credential patterns in chat traffic, access monitoring (item b) can catch anomalous admin-portal or server-side access to chat platforms, and event-log correlation can reveal compromised integration tools making unusual API calls — but the technique's core acts (direct endpoint access to chat applications, normal-appearing reads of message history, and activity within SaaS platforms largely outside the organization's monitoring estate) are not visible to the control's network/system scope, and what is caught depends on what the organization has chosen to instrument rather than on a mechanism that mostly works.
- T1553partialdetects — A.8.16's scope includes security tool logs, configuration file monitoring, event logs, and item (f) explicitly covers checking that executed code is authorized and untampered — all surfaces where trust subversion (certificate store changes, code signing events, trust provider modifications, security policy edits) can leave artifacts. The grade is partial because the clause sets scope by business and security requirements rather than mandating instrumentation of these specific trust surfaces, so a conformant implementation focused on network and application layers sees little of T1553.
- T1553.001partialdetects — A.8.16 item (f) explicitly covers checking that executed code is authorized and untampered, and items (d) and (i) cover security-tool logs and unusual system behaviour — all directly relevant to Gatekeeper bypass — but the clause sets scope rather than mandating macOS-specific file-attribute, code-signing and bundle-integrity instrumentation, so a conformant implementation focused on network and application layers sees nothing of this technique.
- T1553.002partialdetects — A.8.16 explicitly lists monitoring for process injection, unauthorized code execution, tampered binaries, known attack patterns, and deviations from baseline (including code that is not authorized or has been altered), which surfaces adversary use of stolen or forged signing materials when the signed malware runs or exhibits anomalous behavior; however, purely pre-execution acquisition/creation of signing material on external infrastructure or fully stealthy validly-signed benign-looking binaries fall outside the listed scopes and baselines.
- T1553.003partialdetects — A.8.16 explicitly includes monitoring of critical system configuration files (item c), security tool logs (item d), and event logs (item e), all of which can surface the Registry modifications, DLL drops, and DLL loads that constitute SIP/trust provider hijacking; however, the clause sets scope by business and security requirements rather than mandating Registry or file-integrity instrumentation, so a conformant implementation focused only on network traffic would miss the technique entirely.
- T1553.004partialdetects — A.8.16's scope includes monitoring of critical system configuration files (c) and event logs (e), and root certificate installation modifies the certificate store and generates command-execution and event-log artifacts on Windows, macOS and Linux — but the clause sets scope by business requirements rather than mandating certificate-store or file-integrity monitoring, so a conformant implementation focused on network and application layers can miss the installation entirely.
- T1553.005partialdetects — A.8.16's monitoring scope can observe the precursor chain — downloads of container files from external sources (network traffic), mounting of .iso/.vhd images (system events), and execution of files from mounted containers (process monitoring, security tool logs) — but the specific MOTW bypass mechanism itself is not directly detectable as anomalous behaviour, since the technique abuses legitimate Windows functionality (container files not inheriting NTFS alternate data streams) rather than performing a clearly anomalous action, and an implementation focused on network/application layers would miss the file-system-level abuse.
- T1553.006partialdetects — A.8.16 explicitly lists process injection and unauthorized access to configuration files as monitoring targets, but code signing policy modification via kernel memory manipulation or debug mode reboots occurs outside the monitoring scope the clause mandates, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1555partialdetects — A.8.16's monitoring scope explicitly includes access to systems, security tool logs (AV/IDS/EDR), event logs, and unusual system behaviour — all of which can surface credential-store access and credential-dumping tool execution. However, the clause sets scope by business requirements rather than mandating instrumentation depth, so an implementation monitoring only network traffic or only application-layer events is fully conformant and sees nothing of a local credential-store read; what is caught is a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1555.001partialdetects — A.8.16's scope includes event logs relating to system activity and unusual system behaviour, so `security dump-keychain` execution and keychain file access on a monitored macOS endpoint are observable in principle; however, the clause sets scope by business and security requirements rather than mandating host-level instrumentation depth, so an implementation focused on network or Windows-centric monitoring is fully conformant and sees nothing of this macOS-specific credential access.
- T1555.002partialdetects — A.8.16's scope includes system and process behaviour monitoring (e.g. 'unusual system behaviour', security tool logs, resource use), and a host-telemetry implementation can surface memory access to securityd by an unexpected process. But the clause sets scope rather than mandating instrumentation depth: an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-memory credential read, and a root-level adversary can often suppress or evade the very logs the control relies on.
- T1555.003partialdetects — A.8.16's monitoring scope (file access, process behaviour, event logs, security-tool logs) can surface credential-store reads, unusual API calls like CryptUnprotectData, and process-memory scraping, but the clause sets scope by business requirements rather than mandating instrumentation depth, so a conformant implementation focused on network/application layers would miss the in-process activity, mirroring the T1055 anchor.
- T1555.004partialdetects — A.8.16's enumerated scope includes process execution logs, file access events, security tool logs (AV/EDR), and unusual system behavior — all of which can surface vaultcmd.exe or rundll32.exe keymgr.dll execution, access to credential store paths, and known credential-dumping tools. But the clause sets scope by business and security requirements rather than mandating instrumentation depth, so an organization fully conformant with A.8.16 that monitors only network and application layers sees nothing of this in-process/file-system technique, matching the same partial shape as A.8.16 vs T1055.
- T1555.005partialdetects — A.8.16's monitoring scope includes process behaviour, file access, event logs and resource use, which can surface password-manager credential extraction (unauthorised process reading the database file, brute-force authentication failures, anomalous memory/CPU use), but the core extraction is in-memory and a conformant A.8.16 implementation monitoring only network and application layers sees nothing of it — the same scope-determined remainder that put A.8.16 vs T1055 at partial.
- T1555.006partialdetects — A.8.16's scope includes outbound network traffic and behavioural baselines against which unusual user/system activity is flagged; calls to secrets-manager APIs (e.g. `get-secret-value`) are observable as cloud control-plane traffic and as deviations from established secret-access patterns, so monitoring configured to cover cloud API activity can surface them. The grade is partial because the clause sets scope by business requirement rather than mandating cloud audit-log ingestion or secrets-manager telemetry, and many conformant implementations monitor only on-premises systems and would see nothing of API calls made from a compromised workload using valid credentials.
- T1556partialdetects — A.8.16's monitoring scope includes system and application behaviour, and its guidance explicitly lists 'process injection' and 'unauthorized access' as detectable anomalies, but modification of the authentication process itself (e.g., tampering with LSASS or PAM) is a deep system-level change that may not be surfaced by the network, access, or resource-use monitoring the clause primarily describes, leaving detection dependent on the implementer's chosen scope rather than guaranteed by the control.
- T1556.002partialdetects — A.8.16 lists monitoring of critical configuration files (c), code execution authorization (f), and security tool logs (d), all of which can surface a malicious password filter DLL via registry changes to Notification Packages, new DLLs in System32, and LSASS loading events; however, the clause sets scope by organizational requirement rather than mandating host-level instrumentation depth, so a network-only implementation is fully conformant and sees nothing of this host-resident technique.
- T1556.003partialdetects — A.8.16's enumerated scope directly covers PAM tampering: (c) critical configuration files includes /etc/pam.d/, (f) code-tamper checks explicitly addresses recompiled libraries like pam_unix.so, and (e)+(h) plus baseline monitoring cover authentication anomalies from a backdoored PAM. A faithful implementation detects most aspects, but the clause sets scope by organizational requirements rather than mandating file-integrity or code-tamper instrumentation depth, so an implementation focused only on network traffic is conformant and blind to this technique.
- T1556.004partialdetects — A.8.16's scope includes access to networking equipment and critical configuration files, so file-integrity or config-change monitoring on network devices could surface the patched image, but the implanted backdoor authenticates as a normal successful login and the technique is designed to be invisible during operation, so the authentication half is not caught.
- T1556.005partialdetects — A.8.16's enumerated scope covers event logs, configuration files, and security tool logs, which can surface AD property changes (Event ID 4738/4739), PowerShell Script Block logging for Set-ADUser, and LSA secret access; however, the clause sets scope by business requirements rather than mandating AD audit or PowerShell logging, and the offline decryption step is inherently unobservable, leaving a bounded remainder.
- T1556.006partialdetects — A.8.16's scope explicitly covers critical configuration files (hosts file modifications), authentication-related event logs, and access to systems — all surfaces where MFA tampering leaves artifacts. However, the technique often abuses legitimate admin features (e.g., Conditional Access exclusions, new MFA method registration) that look like valid actions, and A.8.16 sets scope by business requirements rather than mandating MFA-specific instrumentation, so detection depends on what the implementation chose to monitor.
- T1556.007partialdetects — A.8.16 explicitly enumerates monitoring of critical configuration files (AD FS config), code execution authorization checks (DLL injection into authentication services), and process injection — all directly relevant to T1556.007's mechanics — but the clause sets scope by business requirement rather than mandating instrumentation depth, so an implementation focused on general network/application layers is fully conformant and may miss authentication-infrastructure tampering.
- T1556.008partialdetects — A.8.16's scope includes monitoring configuration files (Registry changes to network provider keys), checking that executed code is authorized (item f), and detecting unusual system behavior — all relevant surfaces for this technique. However, the malicious DLL is loaded by the legitimate mpnotify.exe process during normal logon, making it stealthy by design; detection requires specific monitoring of network provider registrations and DLL load events during authentication, which A.8.16 does not explicitly mandate.
- T1556.009partialdetects — A.8.16's scope includes critical/admin configuration files and event logs, and modifications to conditional access policies in Entra ID, Okta, AWS IAM, etc. generate audit log entries that fall within that scope — but the clause sets scope by business requirements rather than mandating identity-provider audit ingestion, so detection depends on whether those logs are in the monitored set.
- T1557.001partialresponds — A.8.16's real-time/periodic monitoring of network traffic, anomalous behaviour (including known attack patterns and unauthorized access), security-tool logs, and alerting feeds directly into 5.26 incident response once the poisoning/relay is underway on the monitored network; it does not contain/eradicate the adversary-controlled system or relayed session itself.
- T1557.002partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but ARP cache poisoning operates at the link layer where detection depends on whether the implementer's scope includes network-level traffic analysis, which the clause leaves to business and security requirements rather than mandating.
- T1557.003partialdetects — A.8.16's monitoring scope includes network traffic and anomalous behaviour patterns, and DHCP spoofing produces observable network artifacts (rogue OFFER/ACK messages, unusual DNS server assignments), but the clause determines scope by business requirements rather than mandating the specific DHCP-layer instrumentation needed to catch every spoofing attempt, so what is detected is a slice the implementer chose.
- T1557.003partialresponds — A.8.16's real-time/periodic monitoring of network traffic, anomalous behaviour (incl. known attack patterns and unauthorized access), baseline deviations, and alert/response procedures can surface and feed into incident response for an in-progress DHCP spoofing AiTM or exhaustion attack on the monitored network, but the clause itself performs no containment/eradication and many stealthy or pre-compromise DHCP exchanges fall outside its configurable scope.
- T1557.004partialdetects — A.8.16's network traffic and access monitoring can surface downstream indicators of an evil twin attack (anomalous connections, certificate failures, unauthorized access attempts, traffic to known-bad infrastructure) once victim devices interact with organization resources, but the clause does not mandate wireless/RF monitoring needed to detect the rogue AP itself, and scope is set by business requirements rather than the control.
- T1558partialdetects — A.8.16's monitoring scope includes network traffic and event logs, which can surface Kerberos ticket anomalies like Golden Ticket usage, but the control's scope is determined by business requirements rather than mandating the deep host-level instrumentation needed to detect all ticket theft or forgery.
- T1558.001partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but Golden Ticket abuse uses legitimate Kerberos protocol flows that blend with normal authentication traffic, so detection depends on whether the implementation's scope and baseline include the specific TGT anomalies (e.g. anomalous PAC fields, lifetime, or RC4 usage) rather than being guaranteed by the clause's general mandate.
- T1558.002partialdetects — A.8.16's scope includes event logs, deviations in standard protocols, and unauthorized access attempts — all of which can surface Silver Ticket usage via service-side Kerberos event logs (e.g., 4769) and ticket-characteristic anomalies — but the technique is specifically designed to evade detection by bypassing the KDC, and the clause sets scope by business requirements rather than mandating the deep Kerberos-specific analysis needed to reliably catch forged service tickets.
- T1558.005partialdetects — A.8.16's scope includes event logs relating to system activity and monitoring for unusual system behaviour, so an implementation with host-based file access auditing or process execution telemetry can surface reads of ccache files in /tmp or anomalous use of klist/kinit/Kekeo; however, the clause sets scope by business requirements rather than mandating file-integrity or process-level instrumentation, so a network-focused implementation is fully conformant and sees nothing of this local credential-theft technique.
- T1559partialdetects — A.8.16's scope is set by business and security requirements rather than mandating host-level process telemetry, so a conformant implementation monitoring only network and application layers sees little of local IPC abuse (pipes, COM, DDE between local processes); however, the clause explicitly lists 'unusual system behaviour' and 'deviations in use of standard protocols' as monitoring targets, and network-mediated IPC (e.g. DCOM) plus security-tool logs can surface a slice of the technique.
- T1559.002partialdetects — A.8.16's scope is set by business and security requirements rather than mandating process-level or command-line telemetry, so DDE execution is observable only where the implementation includes process creation logs, parent-child process relationships, and behavioural baselines against 'deviations in use of standard protocols' — a slice the implementer chose, not a bounded remainder of a mechanism that mostly works.
- T1559.003partialdetects — A.8.16's scope is set by business and security requirements rather than mandating instrumentation depth, and XPC service abuse is a macOS-specific technique requiring endpoint telemetry (process execution, code-authorization checks per item f, unusual system behaviour per item d) that an organization may or may not include in its monitoring scope; an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process XPC abuse.
- T1560partialdetects — A.8.16's monitoring scope includes network traffic and system behaviour, which can surface the use of compression/encryption utilities or anomalous data flows prior to exfiltration, but the technique's obfuscation purpose and the control's scope being set by business requirements mean detection is a slice the implementer chooses, not a bounded remainder of a mechanism that mostly works.
- T1560.001partialdetects — A.8.16's monitoring scope includes system and network activity logs and checking that executed code is authorized, which can surface the use of archiving utilities, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of a local archive operation.
- T1560.002partialdetects — A.8.16's scope includes resource use (CPU/disk spikes from compression), event logs, and code-authorization checks that can surface library-based archival, but the clause sets scope by business requirements rather than mandating host-level instrumentation depth — an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process archive operation, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1560.003partialdetects — A.8.16's scope includes resource utilization (CPU, disk, memory), process execution logs, and behavioral baselines that could surface custom archiving activity, but the technique is specifically designed to evade signature-based detection (which A.8.16 explicitly relies on), and the clause's scope is set by business requirements rather than mandating host-based instrumentation depth, so an implementation monitoring only network layers sees nothing of the archiving itself.
- T1561partialdetects — A.8.16 explicitly lists resource monitoring (including hard disks), system access monitoring, event logs, and configuration file monitoring within its scope, all of which would surface disk wipe activity through anomalous I/O patterns, direct disk device access, and configuration changes on network devices; however, the clause sets scope by business and security requirements rather than mandating instrumentation depth, so a network-only implementation is fully conformant and sees nothing of the disk-level activity, leaving a nameable slice rather than a bounded remainder of a mechanism that mostly works.
- T1563partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but session hijacking operates within an already-authenticated session and may not produce a distinct anomalous event unless the monitoring scope includes session-level telemetry, which the clause leaves to business requirements rather than mandating.
- T1563.001partialdetects — A.8.16's scope explicitly covers monitoring network traffic, access patterns, deviations in use of standard protocols, and unusual user/system behaviour against an established baseline — all of which can surface SSH hijacking through anomalous session activity or unexpected lateral movement. However, the technique is designed to evade detection by reusing a legitimate authenticated session over an encrypted tunnel, so what is caught depends on baseline quality and behavioral analysis depth rather than being inherent to the control's mechanism.
- T1564partialdetects — A.8.16's baseline-and-anomaly monitoring can surface some hidden-artifact activity (unusual resource use, deviations from baseline behaviour, unauthorized access patterns, configuration-file integrity), but T1564 is defined as evasion of detection, and many sub-techniques (ADS, hidden directories, virtualization-isolated regions) are designed to be invisible to standard monitoring; scope is set by business requirements, so a conformant implementation may catch only a slice.
- T1564.001partialdetects — A.8.16's scope includes security tool logs (antivirus, IDS, IPS) and event logs that, when configured to monitor file system activity, can surface hidden files and directories; however, the clause sets scope by business requirements rather than mandating file system or EDR monitoring, and the technique is specifically designed to evade standard detection, so a conformant implementation focused on network/access monitoring would miss it.
- T1564.002partialdetects — A.8.16 can surface the account creation and registry/plist/configuration changes that create hidden users where host event and configuration logging are in scope, but because it leaves monitoring depth to business requirements, an implementation without that depth remains conformant and misses it.
- T1564.003partialdetects — A.8.16's scope includes event logs, process monitoring, and configuration-file changes, which can surface hidden-window indicators (PowerShell -WindowStyle Hidden command lines, suspicious parent-child process trees, WindowPosition registry edits, CreateDesktop API calls); but the clause sets scope by business and security requirements rather than mandating host-level telemetry, so a conformant network-focused implementation sees nothing of this UI-concealment technique, mirroring the A.8.16/T1055 partial reading.
- T1564.004partialdetects — A.8.16's monitoring scope includes event logs, security tool logs, and system activity that can surface NTFS file attribute abuse (ADS enumeration, MFT analysis, file integrity monitoring), but the clause sets scope by business requirements rather than mandating host-level file system instrumentation depth — an implementation monitoring only network and application layers is fully conformant and sees nothing of on-disk attribute hiding, so what is caught is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1564.005partialdetects — A.8.16's monitoring scope includes resource use (disk space, I/O patterns), event logs, configuration files, and 'unusual system behaviour' — all of which can surface side effects of a hidden file system (anomalous disk consumption, reserved-space modification, unusual partition activity, fragmented file metadata). However, the technique is explicitly designed to evade security tools by abstracting malicious storage outside the standard file system view, and A.8.16 sets scope by business requirements rather than mandating the deep host telemetry needed to reliably catch a well-implemented hidden FS, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1564.006partialdetects — A.8.16's scope includes resource use (CPU/memory/disk), network traffic, configuration files and unusual system behaviour, which can surface side effects of a hidden VM (anomalous resource consumption, unattributable network traffic, new VM config files, unexpected processes), but the technique is explicitly designed to evade host-based monitoring by hiding artifacts inside the virtual instance where the listed monitoring tools cannot see them.
- T1564.007partialdetects — A.8.16's monitoring scope (process execution, file access, security tool logs, code-authorization checks) can surface VBA stomping through execution behavior, file-structure anomalies (source replaced with zeros while p-code remains), and post-execution network or process activity, but the technique is specifically engineered to evade the source-code and signature-based scanning that dominates A.8.16's typical implementation, and scope is set by business requirements rather than mandating p-code analysis.
- T1564.008partialdetects — A.8.16's scope (set by business/security requirements) covers event logs, security tool logs, and unusual user/system behaviour, so creation of inbox rules via PowerShell cmdlets or Exchange admin actions is observable where mailbox audit logs, PowerShell script-block logging, and Exchange management logs are inside that scope; however, the clause determines scope rather than mandating instrumentation depth, and the technique is specifically designed to hide the very alerts a defender would rely on, so an implementation monitoring only network traffic is fully conformant and sees nothing of the rule creation.
- T1564.009partialdetects — A.8.16's scope includes event logs, code execution authorization checks, and unusual system behaviour monitoring, which can surface resource fork abuse (e.g., extended attribute changes, unauthorized code execution), but the clause sets scope by business requirements rather than mandating macOS file-system or extended-attribute instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1564.010partialdetects — A.8.16's scope includes event logs, security-tool logs, and monitoring for unusual system behaviour (with process injection explicitly named), so a sufficiently instrumented implementation can surface PEB manipulation through WriteProcessMemory telemetry, memory-integrity checks, or behavioural correlation between logged and actual arguments; however, the clause sets scope by business requirement rather than mandating host-memory depth, and the technique is specifically engineered to defeat process-creation monitoring by overwriting the PEB after the benign arguments are captured, so a network- or application-layer implementation fully conformant with A.8.16 sees nothing of the in-process spoofing.
- T1564.011partialdetects — A.8.16's scope includes monitoring for unusual system behaviour and deviations in standard protocol use, and T1564.011 leaves observable artifacts (command-line arguments like `nohup` or `-ErrorAction SilentlyContinue`, processes surviving parent termination) that host-level monitoring can surface; however, the clause sets scope by business requirements rather than mandating command-line auditing or process-tree instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this evasion technique.
- T1564.012partialdetects — A.8.16's scope includes security tool logs (antivirus), configuration files (where AV exclusions live), and event logs that could surface file creation in excluded paths, so the technique is observable where the implementation chooses that scope; but the clause sets scope by business and security requirements rather than mandating file-system or AV-exclusion telemetry, so a fully conformant A.8.16 implementation focused on network and application layers would see nothing of files dropped into trusted exclusions.
- T1564.013partialdetects — The technique occurs on the organization's own Linux systems, so there is a vantage point: the `mount` syscall is auditable, sudo use is logged, and EDR/host-intrusion tools can surface bind-mount abuse — all within A.8.16's scope of event logs and security-tool logs. But the clause sets scope by business and security requirements rather than mandating the host-level audit depth or kernel telemetry needed, so an implementation monitoring only network and application layers is fully conformant and sees nothing of the bind mount, leaving a nameable slice undetected.
- T1564.014partialdetects — A.8.16's monitoring scope is set by business and security requirements rather than mandating xattr inspection; the technique is specifically designed to evade standard file-integrity and content-based detection because the file hash is unchanged and xattrs are invisible to standard tools. Detection is achievable in implementations that include process execution logging, command-line auditing, or behavioural monitoring of decode-and-execute patterns, but a fully conformant A.8.16 implementation focused on network and application layers would see nothing of the xattr payload itself.
- T1565partialdetects — A.8.16 can surface data manipulation through access logs to critical applications, event logs, and behavioural baselines (unusual user/system behaviour), but the clause sets scope by business and security requirements rather than mandating instrumentation depth — an implementation focused on network and perimeter monitoring is fully conformant and sees nothing of in-database or in-application data tampering, and the technique is explicitly designed to hide activity within normal patterns.
- T1565.001partialdetects — A.8.16's scope includes access monitoring, configuration-file monitoring, event logs and behavioural baselines that can surface stored-data manipulation (e.g. file-integrity changes, unusual access patterns, database transaction anomalies), but the clause sets scope by business requirements rather than mandating file-integrity or database-activity monitoring, so a conformant implementation focused on network and access logs sees little of this technique, and manipulations by actors with legitimate access may not deviate from the baseline.
- T1565.002partialdetects — A.8.16's scope includes network traffic, event logs, and deviations in standard protocol use, which can surface some transmitted-data manipulation (e.g., MITM-style interceptions, protocol anomalies), but the clause sets scope by business and security requirements rather than mandating deep packet inspection or host-level integrity checks, so inter-process manipulation and subtle in-transit alterations remain outside what a conformant implementation necessarily sees.
- T1565.003partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered (item f), and its anomalous-behaviour list explicitly names process injection (item d), but runtime data manipulation that alters in-memory values without changing the binary or injecting a process falls outside those specific checks, leaving a nameable gap.
- T1566partialdetects — A.8.16 monitors network traffic, email gateway logs, and unusual user behaviour, which can surface phishing indicators, but the control's scope is determined by business requirements and does not mandate the deep email inspection or sandboxing needed to catch every phish.
- T1566.001partialdetects — A.8.16's monitoring scope includes network traffic, security tool logs, and event logs, which can surface the delivery and execution of a malicious attachment, but the control's scope is determined by business requirements rather than mandating the specific email and endpoint instrumentation depth needed to catch every instance.
- T1566.002partialdetects — A.8.16 monitors network traffic and security tool logs, which can surface the delivery and click of a malicious link, but the control's scope is set by business requirements rather than mandating full email and web proxy instrumentation, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1566.003partialdetects — A.8.16 can detect downstream artifacts of T1566.003 — outbound traffic to malicious destinations, web filter/IDS hits on known-bad links, antivirus alerts on the delivered payload, and anomalous process behaviour after execution — but the spearphishing message itself arrives through third-party services (social media, personal webmail) that sit outside the enterprise monitoring scope, so the social engineering delivery step is invisible to the control.
- T1567.001partialdetects — A.8.16 monitors outbound network traffic and can surface exfiltration to a code repository via HTTPS, but the technique's use of a popular service already accessed by hosts provides cover that reduces detection to a slice rather than the bulk.
- T1567.004partialdetects — A.8.16's monitoring scope explicitly includes outbound network traffic and behavioural baselines (time, frequency, volume of access), so webhook exfiltration from the estate is within its vantage point — unlike T1583.001 which happens on a registrar's infrastructure. However, the technique is specifically designed to blend in with legitimate HTTPS traffic to common SaaS services, and the clause sets scope by business requirements rather than mandating deep content inspection, so what is caught is a slice (volume/timing anomalies, DLP signatures) rather than the bulk of webhook-based exfiltration.
- T1567.004partialresponds — A.8.16 surfaces anomalous outbound traffic, webhook patterns, or deviations from baseline (e.g. unusual HTTPS posts to SaaS endpoints) once underway and feeds them to 5.26 response procedures, but its core mechanism is detection/monitoring rather than containment or eradication of the exfiltration event itself.
- T1568partialdetects — A.8.16's monitoring scope includes outbound traffic and DNS server access, which can surface the resolution event itself, but the technique's dynamic nature and the control's scope being set by business requirements mean detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1568.001partialdetects — A.8.16's scope includes outbound/inbound network traffic and event logs, which encompass DNS queries/responses, and the clause explicitly calls out detecting known attack characteristics and unusual behaviour — both of which Fast Flux DNS exhibits (rapid IP churn, short TTLs). However, the clause sets scope rather than mandating DNS-specific analysis: an implementation monitoring only network flow records or application-layer logs is fully conformant and would miss the DNS pattern analysis Fast Flux detection requires, so the grade is partial rather than mostly.
- T1568.002partialdetects — A.8.16 explicitly lists monitoring for anomalous behaviour, malware-associated traffic, known malicious domains, process injection, and deviations in protocols, which surfaces DGA-generated C2 traffic or fallback channel activity when it crosses monitored networks, logs, or baselines; however the technique's domain generation occurs on adversary infrastructure and many DGA domains are never contacted, leaving a large slice unseen.
- T1568.003partialdetects — A.8.16's network and event-log monitoring scope covers the DNS queries, responses and subsequent outbound connections that DNS Calculation produces, and unusual port usage against returned IPs is a behavioural anomaly the clause explicitly contemplates; however, the calculation itself is in-memory and the technique is purpose-built to make resulting traffic resemble legitimate DNS or normal port use, so detection depends on the depth of behavioural baselining and signature coverage an implementation chooses rather than being guaranteed by the clause's scope.
- T1569.001partialdetects — A.8.16's scope includes event logs, security tool logs, and checks for unauthorized code execution, all of which can surface launchctl abuse on monitored macOS endpoints; however, the clause sets monitoring scope per business requirements rather than mandating host-level or macOS-specific instrumentation, so an implementation focused on network or Windows telemetry is fully conformant and sees nothing of this technique.
- T1569.002partialresponds — A.8.16 surfaces anomalous service-related activity (e.g. process injection, unauthorized access, unusual system behaviour, resource anomalies) in real time or near-real time and feeds it to the 5.26 response procedure for containment/eradication once the technique is underway; partial because the clause sets scope by business requirements rather than mandating universal coverage of every service execution vector or remote sc.exe/PsExec use.
- T1569.003partialdetects — Systemctl abuse leaves observable traces in audit logs, the systemd journal, process execution events, and service-file changes — all within A.8.16's permitted scope (items b, c, e, g) — but the clause sets scope by business requirement rather than mandating host-level instrumentation depth, so a network-only or application-layer implementation is fully conformant and sees nothing of the technique.
- T1572partialdetects — A.8.16 explicitly lists 'deviations in use of standard protocols' and resource-use anomalies (bandwidth, CPU) as monitoring targets, both of which can surface tunneling, but the technique is specifically designed to blend in with legitimate traffic (SSH, HTTPS, DNS) and the clause sets scope by organizational requirement rather than mandating deep packet inspection or behavioral depth, so a fully conformant implementation may see only the outer protocol and miss the encapsulation.
- T1572partialresponds — A.8.16's real-time/periodic monitoring of network traffic, anomalous behaviour, known attack patterns, and alerting feeds directly into 5.26 response procedures once tunneling is underway, but the control's scope is set by organisational requirements rather than mandating coverage of all tunneling variants (e.g. encrypted SSH, DoH) and its mechanism stops at detection/alerting without performing containment or eradication itself.
- T1573partialdetects — A.8.16 monitors network traffic and can detect the use of encryption on non-standard ports or with anomalous patterns, but it cannot inspect the content of an encrypted channel to confirm it is C2, so detection is limited to traffic metadata and behavioral anomalies.
- T1573.001partialdetects — A.8.16 can detect anomalous traffic patterns or connections to known malicious domains, but symmetric encryption conceals the content of command and control traffic, leaving the detection of the encrypted channel itself as a partial capability dependent on behavioral analysis rather than content inspection.
- T1573.002partialdetects — A.8.16's network monitoring can surface some instances of asymmetric crypto C2 through traffic-pattern anomalies, known-malicious destinations, and deviations from standard protocols, but properly implemented asymmetric encryption over standard protocols (e.g., TLS to a legitimate-looking domain) is designed to be indistinguishable from legitimate traffic at the content level, so the encryption itself defeats content inspection and a meaningful share of instances go unseen.
- T1574partialdetects — A.8.16 explicitly lists 'process injection' as an example of unusual system behaviour to monitor for, and item (f) requires checking that executed code is authorized and untampered — both directly relevant to execution-flow hijacking. Configuration-file and event-log monitoring catch many sub-techniques (DLL search order, path interception, registry permissions), but the clause sets scope by business and security requirements rather than mandating host-level instrumentation depth, so an implementation focused on network/application layers is fully conformant and misses in-process hijacks like COR_PROFILER or KernelCallbackTable.
- T1574.001partialdetects — A.8.16's enumerated scope includes security-tool logs (AV/EDR), event logs, and code-authorization checks that can surface DLL sideloading and search-order hijacking, but the clause sets scope by business requirements rather than mandating host-level instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of the DLL-load events, file-system artifacts, or process-behavior anomalies this technique produces.
- T1574.001partialresponds — A.8.16 surfaces anomalous behaviour (incl. process injection and deviations) in real time or near-real time and feeds alerts to the 5.26 response procedure, but the control itself performs none of the core respond activities (contain/eradicate) and the technique's in-process mechanics often leave no observable artefact at the network/application layers the clause explicitly scopes
- T1574.005partialdetects — A.8.16's clause (f) explicitly covers checking that executed code is authorized and has not been tampered with, and file-system events fall under (c)/(e), so the technique is observable where host and file-integrity telemetry are inside the monitoring scope; the grade is partial because the clause sets scope by business and security requirements rather than mandating file-integrity monitoring on installer directories, and an implementation focused on network or application-layer monitoring is fully conformant yet sees nothing of a binary replacement in %TEMP%.
- T1574.006partialdetects — A.8.16's scope includes event logs, configuration file monitoring, code authorization checks, and unusual system behaviour detection — all of which can surface LD_PRELOAD/DYLD_INSERT_LIBRARIES abuse through file integrity on library directories, process telemetry showing unexpected library loads, and environment-variable anomalies. But the clause sets scope by business requirement rather than mandating the specific instrumentation (library-load telemetry, file integrity on /usr/lib, env-var auditing) needed to catch this technique, and the technique's own design — hooking execve, readdir and similar functions to scrub artifacts — actively defeats the behavioural and log-based views A.8.16 relies on, leaving a conformant network-and-access-only implementation blind to it.
- T1574.007partialdetects — A.8.16's scope includes event logs, configuration file monitoring, and 'checking that the code being executed is authorized to run' — all of which can surface PATH interception via process execution telemetry with full paths, file integrity monitoring on PATH directories, and environment-variable change detection. But the clause sets scope by business and security requirements rather than mandating that depth of host-based instrumentation, so a conformant implementation focused on network or application layers would miss the in-host execution and configuration changes this technique produces.
- T1574.009partialdetects — A.8.16's scope covers file system events, process execution, and configuration changes that would surface an unquoted-path hijack (unexpected executable in a parent directory, service path modification, unauthorized binary launch), and item (f) explicitly calls for checking that executed code is authorized — but the clause sets scope by business requirements rather than mandating file-integrity or path-aware process telemetry, so a conformant implementation focused on network and access logs would miss the placement and resolution events that reveal this technique.
- T1574.010partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and a service binary replacement is observable through file integrity monitoring, unexpected process lineage, or anomalous binary execution, but the control's scope is set by business requirements rather than mandating the depth of host instrumentation that would surface every instance.
- T1574.011partialdetects — A.8.16's scope includes monitoring configuration files (item c) and event logs (item e), which would surface registry modifications to service keys and resulting DLL loads, but the clause determines scope by business requirements rather than mandating host-level registry instrumentation — an implementation monitoring only network and application layers is fully conformant and sees nothing of this technique.
- T1574.012partialdetects — A.8.16's scope includes configuration files (registry entries for COR_PROFILER), security tool logs (EDR detecting DLL loads in CLR processes), event logs, and explicitly lists 'process injection' as an example of unusual system behaviour to monitor — all of which surface this technique. But the clause sets scope rather than mandating instrumentation depth, so a network-only or application-layer implementation is fully conformant and sees nothing of the host-level DLL load or registry modification, leaving a nameable slice undetected.
- T1574.013partialdetects — A.8.16 explicitly names 'process injection' and 'unusual system behaviour' as items the monitoring system should identify, and T1574.013 is a form of process injection (the technique text links to T1055), so the clause covers it where the implementation has the right scope and depth; the same partial reasoning as A.8.16 vs T1055 applies — the clause sets scope rather than mandating instrumentation depth, and an implementation monitoring only network/application layers is fully conformant and sees nothing of this in-process PEB manipulation, while a deeper host-telemetry implementation catches a slice.
- T1574.013partialresponds — A.8.16 surfaces anomalous behaviour such as process injection and unusual system behaviour in real time or near-real time, enabling the 5.26 response procedure once the KernelCallbackTable hijack (a stealthy form of execution-flow tampering) is underway; the remainder is that the technique can complete and self-restore before detection occurs, and the clause's scope is set by organisational requirements rather than mandating universal coverage of every in-process callback-table mutation.
- T1574.014partialdetects — A.8.16 explicitly names 'process injection' as an example of unusual system behaviour to monitor, and AppDomainManager injection is a variant of process injection, so the technique is in scope; however, the clause sets monitoring scope rather than mandating instrumentation depth, so an implementation monitoring only network/application layers is fully conformant yet sees nothing of in-process .NET runtime hijacking, leaving detection dependent on the depth chosen by the implementer.
- T1578partialdetects — A.8.16's monitoring scope includes access to systems and event logs, which can surface the creation, deletion, or modification of cloud compute instances, but the control's scope is determined by business requirements rather than mandating cloud API instrumentation, so detection is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1578.001partialdetects — Cloud snapshot creation occurs within the organization's own cloud account and is captured in cloud provider audit logs (CloudTrail, Activity Log, etc.), which fall under A.8.16's 'event logs relating to system and network activity' and 'access to systems' scope; however, A.8.16 sets scope by business and security requirements rather than mandating cloud control-plane monitoring, so an implementation focused on traditional infrastructure is fully conformant and sees nothing of this technique.
- T1578.002partialdetects — Cloud instance creation generates API calls and audit log entries (CloudTrail/Activity Log equivalents) that fall within A.8.16's scope of 'event logs relating to system and network activity' and 'unusual system behaviour,' but the clause sets scope by business requirements rather than mandating cloud control-plane monitoring, so an implementation monitoring only traditional infrastructure is fully conformant and sees nothing of this technique.
- T1578.003partialdetects — Cloud instance deletion is observable through cloud control-plane audit logs (API calls, resource changes) which can fall within A.8.16's scope of 'event logs relating to system and network activity' and 'access to systems,' but the clause sets scope by business requirements rather than mandating cloud-specific instrumentation, so an implementation monitoring only on-premises or network-layer telemetry is fully conformant and sees nothing of the deletion; additionally, the technique is designed to remove forensic artifacts of prior malicious activity, so even when the deletion event is detected, the evidence it was meant to destroy is already gone.
- T1578.004partialdetects — A.8.16's scope includes access to systems, event logs, and security tool logs, so cloud API calls to revert snapshots and management console access are observable if cloud audit logging is within the monitoring scope set by business requirements; however, ephemeral storage resets occur as a side effect of VM stop/restart without explicit events, and the clause determines scope rather than mandating cloud telemetry depth, leaving a bounded remainder.
- T1578.005partialdetects — Cloud control plane modifications generate audit logs (CloudTrail, Azure Activity Log) that fall within A.8.16's scope of 'event logs relating to system and network activity' and 'critical or admin level system and network configuration files,' so a monitoring implementation that includes cloud audit telemetry and baselines normal configuration changes can detect anomalous quota/policy modifications; however, A.8.16 sets scope by business requirements rather than mandating cloud control plane instrumentation depth, and the technique is specifically designed to look like legitimate administrative activity, so a fully conformant A.8.16 implementation monitoring only traditional network and host telemetry would miss it.
- T1580partialdetects — Cloud infrastructure discovery via API calls (DescribeInstances, ListBuckets, etc.) generates event logs and network traffic that fall within A.8.16's scope (items a, b, e), and anomalous enumeration patterns are detectable against the baseline the clause requires. However, scope is set by business and security requirements rather than mandated by the clause, so an implementation monitoring only traditional on-premises estate is fully conformant and sees nothing of cloud API activity — the same 'slice the implementer chose' shape as the T1055 anchor.
- T1589partialdetects — A.8.16 can detect the active-scanning component of T1589 — probing authentication services leaves logs of access attempts that fall within the clause's monitoring scope (unauthorized scanning, successful/unsuccessful access attempts to protected resources) — but the bulk of the technique (OSINT via social media, public websites, phishing for information) occurs entirely off the organization's estate where A.8.16 has no vantage point, so only a slice is observable.
- T1590.002partialdetects — A.8.16 can detect a slice of this technique — specifically DNS zone transfer (AXFR) attempts and unusual DNS query patterns that touch the organization's DNS server and appear in event logs — but the bulk of the technique (passive DNS queries, public database lookups, third-party SaaS enumeration) occurs entirely off the organization's estate, where A.8.16 has no vantage point.
- T1590.004partialdetects — A.8.16's monitoring scope explicitly includes detecting unauthorized scanning of business applications, systems and networks, which catches the active-scanning component of T1590.004, but the technique is also executed through passive OSINT (public databases, search engines, victim-owned websites) that occurs entirely off the monitored estate and leaves no signal A.8.16 can see.
- T1590.006partialdetects — A.8.16's monitoring scope explicitly includes detecting 'unauthorized scanning of business applications, systems and networks,' and active scanning of network security appliances does touch the estate and generate observable traffic. However, T1590.006 also encompasses passive reconnaissance via public data sets and phishing for information, which occur outside the organization's monitored estate and are not visible to A.8.16's instrumentation.
- T1592partialdetects — Some sub-techniques of T1592 touch the estate and are observable through A.8.16's monitoring scope — active scanning of hosts leaves network traces, phishing-for-information attempts appear in inbound traffic, and User-Agent headers sent to adversary servers are visible in outbound traffic logs — but the bulk of the technique (gathering from social media, open databases, search of victim-owned websites, and compromised third-party sites) happens entirely off the organization's estate where A.8.16 has no vantage point, so only a slice is detected.
- T1592.002partialdetects — A.8.16 detects the vectors of T1592.002 that traverse the estate — active scanning of host software (network traffic, banners, ports), phishing-for-information messages, and outbound connections to adversary-controlled collection sites — but cannot detect the vectors that occur entirely off the estate (adversary reading public job postings, network maps, or assessment reports) or that involve remote reading of metadata from victim-hosted files, so the technique is only partially observable from the monitoring vantage point.
- T1592.004partialdetects — A.8.16 can detect a slice of T1592.004 — outbound traffic from the organization's users to adversary-compromised sites designed to harvest client configurations is within the monitoring scope (item a), and phishing emails sent to gather this information surface in security tool logs (item d). However, the actual gathering on adversary infrastructure and the public-data-sets vector (job postings, resumes, network maps) occur entirely outside the organization's estate, so the technique is only partially visible.
- T1594partialdetects — T1594 touches the organization's estate (their public web servers), so unlike T1583.001 there IS a vantage point: web server access logs fall under A.8.16's 'event logs relating to system and network activity' and 'inbound network traffic,' and the clause explicitly lists monitoring for 'unauthorized scanning of business applications, systems and networks.' However, scope is set by business and security requirements rather than mandated, so a conformant implementation monitoring only network perimeter or security-tool logs may see nothing of someone browsing a public site; detection of reconnaissance patterns (wordlist scanning, hidden directory enumeration) requires web server log analysis to be in scope and baselined, which is a slice the implementer chooses rather than a bounded remainder of a mechanism that mostly works.
- T1595partialdetects — A.8.16 explicitly lists monitoring of inbound/outbound traffic, unauthorized scanning of networks/systems/applications, and anomalous behaviour patterns that surface active scanning (especially when it triggers known attack characteristics or deviates from baseline), but the control's scope is set by the organization and limited to its own estate — it cannot observe pre-compromise scanning of third-party or unrelated infrastructure that the technique also permits.
- T1595.001partialresponds — A.8.16 surfaces the scan via network traffic, anomalous patterns or unauthorized scanning alerts (items a, d, g, h, i in guidance) once it touches monitored assets, feeding real-time alerts and response procedures per 5.26; partial because pre-compromise external scans on unowned IP space often generate no organizational event or artifact to contain/eradicate, leaving a large slice untouched per the event-lane anchor for this exact pair.
- T1595.003partialdetects — A.8.16's monitoring scope includes network traffic and anomalous behaviour patterns, and wordlist scanning generates observable HTTP requests against the organisation's own web servers, but the control's scope is determined by business requirements rather than mandating universal instrumentation, so detection is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1598partialdetects — A.8.16's scope explicitly includes inbound traffic and security tool logs (email security, IDS, DLP), giving it vantage on phishing messages arriving and on outbound credential transmission, but the core event — a user being socially engineered into divulging information — is not directly observable by automated monitoring, and a conformant implementation focused on network/application layers may miss user-side deception.
- T1598.002partialdetects — A.8.16's scope explicitly covers inbound network traffic and security tool logs including web filters and antivirus, which can surface spearphishing attachments at the email perimeter; however, the reconnaissance nature of T1598.002 means the information elicitation itself (the target being tricked into divulging data) is not reliably observable, and scope is set by business requirements rather than mandating email/attachment instrumentation depth.
- T1598.003partialdetects — A.8.16's scope includes inbound network traffic and security tool logs (web filters, IDS/IPS, firewalls) that can surface spearphishing links arriving at the organization's perimeter, but the clause sets scope by business requirements rather than mandating email/web instrumentation depth, so a fully conformant implementation monitoring only certain layers may miss the technique.
- T1600partialdetects — A.8.16 explicitly names networking equipment, configuration files, code-authorization checks (directly relevant to T1601 system image modification referenced by T1600), and 'deviations in use of standard protocols' as anomalous behaviour to monitor — all of which can surface weakened encryption on network devices. However, the clause sets monitoring scope by business and security requirements rather than mandating depth on network devices, so an implementation focused on servers and applications is fully conformant and sees nothing of this technique, leaving a nameable slice rather than a bounded remainder of a working mechanism.
- T1600.001partialdetects — The technique occurs on the organization's own network devices, so configuration-file monitoring (item c) and event logs (item e) could surface changes to encryption parameters, and traffic monitoring (item a) could in principle detect weakened cipher negotiations; however, A.8.16 sets scope by business requirements rather than mandating the depth of device-config or cryptographic-handshake inspection needed to reliably catch a subtle key-size reduction, so what is detected is a slice the implementer chose rather than a bounded remainder of a mechanism that mostly works.
- T1600.002partialdetects — The technique occurs on the organization's own network devices, which A.8.16 explicitly includes in monitoring scope (networking equipment, configuration files, resource use); disabling hardware encryption would manifest as configuration changes and CPU spikes from software encryption. However, A.8.16 sets scope by business requirements rather than mandating instrumentation depth, so an implementation monitoring only traffic and access logs — fully conformant with the clause — would miss this in-device change.
- T1601partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered (item f), which can surface a modified system image, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of the modification.
- T1601.001partialdetects — A.8.16 explicitly includes 'networking equipment' in its monitoring scope and item (f) calls out checking that code has not been tampered with, which directly addresses stored-image modification via TFTP/FTP/SCP transfers and file-integrity monitoring; however, scope is set by business and security requirements rather than mandated, and in-memory patching via debug commands or ROMMONkit manipulation leaves little observable surface for the monitoring mechanisms listed, so what is caught is a slice rather than the technique's full footprint.
- T1601.002partialdetects — A.8.16 explicitly covers networking equipment (b), critical configuration files (c), and code-tampering checks (f), all of which are directly relevant to detecting a system image downgrade on a network device; however, the clause sets scope by business and security requirements rather than mandating instrumentation depth on network devices, so an organization fully conformant with A.8.16 may monitor only traffic and not file integrity or boot configuration on embedded devices, leaving the file replacement and reboot unobserved.
- T1602partialdetects — A.8.16's scope explicitly includes access to networking equipment, critical applications, and configuration files, and network traffic to/from management systems is observable; however, the clause sets scope by business requirements rather than mandating instrumentation of configuration repositories specifically, and adversaries using legitimate-looking queries (SNMP, config management protocols) may not deviate from baseline without specific tuning, so detection is a slice of conformant implementations rather than a bounded remainder of a working mechanism.
- T1602.002partialdetects — A.8.16 explicitly requires monitoring of network traffic, access to networking equipment, critical configuration files, security-tool logs, event logs, resource use, and anomalous behaviour including unauthorized access/scanning and known attack patterns; this surfaces the technique of dumping network device configs via management protocols or file access on the monitored estate, with the bounded remainder being pre-compromise external reconnaissance that never touches organizational assets.
- T1606partialdetects — A.8.16's monitoring scope can include the identity-provider APIs, authentication events and access logs involved in credential generation, so unusual patterns in token issuance or API calls (e.g. AWS STS AssumeRole, Zimbra preauth) are observable where telemetry is in scope; however, the forging is designed to produce credentials indistinguishable from legitimate ones, much of the activity occurs on adversary infrastructure or through legitimate-looking API calls, and the clause sets scope rather than mandating depth, leaving a bounded remainder that is not detected.
- T1606.001partialdetects — A.8.16's behavioral baseline and anomaly monitoring (unusual time/location/frequency of access, deviations from expected user behaviour) can surface some instances of forged-cookie use, but the forging itself occurs on adversary infrastructure outside the control's scope, and a technically valid cookie that mimics the legitimate user's pattern defeats behavioural detection — so what is caught is a slice, not the bulk.
- T1606.002partialdetects — A.8.16's behavioral baseline monitoring (unusual time/location/frequency of access, deviations from expected user behaviour) can surface the USE of a forged SAML token when the adversary's downstream activity diverges from the legitimate user's pattern, but the forgery itself occurs at the IdP and produces a cryptographically valid authentication event indistinguishable from a legitimate one, and an adversary who mimics the legitimate user's behaviour evades the baseline entirely.
- T1609partialdetects — A.8.16's monitoring scope includes access to systems and unusual system behaviour, which can surface anomalous container administration commands, but the clause determines scope rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-container command execution.
- T1610partialdetects — Container deployment is a discrete event that generates artifacts across multiple monitoring domains explicitly listed in A.8.16: API calls to Docker/Kubernetes (network traffic), access to orchestration platforms (access logs), container creation events (system/event logs), and resource utilization changes (performance monitoring) — all standard monitoring targets similar to the T1110.001 anchor; the bounded remainder is organizations whose monitoring scope excludes container infrastructure or deployments executed through compromised legitimate credentials that blend with normal activity.
- T1611partialdetects — Container escape is observable through host-level telemetry (system calls, file mounts, privilege escalation, kernel module loads, security tool logs) that A.8.16 enumerates, but the clause sets scope by business requirements rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-container escape — the same shape as the T1055 anchor.
- T1612partialdetects — A.8.16's monitoring scope covers event logs, network traffic, resource use, and code-execution authorization — all of which surface artifacts of a Docker API build request (daemon logs, C2 download traffic, CPU/disk consumption, file-system writes). However, the technique is explicitly designed to evade defenses by using vanilla base images and building locally, and A.8.16 sets scope by business requirements rather than mandating container-aware instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-host image build.
- T1613partialdetects — A.8.16's scope is set by business and security requirements rather than mandating container-specific instrumentation; API calls to Docker/Kubernetes and dashboard access generate logs and traffic that monitoring catches where container infrastructure is in scope, but a conformant implementation monitoring only traditional estate would miss this technique entirely.
- T1614partialdetects — A.8.16's network traffic monitoring can surface the network-based forms of T1614 (queries to cloud instance metadata services, outbound calls to geolocation IP-lookup services) as anomalous connections, but the in-process discovery methods — GetLocaleInfoW, reading timezone/keyboard layout/language settings — generate no network traffic and no loggable event under A.8.16's scope, which is set by organizational requirements rather than mandating host-level API instrumentation.
- T1614.001partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, but system language discovery via registry queries or API calls is a low-volume, legitimate-looking activity that blends into normal system operation and is unlikely to trigger the baseline anomalies the control is configured to surface unless the organization has specifically instrumented for it.
- T1615partialdetects — T1615 runs commands (gpresult, Get-DomainGPO) on a system that generate process execution events, command-line logs, and LDAP traffic to domain controllers — all within A.8.16's enumerated scope (event logs, security tool logs, system activity). However, the clause sets scope by business and security requirements rather than mandating command-line auditing or process telemetry depth, so an implementation monitoring only network flows is fully conformant and sees only generic LDAP queries, not the discovery technique itself.
- T1619partialdetects — Cloud storage enumeration via provider APIs (ListObjectsV2, List Blobs) generates outbound network traffic and access events that fall within A.8.16's monitoring scope, but the clause sets scope by business and security requirements rather than mandating cloud-API instrumentation, so an implementation monitoring only on-premises traffic is fully conformant and sees nothing of this technique.
- T1620partialdetects — A.8.16's item (f) explicitly requires checking that code being executed is authorized and untampered, and item (d) names process injection as an example of unusual system behaviour to detect — both directly relevant to reflective loading — but the clause sets scope rather than mandating host-level memory instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of in-memory reflective loads, mirroring the T1055 anchor's reasoning.
- T1620partialresponds — A.8.16's real-time/continuous monitoring of anomalous behaviour (incl. process injection, unauthorized code execution, memory/resource deviations, and malware-like activity) surfaces T1620 once it is underway inside a process; alerts and procedures then feed the 5.26 response workflow for containment/eradication, exactly as the event-lane definition of `responds` requires.
- T1622partialdetects — A.8.16's monitoring scope can surface debugger-evasion artifacts where process-level telemetry is in scope — the API calls (IsDebuggerPresent, NtQueryInformationProcess), OutputDebugStringW flooding, SEH exception patterns, and the behavior change when malware disengages are all observable to EDR/Sysmon-class instrumentation. But the clause sets scope by business requirement rather than mandating host-process instrumentation depth, so a conformant implementation focused on network traffic, access logs and configuration files sees nothing of an in-process check, and the core technique — the malware querying its own environment — is largely silent at the layers A.8.16 most directly addresses.
- T1647partialdetects — Plist files are configuration files explicitly within A.8.16's monitoring scope (item c: critical or admin level system and network configuration files), and security tools (item d: antivirus, IDS/IPS, DLP) commonly include file integrity monitoring that surfaces plist modifications; however, the clause sets scope by business and security requirements rather than mandating file integrity monitoring of plist files, so an implementation focused on network and application layers is fully conformant and sees nothing of this technique.
- T1648partialdetects — A.8.16's scope includes network traffic, security tool logs (cloud audit tools like CloudTrail fall here), event logs, and resource utilization — all of which can surface serverless execution abuse via anomalous API calls, function creation/invocation patterns, and crypto-mining resource consumption. However, the clause sets scope by business requirements, and an implementation focused on traditional on-premises infrastructure may not comprehensively cover cloud API logs and serverless-specific telemetry; the execution itself happens in cloud provider infrastructure, so what is detected is the organization's-side artifacts rather than the execution itself.
- T1649partialdetects — A.8.16's monitoring scope (access to systems, event logs, security tool logs) can surface certificate-related activity — AD CS enrollment events, certificate store access, certificate-based authentication anomalies — but the clause sets scope by business requirements rather than mandating instrumentation depth, and stealthy variants (golden certificate forgery using stolen CA keys, in-memory certificate theft via crypto APIs) leave little observable telemetry, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1651partialdetects — A.8.16's scope covers event logs, access to systems, security tool logs, and network traffic — all of which surface T1651's artifacts (cloud management API calls, VM agent process executions, management-plane-to-VM traffic). Detection is not full because the technique abuses legitimate cloud services, so distinguishing adversary use from normal administration depends on baselines and thresholds the organization has chosen to configure, and cloud control-plane API monitoring may sit outside the implemented scope.
- T1652partialdetects — A.8.16 detects T1652 where the monitoring scope includes process execution telemetry, command-line auditing, or EDR — driverquery.exe, lsmod, and EnumDeviceDrivers calls produce observable artifacts in event logs and security-tool logs. But the clause sets scope by business and security requirements rather than mandating host-level instrumentation depth, so a conformant network-only or application-layer implementation sees nothing of a local enumeration, and admin-context enumeration blends with normal baseline behavior.
- T1657partialdetects — A.8.16 can detect some forms of financial theft — unauthorized access to financial systems, anomalous user behavior around banking applications, and known attack signatures associated with account compromise or technical theft — but the actual transfer of funds is often a legitimate-looking business action (BEC, authorized-looking wire transfers, pig-butchering victim compliance) that produces no technical indicator on the monitored estate, and the extortion demand itself is a business process outside the monitoring scope.
- T1665partialdetects — A.8.16 monitors network traffic, DNS, and system behaviour, so on-estate indicators of infrastructure hiding (anomalous proxy/VPN use, traffic-filtering attempts against security tools, unusual DNS patterns, masked-domain resolution) are within its scope; however, T1665 is explicitly designed to evade that monitoring by filtering defensive traffic, blending with trusted hosting, and masking destinations, and the off-estate resource-development half (acquire/compromise infrastructure) is not observable, so detection is a slice rather than a bounded remainder of a working mechanism.
- T1666partialdetects — Cloud control plane API calls (CreateAccount, LeaveOrganization, subscription transfers) generate event logs within the organization's own estate that fall under A.8.16's scope of 'event logs relating to system and network activity,' but the technique is explicitly designed to evade defenses by creating new accounts/subscriptions outside existing Service Control Policies and by transferring subscriptions to adversary-controlled tenants, so a conformant A.8.16 implementation focused on traditional network/system telemetry may miss the control-plane activity, and post-transfer activity is outside the victim's monitoring scope entirely.
- T1668partialdetects — A.8.16's scope covers configuration files, event logs, access logs and resource use, so the artifacts of T1668 (patches applied, services disabled, privileges stripped, malware removed) are within the monitored surface; however, the technique is specifically designed to mimic legitimate defensive administration, so the signal-to-noise ratio is poor and detection depends on baselining and correlation that not every conformant implementation performs.
- T1669partialdetects — A.8.16 can detect anomalous network connections and unauthorized access attempts to Wi-Fi networks, but the control's scope is determined by business requirements and may not cover all wireless access points or detect passive discovery of Wi-Fi networks.
- T1671partialdetects — A.8.16's monitoring can detect OAuth application integration activity (new app registrations, consent grants, service principal creation, token issuance) when the organization's monitoring scope includes SaaS platform audit logs, which fall plausibly under 'critical applications' and 'event logs'; however, the clause determines scope organizationally rather than mandating SaaS/identity-provider log ingestion, so an implementation focused only on network and host telemetry is fully conformant and sees nothing of this technique.
- T1673partialdetects — VM discovery commands (esxcli, vim-cmd) and vCenter access produce observable artifacts — command execution logs, authentication/access logs, and management-plane traffic — that fall within A.8.16's scope of system access logs, event logs, and unusual system behaviour against a baseline; however, the clause sets monitoring scope by business requirements rather than mandating hypervisor instrumentation, so an implementation focused on general systems but not on ESXi/vCenter management will be fully conformant and see nothing of this technique.
- T1674partialdetects — A.8.16 explicitly names 'keystroke logging' and 'process injection' as examples of unusual system behaviour to detect, and input injection (simulated keystrokes) is observable through the same mechanisms — process behaviour, USB device events, resource usage spikes, and anomalous input patterns. However, the clause sets scope by business requirements rather than mandating instrumentation depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process input-injection tool or a HID device event, leaving what is caught a slice the implementer chose.
- T1675partialdetects — A.8.16's scope is set by business and security requirements rather than mandating ESXi-specific instrumentation; the technique produces observable artifacts (vSphere SDK API calls, VMware Tools daemon activity, guest VM process execution originating from ESXi, management-plane network traffic) that fall within the clause's listed monitoring items (access to servers, event logs, system/network activity), but an organization monitoring only application and network layers without including ESXi hypervisor infrastructure is fully conformant with A.8.16 and sees nothing of this technique.
- T1677partialdetects — A.8.16's monitoring scope covers configuration files (CI configs), code execution authorization checks, resource use, and network traffic — all of which can surface aspects of T1677, particularly the self-hosted runner scenario where execution occurs on the organization's estate. However, the technique's primary scenarios (Direct, Indirect, and Public Pipeline Execution) run on SaaS CI/CD infrastructure (GitHub Actions, GitLab CI) outside the organization's monitoring vantage point, so the core malicious execution in the pipeline is not visible to A.8.16's instrumentation.
- T1678partialdetects — A.8.16's monitoring scope includes resource use (CPU, memory, bandwidth), performance, event logs, and baseline comparison of access timing — which can surface API hammering, ping loops, and unusual scheduled tasks. But T1678 is specifically designed to evade detection by blending with normal timing, and pure sleep/delay calls produce minimal signal; the technique's core purpose is to look like benign activity, so what monitoring catches is a slice rather than the bulk.
- T1679partialdetects — A.8.16's monitoring scope includes checking that executed code is authorized and untampered, and its anomaly baseline explicitly lists process injection and unusual system behaviour, but selective exclusion is a deliberate absence of action that leaves no positive indicator in the monitored artifacts unless the exclusion pattern itself is anomalous against the baseline.
- T1680partialdetects — A.8.16's scope includes event logs relating to system activity, which can capture process execution and command-line arguments (e.g., `wmic logicaldisk get`, `Get-PSDrive`, `esxcli`), and the baseline-of-normal-behaviour framing can flag unusual enumeration; but the clause sets scope by business and security requirements rather than mandating command-line auditing depth, so an implementation monitoring only network and application layers is fully conformant and sees nothing of this local discovery.
- T1684partialdetects — A.8.16 monitors network, system, and application traffic and logs for anomalous behaviour, which can surface the technical indicators of a social engineering attempt (e.g., unusual access, unauthorized changes, or execution of adversary-supplied instructions), but the purely human-to-human trust-building conversation that precedes the technical action is outside its monitoring scope, leaving a nameable remainder.
- T1684.001partialdetects — A.8.16's monitoring scope (network traffic, event logs, anomalous behaviour against baseline) can surface technical indicators of impersonation such as spoofed email headers, anomalous sender patterns, and deviations in communication behaviour, but the technique's core mechanism is human deception through a trusted identity, which monitoring does not directly observe and which depends on the implementer's chosen scope rather than the clause's mandate.
- T1684.002partialdetects — A.8.16's scope includes inbound network traffic and logs from security tools (which can encompass email gateways and their authentication-failure logs), so a scoped implementation can surface spoofed emails via failed SPF/DKIM checks and anomalous sender patterns; however, the clause sets scope rather than mandating email-header inspection, so an implementation monitoring only other layers is fully conformant and sees nothing of this technique.
- T1685partialresponds — A.8.16 requires real-time/periodic monitoring, anomaly detection against baseline, alerts on events like unauthorized access/tampering indicators, and dedicated personnel/procedures to respond to positive indicators (explicitly referencing 5.26 incident response), which acts on the T1685 event once underway to contain/eradicate; partial because scope is set by org requirements (not mandating coverage of all tampering vectors like ETW manipulation or driver-based bypasses) and it detects/surfaces rather than fully containing all instances.
- T1685.001partialdetects — A.8.16's monitoring scope includes event logs (item e), which is exactly what T1685.001 targets — the technique is designed to defeat the primary detection channel. However, A.8.16 also covers configuration file monitoring (item c, which can catch registry modifications to EventLog keys), security tool logs (item d, which may be independent of Windows Event Log), and process execution visibility, so the disabling action itself remains partially observable through these alternative channels. The grade is partial because the technique's explicit purpose is to limit the data A.8.16 relies on, while bounded detection of the disabling act survives through non-event-log channels.
- T1685.002partialdetects — A.8.16's scope includes event logs, security tool logs, and configuration file monitoring, so changes to cloud logging configurations (e.g., CloudTrail disablement, audit bypass associations) can be detected when those specific data sources are within the monitored scope and when the disablement action is itself logged before taking effect; however, the technique is specifically designed to blind the monitoring system by removing the very data source needed for detection, and A.8.16 sets scope by business requirement rather than mandating cloud audit-log or cloud configuration monitoring, so a fully conformant implementation can miss the technique entirely.
- T1685.003partialdetects — A.8.16's scope (file integrity of security tools, process behaviour, configuration changes, telemetry gaps) can surface indicators of UI spoofing — modified tool binaries, new processes mimicking tool UIs, missing logs from disabled tools — but the technique is specifically engineered to make monitoring appear normal, and a conformant A.8.16 implementation whose scope is set by business requirements rather than mandated instrumentation depth may not catch the spoofed interface itself.
- T1685.004partialdetects — A.8.16 enumerates monitoring of configuration files (item c, which includes audit.conf/audit.rules), unplanned process termination (anomalous behavior a, which catches auditd being killed), and deviations in standard protocol use — all of which surface aspects of this technique. But the technique specifically targets the audit/logging infrastructure itself, and an implementation whose monitoring depends on auditd is fully conformant with A.8.16 yet blind to its own tampering; the clause sets scope rather than mandating an independent monitoring channel, so detection is a slice the implementer chose rather than a bounded remainder of a working mechanism.
- T1685.006partialdetects — A.8.16's scope includes event logs, system configuration files, and system activity, so the act of clearing logs (file deletion/truncation events, command execution, sudden gaps in log continuity) is observable in principle; however, the clause sets scope by business and security requirements rather than mandating file-integrity or command-audit instrumentation, so an implementation monitoring only network and application layers is fully conformant and sees nothing of an in-process log wipe.
- T1685.007partialdetects — A.8.16 explicitly lists process injection and unusual system behaviour as monitoring targets, and the guidance requires checking that executing code is authorized and untampered, so the control's scope covers the observable execution of the technique; however, the technique's purpose is to impair the autonomous defences that perform the monitoring, and a successful prompt injection that causes the LLM to ignore or abort analysis creates a blind spot within the monitoring system itself, leaving a slice of the technique's activity undetected.
- T1686.001partialdetects — Cloud firewall rule modifications generate audit logs (CloudTrail, Azure Activity Log) and configuration changes that fall under A.8.16's explicit scope of configuration file monitoring (item c), event logs (item e), and security tool logs (item d); the technique occurs on the organization's own cloud estate, unlike T1583.001 which sits on a registrar's infrastructure, so the monitoring vantage point exists. The remainder is scope-dependent: organizations not actively monitoring cloud control plane API calls, and subtle rule additions that resemble legitimate changes, may not surface in real time.
- T1686.002partialdetects — A.8.16's scope explicitly includes networking equipment, critical configuration files, and outbound/inbound traffic, and its baseline-comparison logic would flag the traffic effects of firewall rule changes; however, the clause determines scope from business and security requirements rather than mandating instrumentation depth, so an implementation focused on host systems and perimeter traffic without network-device configuration or syslog ingestion is fully conformant and would miss the rule modifications themselves.
- T1687partialdetects — A.8.16 explicitly lists monitoring of security-tool logs, anomalous behaviour patterns (including process injection, malware activity, unauthorized access, and known attack characteristics), resource use, and real-time alerting against a baseline, which surfaces exploitation of defensive components once it produces observable anomalies on the monitored estate; partial because the technique can be entirely silent, target unmonitored defensive layers, or impair the monitoring system itself before detection occurs.
- T1688partialdetects — A.8.16's scope covers configuration file changes (BCD modifications), registry alterations, and event logs that would surface the precursors and the safe mode boot itself, but the technique is specifically designed to disable the endpoint monitoring tools (EDR) that constitute the bulk of A.8.16's instrumentation during the safe mode session, leaving the core execution window unobserved at the host level.
- T1689partialdetects — A.8.16's enumerated monitoring items cover the main downgrade vectors — network traffic monitoring can detect protocol downgrades (the clause explicitly lists 'deviations in use of standard protocols'), and event logs can capture use of older PowerShell versions — but the clause sets scope by organizational requirement rather than mandating instrumentation depth, so an implementation monitoring only some layers is fully conformant and would miss other downgrade vectors such as boot manager changes.
- T1690partialdetects — The technique occurs on the estate (environment variable changes, PSReadLine option commands, history file modifications) and is observable where host telemetry and command execution logging fall inside A.8.16's scope, but the clause sets scope by business requirements rather than mandating instrumentation depth — an implementation monitoring only network traffic or access logs is fully conformant and sees none of it, mirroring the T1055 anchor's reasoning.
- T1696.001partialdetects — A.8.16's scope includes inbound application traffic and detection of unauthorized scanning/automated access, so monitoring can surface bot-like patterns (request volume, timing, lack of normal user behavior) against the SaaS endpoint, but the technique is specifically designed to defeat the very bot-detection mechanism (CAPTCHA) and the clause sets scope by organizational requirements rather than mandating the application-layer instrumentation depth needed to catch sophisticated automation.
- T1697partialdetects — Monitoring can surface some indicators of ASP abuse through access-pattern anomalies (unusual time, location, frequency against the established baseline) and authentication event logs, but the technique is designed to look like legitimate valid-account access, so the theft vectors (phishing, generation on a compromised account, reading from disk) leave little the monitoring scope covers, and successful ASP authentications are indistinguishable from normal logins without specific correlation.
- T1698partialdetects — A.8.16 monitoring can surface some downstream effects of indirect prompt injection — unusual API calls, anomalous data access patterns, and unexpected network traffic generated by a compromised LLM — but the injection itself occurs at the LLM application layer that standard monitoring scope (set by business requirements, not mandated depth) does not inherently cover; an implementation monitoring only network and traditional application layers is fully conformant and sees nothing of the prompt being processed.
- T1699partialdetects — Direct prompt injection occurs at the LLM application interface on the organization's estate, generating network traffic and application logs within A.8.16's scope, but the technique's malicious signal is in natural-language prompt content rather than in network patterns or system behavior, so an A.8.16-conformant implementation monitoring only network and system layers (without LLM-specific content analysis) sees nothing of it.
- T1700partialdetects — A.8.16's network, event-log and resource-use monitoring can surface some indicators of AI-enabled exfiltration (unusual outbound volumes, anomalous API/tool calls, deviations from a behavioural baseline), but the technique's core mechanism—prompt injection steering a sanctioned agent—operates inside legitimate tool use that traditional monitoring is not scoped to inspect, and the resulting data transfer often resembles normal agent activity, leaving a substantial unmonitored remainder.
Prevented OWASP Web Top 10 (2025) risks (24)
OWASP Web Top 10 (2025) risk categories this ISO control helps prevent or mitigate — our analysis (mostly llm_verified).
Direction: ← other covers this;
→ this covers other (F/M/P = full / mostly /
partial). gov = governs / implements (a mandate, not coverage).
Why these map — rationale
- A01partialfinds — A.8.16's monitoring for anomalous behaviour, unauthorized access attempts, and deviations from baseline directly surfaces many Broken Access Control realizations (e.g. unauthorized scanning, successful/unsuccessful protected-resource attempts, unusual access patterns), but does not discover design or code defects like missing authorization checks, IDOR, or CSRF before they are exercised.
- A02partialfinds — A.8.16's monitoring of traffic, access, config files, logs, resource use, baselines, attack signatures, unauthorized attempts and anomalous behaviour directly surfaces misconfigurations that leave attack surface (weak defaults, incomplete hardening, exposed settings) via real-time/periodic detection and alerts.
- A03partialfinds — A.8.16's monitoring of logs, traffic, resource use, anomalous behaviour, malware signatures, and known attack patterns can surface indicators of already-compromised supply-chain components or pipelines after the fact (e.g. via C2 traffic or process anomalies), but does not discover vulnerable/outdated dependencies themselves nor inspect build/signing infrastructure.
- A05partialfinds — A.8.16's monitoring of traffic, logs, anomalous behaviour, known attack patterns (including injection signatures), and unauthorized access can surface realized injection attempts or their effects in runtime data, but does not inspect code or find the underlying neutralization defect itself.
- A07partialfinds — A.8.16's monitoring of access attempts, anomalous behaviour, unauthorized access, and known attack patterns (including those tied to credential stuffing or session anomalies) surfaces many authentication failures after they occur, but does not discover design or implementation defects such as weak password reset flows or flawed session management before exploitation.
- A08partialfinds — A.8.16's monitoring for anomalous behaviour, malware signatures, unauthorized changes, process injection, tampered code execution and baseline deviations directly surfaces many integrity failures (e.g. unsigned updates, CI/CD anomalies, insecure deserialization side-effects) after they occur, but does not discover all members such as purely static trust of unsigned code or deserialization logic bugs before exploitation.
- A10partialfinds — A.8.16's monitoring for anomalous behaviour, unauthorized access attempts, malware indicators, unusual system behaviour and real-time alerts can surface many (but not all) exceptional-condition leaks, fail-open states or inconsistent error paths after they occur.
Control IDs, short titles and the structured attribute table (control type, CIA properties, cybersecurity-concept, operational capability, security domain) are facts from ISO/IEC 27001:2022 Annex A / ISO/IEC 27002:2022. The full implementation guidance prose lives in ISO/IEC 27002:2022 — not reproduced here. Cross-walks to NIST 800-53, NIST CSF 2.0, OWASP ASVS, CWE, MITRE ATT&CK and OWASP Web Top 10 are our own analysis (mostly llm_verified), not an ISO, NIST, MITRE or OWASP product — how ours compare.