@blamejs/exceptd-skills 0.19.16 → 0.19.17

package/data/zeroday-lessons.json

@@ -1,7 +1,7 @@
 {
   "_meta": {
     "schema_version": "1.1.0",
-    "last_updated": "2026-08-10",
+    "last_updated": "2026-08-11",
     "last_threat_review": "2026-05-17",
     "purpose": "Zero-day learning loop output. Each entry maps a CVE to: attack vector, defense chain analysis, framework coverage, new control requirements generated, and exposure scoring. v1.1.0 (2026-05-15): every entry now carries ai_discovered_zeroday boolean + ai_discovery_source enum + ai_discovery_date + ai_assist_factor ladder, per AGENTS.md Hard Rule #7.",
     "note": "Never delete entries. Closed gaps are marked status: closed. History is data.",
@@ -20101,7 +20101,39 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-135",
+        "name": "CONTROL-PANEL-PLUGIN-USER-SURFACE-MUST-NOT-REACH-ROOT-OPERATIONS",
+        "description": "CWP Control Web Panel is the constrained user surface this control governs, and the packet places the defect exactly where the control forbids it: shell metacharacters supplied in the t_total parameter of a filemanager changePerm request reach a shell, so a request-supplied string on the panel's file-manager surface is what selects the command the server runs. Bound to this product the control means the panel's privileged filesystem operations — changing permissions on a hosted account's files is one of them — sit behind their own authorization boundary and are invoked through an argv-array interface, so no parameter of a file-manager request can extend the command line, and the panel authorizes the caller before the operation runs rather than relying on a check the request path reaches afterwards. This is also why the least-privilege gap recorded on this entry does not close the path: the packet's stated access requirement is knowing a valid non-root username, not holding that user's session, so the attacker never authenticates as any panel account, per-account privilege scoping is never consulted, and an AC-6 attestation passes cleanly while an unauthenticated caller reaches the shell. Distinguishing test: from an unauthenticated client against a staging CWP instance, send a filemanager changePerm request whose t_total value carries shell metacharacters and confirm it is refused before any command runs. Precondition: the vendor update is what repairs the endpoint — this control states the property to verify, it does not implement it. Until that update and its restart land, restricting which networks may reach the panel's web surface bounds who can send the request but leaves the endpoint fully exploitable to anything inside the permitted range, and it is unavailable where the panel must stay reachable for hosted-account administration.",
+        "evidence": "Packet vector: 'CWP Control Web Panel (formerly CentOS Web Panel) contains an OS command Injection vulnerability that allows unauthenticated remote code execution via shell metacharacters in the t_total parameter in a filemanager changePerm request. A valid non-root username must be known.' CWE-78. CISA KEV listed 2025-11-04; active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 77. NIST SP 800-53 AC-6 (Least Privilege) is recorded among this entry's citing framework gaps. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "NIST-800-53-AC-6",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "A CWP server answers unauthenticated callers by design and the packet's path ends in command execution on it, so the control's premise holds here without any appliance framing: the exploit needs no credential, only a known non-root username, and the packet records a public PoC with confirmed in-the-wild exploitation. Any instance that was reachable by untrusted callers before the update landed therefore has to be treated as one that may already have run attacker commands. For this product the runbook is to capture the panel and server configuration and the hosted-account inventory, rebuild the host from a known-good image, and rotate everything the panel held or could reach — panel administrator credentials, hosted-account and database passwords, and any API keys stored on it — rather than applying the update in place. Patch-in-place closes the changePerm injection and removes nothing written through it: a web shell under a hosted document root, an added cron entry, or an attacker-created account survives the upgrade intact, and none of those are what a build-version check inspects. Precondition: this is the response for an instance whose reachable window overlapped the exploitation the packet records; an instance that can be shown to have been unreachable by untrusted callers throughout is a straightforward patch item. Either way the packet registers no live-patch path, so the vendor fix carries the service restart or system reboot, and a server updated but not restarted is not yet remediated.",
+        "evidence": "Packet attack vector: 'an OS command-injection flaw (CWE-78) enabling unauthenticated remote command execution on the hosting-control server. CISA KEV-listed 2025-11-04 with confirmed in-the-wild exploitation.' Vector states exploitation requires only that 'A valid non-root username must be known.' active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "CWP ships and updates on its own vendor channel, so a remediation programme whose patch obligation is written as operating-system patching never schedules this update at all — the host's OS package inventory stays current on every report while the panel that answers unauthenticated requests stays on the vulnerable build. For this CVE the clock runs from the KEV listing of 2025-11-04, with a public PoC and confirmed exploitation already in hand at that date, and completion has to be measured per host by the CWP build actually running rather than by an approved change record. The packet registers no live-patch tool and states the vendor patch typically requires a service restart or system reboot per the KEV requiredAction, so a server that took the update without restarting still runs the vulnerable code and must be counted as exposed until it does — on a shared hosting box that restart is also the step most likely to be deferred, because taking it interrupts every hosted site, and a deferral recorded as 'patched' is the specific way this remediation goes wrong.",
+        "evidence": "CISA KEV listed 2025-11-04; active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 77; CWE-78. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' Citing gaps on this entry include ASD Essential Eight 'Patch operating systems' and NIST SP 800-53 SI-2 (Flaw Remediation).",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2"
+        ]
+      }
+    ]
   },
   "CVE-2025-11371": {
     "name": "Gladinet CentreStack and Triofox Files or Directories Accessible to External Parties Vulnerability",
@@ -20819,7 +20851,40 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-122",
+        "name": "EOL-ASSET-DECOMMISSION",
+        "description": "This entry is the case the control exists for, and the packet states both halves. patch_available is true — a fixed build carrying this JavaScriptCore fix exists — and the vector itself states that the impacted product could be end-of-life and/or end-of-service and that users should discontinue product utilization. Those facts together define the requirement: on a unit whose hardware can still take an Apple security update, reaching the fixed build is an interim state; on a unit that can no longer take one, the terminal state is removal or replacement, because that device is exposed not only to this code-execution flaw but to everything found in that engine since its last build. Scope to what the packet names — macOS, iOS, tvOS, Safari and watchOS — and inventory those. The packet ties the CWE-94 sink to JavaScriptCore in those products and gives no mapping into other vendors' browsers or other software that embeds a web engine, so treating every renderer in the estate as an instance of this CVE manufactures findings and replacement work against software no evidence implicates. In operational terms: enumerate every macOS, iOS, tvOS, watchOS and Safari install that renders untrusted web content, record for each whether its hardware can still receive a build carrying the fix, put those on the interim clock against the 2025-10-20 KEV listing, and put the rest on a dated replacement schedule — a risk acceptance with no removal date leaves a KEV-listed flaw with a public PoC and confirmed exploitation in service indefinitely. Precondition on the interim half: live_patch_available is false and the packet records that the vendor patch typically requires a service restart or system reboot per the KEV requiredAction, so a device that downloaded the update but has not restarted still runs the vulnerable code and is not remediated. And because exploitation is confirmed and delivery is attacker-controlled web content, a device that rendered untrusted content while exposed belongs on the incident path rather than being closed on the patch record.",
+        "evidence": "Packet fields for this entry: name 'Apple Multiple Products Unspecified Vulnerability', CWE-94. Vector: 'Apple macOS, iOS, tvOS, Safari, and watchOS contain an unspecified vulnerability in JavaScriptCore that when processing web content may lead to arbitrary code execution. The impacted product could be end-of-life (EoL) and/or end-of-service (EoS). Users should discontinue product utilization.' cisa_kev true with kev_date 2025-10-20, active_exploitation confirmed, poc_available true, cvss 8.8, rwep_score 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' AU-Essential-8-Patch (Patch operating systems), ISO-27001-2022-A.8.8 (Management of technical vulnerabilities), NIST-800-53-SI-2 (Flaw Remediation) and NIS2-Art21-patch-management (Vulnerability handling and disclosure) are recorded as citing gaps against this entry — every one of them measures whether a fix was applied within a timescale, and none of them has a terminal state for a device on which no future fix will ever be installable.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-patch-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "The population this control has to bite on for this entry is the one the decommission requirement produces: macOS, iOS, tvOS, watchOS and Safari installs that cannot reach a build carrying the JavaScriptCore fix. For those there is no future build that clears them, so the fixed build has to function as an access condition — organizational mail, VPN, document stores and identity access denied to a device below it — rather than as a stale row on a patch-compliance dashboard that ages indefinitely. The same condition covers the shorter window on devices that can take the fix: the packet records that the vendor patch typically requires a restart, so a device between download and restart is still in the exposed population and needs an access-side answer, not a patch-side one. The distinguishing test is to enrol a device pinned below the fixed build and confirm the policy actually denies it access to protected resources; an estate that surfaces the stale build on a report while the device keeps its mail and VPN has recorded the exposure rather than removed it. Precondition, and it is the reason this cannot be logged as the mitigation: an access condition bounds what an attacker gains from a compromised device, it does nothing to the device itself. It does not remove the JavaScriptCore defect, does not evict an implant on a device already compromised — and the packet's note that Apple flaws of this class are typically used in targeted-spyware chains makes that the case to plan for — and it reaches only devices the management channel enrols. Personally-owned or unenrolled Macs, iPhones, Apple TVs and Watches touching organizational data are not covered by it, and the only lever against those is removing the access path they use.",
+        "evidence": "Packet fields for this entry: CWE-94 code execution in JavaScriptCore across Apple macOS, iOS, tvOS, Safari and watchOS, with the vector stating the flaw 'when processing web content may lead to arbitrary code execution' and that the impacted product could be end-of-life and/or end-of-service. cisa_kev true, kev_date 2025-10-20, active_exploitation confirmed, poc_available true, cvss 8.8, rwep_score 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' attack_vector adds that Apple zero-days of this class are typically used in targeted-spyware chains. AU-Essential-8-Patch (Patch operating systems) and UK-CAF-B4 (System security) are recorded as citing gaps against this entry; both score patch cadence and system-protection policy rather than making the fixed build a precondition of access.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-121",
+        "name": "MOBILE-ZERO-CLICK-HARDENING",
+        "description": "The delivery path the packet records is attacker-controlled web content processed by JavaScriptCore, and its attack_vector notes that Apple flaws of this class are typically used in targeted-spyware chains. For the cohort plausibly inside that targeting set — executives, journalists, legal and security staff — the reboot-gated update is not fast enough on its own, so place their macOS and iOS devices in Apple's reduced-attack-surface mode, where untrusted web content, message attachments, link previews and fonts are not processed automatically. That narrows the path into the CWE-94 sink during the window that opened before anyone in the estate learned of the flaw: the packet records confirmed exploitation and a public PoC, so the exposure predates the 2025-10-20 KEV listing and the assignment cannot be a reaction to it. The mode only helps if it was already on when the content arrived, which makes it a standing posture assigned to the high-risk cohort ahead of the next disclosure rather than a response to this one. Preconditions, and they bound this tightly. The mode reduces what is processed automatically; it does not repair JavaScriptCore, and content the user deliberately chooses to load in a still-permitted path still reaches the engine. It gives nothing against a device already compromised — a device in this cohort that rendered untrusted content while exposed belongs on the incident path, not the hardening path. And on a device the decommission requirement identifies as unable to reach the fixed build, this is the only remaining lever and it is a holding measure until that device is replaced, not a substitute for replacing it.",
+        "evidence": "Packet fields for this entry: vector 'Apple macOS, iOS, tvOS, Safari, and watchOS contain an unspecified vulnerability in JavaScriptCore that when processing web content may lead to arbitrary code execution', CWE-94. attack_vector: 'a code-execution flaw (CWE-94) reachable via attacker-controlled web/media content. CISA KEV-listed 2025-10-20 with confirmed in-the-wild exploitation (Apple zero-days of this class are typically used in targeted-spyware chains).' cisa_kev true, kev_date 2025-10-20, active_exploitation confirmed, poc_available true, cvss 8.8, rwep_score 77. patch_available true, live_patch_available false, with live_patch_notes recording that the vendor patch typically requires a service restart or system reboot. UK-CAF-B4 (System security) is recorded as a citing gap against this entry; it addresses protection of deployed systems but names no reduced-attack-surface posture for a targeted cohort during the period before a reboot-gated fix lands.",
+        "gap_closes": [
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2025-2746": {
     "name": "Kentico Xperience CMS Authentication Bypass Using an Alternate Path or Channel Vulnerability",
@@ -21195,7 +21260,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet gives AEM Forms on JEE an unauthenticated remote code-execution path, confirmed in-the-wild exploitation, a public PoC and a KEV listing dated 2025-10-15, so for any instance that was network-reachable by untrusted callers before the update landed the question remediation must answer is not whether the flaw is closed but whether something already ran. On a JEE deployment the artifacts that outlive the patch are ordinary parts of the platform: a deployed application archive, a JSP or class file written under the application server's document or work directories, a scheduled job, or a credential lifted from the server's configuration. The vendor update removes none of them and a build-version check inspects none of them. The requirement for this product is therefore to capture the AEM Forms configuration and deployed-application inventory, rebuild the server from a known-good image, and rotate the credentials that instance held — application-server administrative accounts, repository and datastore credentials, and any service identity it authenticates as — instead of updating in place. Precondition: this applies to instances whose reachable window overlapped the exploitation the packet records; where an instance can be shown to have been unreachable by untrusted callers throughout, the vendor update on its own is the proportionate response. Either way the packet registers no live-patch path, so the fix carries the service restart or system reboot the vendor requires, and an instance updated but not restarted is not yet remediated.",
+        "evidence": "Packet vector: 'Adobe Experience Manager Forms in JEE contains an unspecified vulnerability that allows for arbitrary code execution.' Attack vector: 'a code-execution flaw (CWE-94) enabling unauthenticated remote code execution on the AEM Forms server. CISA KEV-listed 2025-10-15 with confirmed in-the-wild exploitation.' active_exploitation confirmed; poc_available true; CVSS 8.8; RWEP 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "AEM Forms on JEE runs inside an enterprise application server whose restarts are normally negotiated into a change window, and the packet states the vendor patch typically requires a service restart or system reboot with no live-patch tool registered — so the specific way this remediation fails is an instance that has taken the update but not restarted being recorded as patched while it still runs the vulnerable code. The clock for this CVE runs from the KEV listing of 2025-10-15, with a public PoC and confirmed exploitation already in hand at that date, and completion has to be measured per instance by the build running after the restart rather than by a deployment ticket being closed. Enumerate the JEE Forms instances specifically rather than treating this as an estate-wide Experience Manager item: the packet names Adobe Experience Manager Forms in JEE as the affected product and ties the code-execution path to nothing else, so widening the remediation to every Experience Manager deployment manufactures work against software this entry does not implicate — widen it only where a verified source identifies another deployment carrying the same component. Because the path needs no authentication, priority follows reachability: instances that untrusted callers can reach are the population to take first.",
+        "evidence": "Packet names the affected product as 'Adobe Experience Manager Forms in JEE' with 'an unspecified vulnerability that allows for arbitrary code execution' (CWE-94), reachable per the attack vector as 'unauthenticated remote code execution on the AEM Forms server'. CISA KEV listed 2025-10-15; active_exploitation confirmed; poc_available true; CVSS 8.8; RWEP 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' Citing gaps include ASD Essential Eight 'Patch operating systems' and NIST SP 800-53 SI-2 (Flaw Remediation).",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2"
+        ]
+      }
+    ]
   },
   "CVE-2025-47827": {
     "name": "IGEL OS Use of a Key Past its Expiration Date Vulnerability",
@@ -21830,7 +21918,35 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-122",
+        "name": "EOL-ASSET-DECOMMISSION",
+        "description": "This entry states both halves of the control itself: a vendor patch is recorded as available, and the packet's own vector says the impacted product could be end-of-life and/or end-of-service and that users should discontinue product utilization. Taken together they define the requirement — a host still exposed to this after the 2025-10-06 re-listing is a host whose browser may have no ongoing update path, so reaching the vendor fix is an interim state and the terminal state is taking Internet Explorer out of service on that host, or replacing the host where the workload depends on the browser. Scope the inventory to what the packet names: Internet Explorer rendering attacker-controlled web content. The packet provides no mapping from this uninitialized-memory defect into any other browser or HTML-rendering component, so treating every renderer in the estate as an instance of this CVE manufactures findings and removal work against software no evidence implicates. In operational terms: enumerate every host that can still use Internet Explorer to render untrusted web content, record per host whether it can take the vendor update and be restarted, put those hosts on the interim clock, and put every host that cannot on a dated removal or replacement schedule — a risk acceptance with no removal date leaves a KEV-listed flaw with a public PoC and confirmed in-the-wild exploitation in service indefinitely. Precondition on the interim half, which is where this is routinely over-claimed: the packet registers no live-patch path and states the vendor patch typically requires a service restart or system reboot, so a host that received the update but was not restarted still runs the vulnerable code and is not remediated; and applying the fix for this defect says nothing about anything found in that browser since, which is why the requirement cannot end at 'every install reports the fixed build' — that phrasing marks an abandoned browser compliant while it stays exposed. Because active exploitation is confirmed and the delivery path is a page the victim visits, a host that browsed untrusted content while exposed belongs on the incident path rather than being closed on the patch record.",
+        "evidence": "Packet vector: 'Microsoft Internet Explorer contains an uninitialized memory corruption vulnerability that could allow for remote code execution. The impacted product could be end-of-life (EoL) and/or end-of-service (EoS). Users should discontinue product utilization.' Packet attack_vector: an uninitialized-memory / use-after-free corruption flaw (CWE-94) in Internet Explorer, exploitable by an attacker-controlled web page for code execution in the browser, and 'the legacy re-listing exists because long-tail unpatched estates remain exposed'. CISA KEV listed 2025-10-06; active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "This entry is the case where a remediation clock anchored on patch availability produces no action at all. The packet describes the 2025-10-06 entry as a legacy re-listing that exists because long-tail unpatched estates remain exposed — that is, the vendor fix predates the listing by a wide margin — so a flaw-remediation programme that ages this CVE from its original patch date reads it as long since closed and nothing fires, while the KEV listing is the statement that it is being exploited now. Applied to this CVE the control means the clock starts at the KEV listing date, not the patch date, and the verified mitigation recorded per host is one of exactly two things the packet supports: the vendor update with its restart taken, or — following the packet's own instruction that users should discontinue product utilization — a documented, dated discontinuation of Internet Explorer on that host. Precondition, and it is what usually gets an SLA marked met while the flaw stays live: the packet registers no live-patch path and states the vendor patch typically requires a service restart or system reboot, so a host that has installed the update without restarting still runs the vulnerable code and must not be counted as mitigated; and on a host with no supported update path the clock can only be met by the discontinuation route, since there is no compensating rule in this packet that keeps the browser safely rendering untrusted pages. Distinguishing test: pull the remediation due-date the vulnerability-management programme actually assigned this CVE and confirm it is anchored on 2025-10-06 — an estate whose report ages this from the original fix shows it closed years ago while unrestarted and unsupported installs remain exposed to confirmed in-the-wild exploitation with a public PoC available.",
+        "evidence": "CISA KEV listed 2025-10-06; active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 77. Packet attack_vector: 'the legacy re-listing exists because long-tail unpatched estates remain exposed'; delivery is an attacker-controlled web page giving code execution in the browser. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' Packet vector: the impacted product could be end-of-life (EoL) and/or end-of-service (EoS); users should discontinue product utilization.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2021-43226": {
     "name": "Microsoft Windows Privilege Escalation Vulnerability",
@@ -22185,7 +22301,32 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "This entry is the case where the control's 'whichever is later' clause is the whole point: the packet records a vendor patch as available and a KEV listing dated 2025-10-02, so the trigger that governs is the listing, and the requirement is that a CVE bearing a 2014 identifier re-enters the remediation queue on the expedited clock instead of being filed as historical by a vulnerability-management process that ages work by CVE year. The population to put on that clock is not every host with bash installed — it is the hosts where the packet's condition holds, namely those where attacker-controlled data reaches a bash environment, with CGI named as the example — so the first deliverable is the enumeration of those service paths, and those hosts are remediated before the general fleet. Completion is measured per host as the installed bash package plus the service restart or system reboot that the packet's live-patch note records as typically required by the KEV requiredAction, since the packet registers no live-patch tool for this entry and a host that installed the package without that restart is not the state the packet describes as remediated. Precondition: this is a scheduling control and reduces nothing by itself; it commits the clock and the ordering, and the work still has to be done. It also says nothing about a host already exploited — the packet records active exploitation as confirmed with a public exploit available, so a host that exposed a bash-reachable path during the exposure window needs triage and credential review rather than closure on a package version.",
+        "evidence": "Packet: GNU Bash contains an OS command injection vulnerability (CWE-78) in Bash environment-variable parsing, described as a Shellshock-family flaw, which allows remote attackers to execute arbitrary commands via a crafted environment, enabling remote command execution wherever attacker-controlled data reaches a Bash environment such as CGI. cisa_kev true, kev_date 2025-10-02, active_exploitation confirmed, poc_available true, CVSS 9.8, RWEP 77. patch_available true; live_patch_available false with live_patch_notes: no live-patch tool registered for this entry, and the vendor patch typically requires a service restart or system reboot per the KEV requiredAction.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-018",
+        "name": "SCANNER-PAPER-COMPLIANCE-TEST",
+        "description": "For this CVE the paper-compliance failure is a bash package version read, and the packet gives two reasons it is not evidence. First, the packet places the flaw in the Shellshock family, so what has to be shown is that a host runs a build carrying the fix for this member — a build merely newer than the family's first fix is not the same claim, and an estate that closed its Shellshock work years ago on a family-level attestation can hold hosts this member still reaches. Second, exposure here is conditional rather than universal: the packet's condition is that attacker-controlled data reaches a bash environment, with CGI as the named example, so a scan that scores every host carrying bash identically measures neither the real exposure nor its absence. The operational test the control demands is therefore behavioural, and the packet supplies its basis — the trigger is a crafted environment and a public exploit exists: on a staging copy of each enumerated service path, deliver the crafted environment and confirm no injected command runs, then repeat it after the service restart or system reboot the packet's live-patch note records as typically required, since the check run before that restart can pass on the updated package while the exposed path still behaves as before. Scope the claim to what the scanner actually owns: a host-package scan is evidence only for the bash copies the package manager manages, so a clean result covers those copies and no others, and any copy shipped inside an image or appliance that the host scan does not enumerate has to be identified from its own vendor's build information and updated through that vendor's path rather than inferred clean. Precondition: this control changes what counts as proof, not the exposure — it neither patches a host nor detects one already exploited, and with active exploitation confirmed a path that was reachable during the window belongs on the incident path regardless of what the re-test now returns.",
+        "evidence": "Packet: OS command injection (CWE-78) in Bash environment-variable parsing, described as a Shellshock-family flaw; GNU Bash allows remote attackers to execute arbitrary commands via a crafted environment; exploitation is enabled wherever attacker-controlled data reaches a Bash environment such as CGI. poc_available true, active_exploitation confirmed, cisa_kev true with kev_date 2025-10-02, CVSS 9.8, RWEP 77. patch_available true; live_patch_available false, live_patch_notes recording no live-patch tool for this entry and a vendor patch that typically requires a service restart or system reboot per the KEV requiredAction.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "AU-Essential-8-Patch"
+        ]
+      }
+    ]
   },
   "CVE-2017-1000353": {
     "name": "Jenkins Remote Code Execution Vulnerability",
@@ -25405,7 +25546,40 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-042",
+        "name": "INCOMPLETE-PATCH-SEQUENCE-SEVERITY-MULTIPLIER",
+        "description": "The packet states it outright: CVE-2025-53771 is a patch bypass for this CVE, and the CVE-2025-53771 updates include more robust protection than the ones shipped for CVE-2025-49706. That inverts the usual reading of a remediation record — a SharePoint Server that took the CVE-2025-49706 update and closed the ticket carries a 'patched' verdict against a fix the vendor has since superseded, so the flaw-remediation attestation reads clean while the authentication path it was meant to close stays reachable through the bypass. For this CVE the remediated state is therefore defined by the later build: verify each SharePoint Server against the build carrying the CVE-2025-53771 protections rather than against the update issued for CVE-2025-49706, and hold the two as one open remediation item instead of two closed ones. The packet also names this as the ToolShell chain entry point and records it as chainable with CVE-2025-49704, so the exposure is a sequence on one authentication primitive rather than a discrete ticket — the intake expectation is that further bypasses on the same path are likely and warrant standing monitoring, not a closed record. Precondition, and it is this control's limit: it changes scoring, scheduling and record-keeping and remediates nothing by itself. The packet registers no live-patch path and a vendor patch typically requiring a service restart or system reboot, so a server that has taken the superseding update but not been restarted is still running the vulnerable code.",
+        "evidence": "Packet vector for CVE-2025-49706: 'This vulnerability could be chained with CVE-2025-49704. CVE-2025-53771 is a patch bypass for CVE-2025-49706, and the updates for CVE-2025-53771 include more robust protection than those for CVE-2025-49706.' Packet attack_vector: 'improper authentication (CWE-287) on SharePoint Server — the ToolShell chain entry point — letting an unauthenticated attacker reach the RCE primitives.' cwe_refs CWE-287; cisa_kev true; kev_date 2025-07-22; active_exploitation 'confirmed'; cvss 9.1; rwep_score 83; poc_available true; patch_available true; live_patch_available false; live_patch_notes 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' Cited as insufficient on this entry: NIST SP 800-53 Rev 5 SI-2 'Flaw Remediation', ISO/IEC 27001:2022 A.8.8, EU NIS2 'Cybersecurity risk-management measures (vulnerability handling and disclosure)', ASD Essential Eight 'Patch operating systems'.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-handling",
+          "AU-Essential-8-Patch"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-129",
+        "name": "ENTERPRISE-ERP-MANAGEMENT-FUNCTION-AUTHENTICATION",
+        "description": "SharePoint Server's authentication check is where the product decides who a caller is, and CWE-287 on this entry is that decision failing. The packet's vector records spoofing performed over a network — a caller presenting an identity it does not hold — and the packet's attack_vector records the consequence as an unauthenticated attacker reaching the RCE primitives at the ToolShell chain entry point. Under either framing the requirement is the same, and that is the point: the privileged SharePoint functions behind the check must establish their caller's authority themselves rather than inheriting a verdict settled once at the front door, and the server's surface must be segmented so a caller with no operational need cannot present the spoofed request at all. This is exactly why the least-privilege and identity-and-access controls cited as insufficient on this entry never engage — the attacker holds no SharePoint account, so no per-account privilege scoping is consulted, no credential is guessed and no second factor is ever requested; an identity-and-access review and a least-privilege attestation both pass cleanly while an unauthorized caller is inside. Distinguishing test: from a segment with no operational need to reach the server, send a staging SharePoint Server requests bearing a spoofed caller identity and confirm each privileged function refuses before it runs; confirming that every named SharePoint user authenticates correctly tests nothing on this path. Precondition: the repair to the authentication decision is the vendor's — this control names the property to verify and the segmentation that bounds who can attempt the spoof; segmentation limits the caller population and closes nothing against anything already inside the permitted segment.",
+        "evidence": "Packet vector for CVE-2025-49706: 'Microsoft SharePoint contains an improper authentication vulnerability that allows an authorized attacker to perform spoofing over a network', with the recorded impact being viewing sensitive information and making changes to disclosed information. Packet attack_vector: 'improper authentication (CWE-287) on SharePoint Server — the ToolShell chain entry point — letting an unauthenticated attacker reach the RCE primitives. CISA KEV-listed 2025-07-22 with confirmed in-the-wild exploitation.' cwe_refs CWE-287; cvss 9.1; rwep_score 83; poc_available true; patch_available true; live_patch_available false. Cited as insufficient on this entry: UK NCSC CAF v3.2 B2 'Identity and access control' and NIST SP 800-53 Rev 5 AC-6 'Least Privilege'.",
+        "gap_closes": [
+          "UK-CAF-B2",
+          "NIST-800-53-AC-6"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet places this flaw at the entry of a chain that reaches the RCE primitives, chainable with CVE-2025-49704, with exploitation confirmed from the 2025-07-22 KEV listing and a public PoC — so a SharePoint Server that was network-reachable on an affected build across that window is a triage subject rather than a patch ticket. The update repairs the authentication decision; it revokes nothing obtained while that decision was failing. For an authentication-bypass entry point that means the credentials and cryptographic material the server's application identity holds are rotated as part of remediation rather than as a follow-up, and the content the server hosts and serves is compared against a known-good baseline rather than inferred clean from the absence of an alert. That last point is the specific reason the anti-malware control cited as insufficient on this entry cannot carry the remediation: an artifact written through the application's own request path by an attacker who authenticated as nobody is not a signature-matched sample, so a fully deployed and current anti-malware estate reports clean over it. Preconditions and limits: the rebuild-and-rotate scope is the server's own hosts and its own secrets — an identity outside that scope reached from the foothold is separate containment; and the triage is bounded by an exposure window the operator can actually establish, so a deployment without retained per-request access logs for the period cannot scope the hunt at all, in which case rebuild is the safe default rather than an evidence-driven choice. Note also that the vendor update alone does not settle the exposure here: the packet records CVE-2025-53771 as a patch bypass for this CVE, so the update must be the superseding build and the restart it requires must be completed.",
+        "evidence": "Packet attack_vector for CVE-2025-49706: 'improper authentication (CWE-287) on SharePoint Server — the ToolShell chain entry point — letting an unauthenticated attacker reach the RCE primitives. CISA KEV-listed 2025-07-22 with confirmed in-the-wild exploitation.' Packet vector: 'This vulnerability could be chained with CVE-2025-49704. CVE-2025-53771 is a patch bypass for CVE-2025-49706'. active_exploitation 'confirmed'; kev_date 2025-07-22; poc_available true; cvss 9.1; rwep_score 83; patch_available true; live_patch_available false, with live_patch_notes recording no registered live-patch tool and that the vendor patch typically requires a service restart or system reboot per the KEV requiredAction. Cited as insufficient on this entry: CIS Controls v8 10.1 'Deploy and Maintain Anti-Malware Software'.",
+        "gap_closes": [
+          "CIS-Controls-v8-10.1"
+        ]
+      }
+    ]
   },
   "CVE-2025-53770": {
     "name": "Microsoft SharePoint Deserialization of Untrusted Data Vulnerability (variant: CVE-2025-53770)",
@@ -27671,7 +27845,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-127",
+        "name": "UNSUPPORTED-NETWORK-DEVICE-ENUMERATION-AND-RETIREMENT",
+        "description": "This packet carries the two facts that have to be reconciled per unit before anything else: a vendor patch is recorded as available, and the entry's own vector states the impacted products could be end-of-life and/or end-of-service and that users should discontinue product utilization. So the first requirement is an inventory that names every ASUS Lyra Mini and ASUS GT-AC2900 in service with its running firmware and a support status obtained from the vendor rather than assumed in either direction. Units whose model still has supported firmware are patch items on the clock that opened with the 2025-06-02 KEV listing; units with no supported firmware cannot be patched at all and are replacement items with a dated schedule, because a risk acceptance with no removal date leaves a device carrying a public PoC and confirmed in-the-wild exploitation in service indefinitely, at a position where it is the network boundary for everything behind it. Scope the inventory to the two models the packet names — no mapping from this authentication flaw into other ASUS models or other consumer routers is established here, and treating every router in the estate as an instance of this CVE manufactures replacement work against hardware no evidence implicates; widen only where a verified source identifies another model carrying the same defect. Precondition on the interim half, which is where this is normally over-claimed: the packet registers no live-patch path and notes the vendor patch typically requires a service restart or system reboot per the KEV requiredAction, so a unit that has taken the firmware but not restarted still runs the vulnerable code and is not remediated; and reaching the fixed firmware closes this defect only, saying nothing about anything found in a model since — which is exactly why the requirement cannot end at 'every unit reports the fixed firmware'. Distinguishing test: produce, per unit, the model, hardware revision, running firmware and the vendor's support status for it, and confirm each unit is either at the fixed firmware or on a dated removal schedule — a patch-compliance report that lists an unsupported model as 'no update available' and closes the item marks an abandoned device compliant while it stays exposed.",
+        "evidence": "Packet name: 'ASUS Routers Improper Authentication Vulnerability'. Vector: 'ASUS Lyra Mini and ASUS GT-AC2900 devices contain an improper authentication vulnerability that allows an attacker to gain unauthorized access to the administrative interface. The impacted products could be end-of-life (EoL) and/or end-of-service (EoS). Users should discontinue product utilization.' cwe_refs CWE-287; cisa_kev true with kev_date 2025-06-02; active_exploitation confirmed; poc_available true; cvss 9.1; rwep_score 77; patch_available true; live_patch_available false with live_patch_notes 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The trigger condition this control exists for is present here in the form the packet documents: a pre-authentication flaw on the administrative interface of the device that sits at the network boundary, with confirmed in-the-wild exploitation and a public PoC. The attacker reaches the administrative interface without holding any account, and administrative access means the device's configuration and its stored credentials are attacker-writable — which firmware does not revert. So for any ASUS Lyra Mini or GT-AC2900 that was reachable during the exposure window, the default response is capture the running configuration for analysis, reset to factory defaults and re-provision from a known-good configuration, and rotate the device's administrative credential along with any credential held in or reachable from that configuration — not an in-place firmware upgrade that leaves the attacker's settings intact underneath a version number that now reads clean. Precondition, and it is the half most often skipped: a rebuild restores configuration integrity, it does not deliver fixed code. A unit rebuilt onto firmware that still lacks the fix is exploitable again the moment it is reachable, so the rebuild has to land on the fixed firmware, and where the model has no fixed firmware the rebuild is not a remediation at all — that unit belongs on the replacement schedule rather than being cycled through factory resets. The rebuild also depends on a known-good configuration baseline recorded before the exposure window; where none exists the configuration must be re-entered by hand, because a backup taken during the window carries the attacker's settings forward into the 'clean' device. Distinguishing test: compare each unit's running configuration — administrative accounts, and whether the administrative interface answers from outside the local network — against a known-good baseline, rather than confirming only that the firmware version is current. A fleet where every unit reports fixed firmware while carrying attacker-set administrative configuration passes a patch attestation and stays compromised.",
+        "evidence": "Packet vector: 'ASUS Lyra Mini and ASUS GT-AC2900 devices contain an improper authentication vulnerability that allows an attacker to gain unauthorized access to the administrative interface.' attack_vector: 'an improper-authentication flaw (CWE-287) letting an unauthenticated attacker bypass authentication on the router's administrative interface. CISA KEV-listed 2025-06-02 with confirmed in-the-wild exploitation.' active_exploitation confirmed; poc_available true; cvss 9.1; rwep_score 77; patch_available true; live_patch_available false with live_patch_notes recording that the vendor patch typically requires service restart or system reboot.",
+        "gap_closes": [
+          "UK-CAF-B2"
+        ]
+      }
+    ]
   },
   "CVE-2025-3935": {
     "name": "ConnectWise ScreenConnect Improper Authentication Vulnerability",
@@ -27731,7 +27928,29 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet's own conditional is why this is a rebuild-and-rotate case rather than a patch ticket: the improper-authentication flaw allows a ViewState code injection attack, which allows remote code execution if machine keys are compromised. Those keys are secret material held by the ScreenConnect server, and applying the vendor update does not change them — so an estate that patches and closes the finding has repaired the injection path while leaving in place the one condition the packet names as turning it into code execution. Bound to this product, remediation of an on-premises ScreenConnect server that was reachable and unpatched during the window opened by the 2025-06-02 KEV listing means: export the server's configuration for offline review before touching it; regenerate the machine keys the ViewState path depends on rather than carrying them across the upgrade; rotate every credential, session and access token the server issued or stored; and default to rebuilding from vendor media wherever the server's state cannot be attested. The packet records a vendor patch with no live-patch path and a fix that typically requires a service restart or system reboot, so a server that has taken the update but not restarted is still running the vulnerable code and must be counted as exposed rather than remediated — a deferred restart recorded as 'patched' is the specific way this goes wrong. Precondition, and it is the one this control is most often recorded without: rotation only holds if the regenerated keys are not reinstated from a pre-incident backup or a configuration snapshot captured during the exposure window, because a rebuild that restores the old machine keys restores exactly the condition the packet names. Rotation also does not retroactively invalidate use already made of the keys before they were changed, so a server inside that window belongs on the incident path and not on the patch record.",
+        "evidence": "Packet vector: 'ConnectWise ScreenConnect contains an improper authentication vulnerability. This vulnerability could allow a ViewState code injection attack, which could allow remote code execution if machine keys are compromised.' attack_vector: an improper-authentication flaw (CWE-287) letting an unauthenticated attacker bypass authentication via ASP.NET ViewState / machine-key abuse. CISA KEV-listed 2025-06-02, active_exploitation confirmed, poc_available true, CVSS 9.8, RWEP 77. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-078",
+        "name": "ENDPOINT-MGMT-DEPLOYMENT-CHANNEL-INTEGRITY",
+        "description": "ScreenConnect is a remote-support server and the packet's stated outcome is code execution on it, which by definition puts everything that server's process can read or write inside the attacker's reach — not the operator console alone, but the artifacts the server serves out, its session and connection configuration, and any credential material stored alongside them. Treat that content as a privileged distribution channel rather than as ordinary application data: file-integrity-monitor the ScreenConnect installation and every directory it serves artifacts from, alert on writes that do not correspond to a sanctioned administrator action or a vendor update, and track the server as a management-plane asset on KEV-priority patching rather than as an internal line-of-business application. This is the half of the response that retains value after the update lands, because the fix closes the authentication bypass but removes nothing placed through it beforehand, and a modified artifact served by a legitimate, trusted support server is not something authentication logging or signature-based endpoint tooling has an artifact to match. Precondition: the monitoring detects a change only against a baseline that predates the exposure window opened by the 2025-06-02 KEV listing — stood up afterwards it baselines whatever is already present and will report clean. Where no such baseline exists, the served content has to be compared against vendor-supplied originals or replaced from them, not certified by the monitor. Note also that the packet conditions the code-execution outcome on machine-key compromise, so this scope applies to servers where that condition cannot be excluded rather than to every install by default.",
+        "evidence": "Packet: ConnectWise ScreenConnect, CWE-287, with the vector stating the ViewState code injection 'could allow remote code execution if machine keys are compromised' and the attack_vector describing an unauthenticated attacker bypassing authentication via ASP.NET ViewState / machine-key abuse. CISA KEV-listed 2025-06-02 with active_exploitation confirmed and poc_available true; CVSS 9.8, RWEP 77. patch_available true; live_patch_available false, with the vendor patch typically requiring a service restart or system reboot per the KEV requiredAction. The entry's citing framework gaps include EU NIS2 Art. 21 supply-chain security measures.",
+        "gap_closes": [
+          "NIS2-Art21-supply-chain"
+        ]
+      }
+    ]
   },
   "CVE-2025-35939": {
     "name": "Craft CMS External Control of Assumed-Immutable Web Parameter Vulnerability",
@@ -28252,7 +28471,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The packet puts this flaw within reach of an unauthenticated attacker and records the traversal being used in the wild to write to startup paths for code execution, so there is no interior authentication step between a published exploit and the Srimax Output Messenger server — the exposure is the interval between the 2025-05-19 KEV listing and the fixed build actually running. That makes it a KEV-clock item rather than an application-maintenance-window item. Where the clock has to run to is the part that matters for this product: the packet registers no live-patch tool and states the vendor patch typically requires a service restart or system reboot per the KEV requiredAction, so a server that has taken the update but has not been restarted is still running the vulnerable code and must be counted as exposed. That restart is also the step most likely to be deferred on a messaging server people are working in through the day, and a deferral recorded as 'patched' is indistinguishable on a compliance dashboard from a real fix — which is the specific way this remediation goes wrong. Distinguishing test: read the version off the running Output Messenger service and confirm the restart completed, rather than accepting the change record or the deployment console as evidence. Precondition: this is a deployment clock and nothing more. It settles nothing about a server that was already reachable on an affected build, where the packet's read-and-write primitive makes the question a triage one.",
+        "evidence": "Packet fields for CVE-2025-27920 ('Srimax Output Messenger Directory Traversal Vulnerability'): cwe_refs CWE-22; cisa_kev true; kev_date 2025-05-19; active_exploitation 'confirmed'; cvss 7.5; rwep_score 77; poc_available true; patch_available true; live_patch_available false; live_patch_notes 'No live-patch tool registered for this entry at bulk-import time. Vendor patch typically requires service restart or system reboot per the KEV requiredAction.'; ai_discovered false. Packet attack_vector: 'a directory-traversal flaw (CWE-22) in Srimax Output Messenger, letting an unauthenticated attacker read or write files outside the intended directory (used in the wild to write to startup paths for code execution)'. Cited as insufficient on this entry: ASD Essential Eight 'Patch operating systems', ISO/IEC 27001:2022 A.8.8 'Management of technical vulnerabilities', UK NCSC CAF B4 'System security'.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet does not stop at file disclosure — it records the traversal being used in the wild to write files outside the intended directory, specifically to startup paths, for code execution. That makes patch-in-place the wrong default for any Srimax Output Messenger server that was reachable on an affected build: the update closes the traversal and removes nothing already written through it. The sequencing is what is specific to this CVE and is easy to get backwards. The packet records the vendor fix as typically requiring a service restart or system reboot, and a reboot is exactly the event that runs whatever an attacker placed in a startup or autorun location — so the enumeration has to come first: compare the server's startup and autorun set, and the filesystem regions the service can write, against a known-good baseline BEFORE taking the restart the patch needs, and rebuild from known-good where that comparison cannot be made cleanly. The read half of the primitive needs its own action rather than the same one: the packet describes access to sensitive files outside the intended directory leading to configuration leakage, so every credential and key present in the server's configuration is treated as disclosed and rotated, not inspected for evidence of use. Preconditions and limits: the packet names no implant, tool or artifact, so the trigger for this runbook is reachability during the exposure window rather than an observed indicator; and rotation reaches only material the server itself held — any credential reused elsewhere in the estate is a separate containment question this control does not cover.",
+        "evidence": "Packet attack_vector for CVE-2025-27920: an unauthenticated attacker can 'read or write files outside the intended directory (used in the wild to write to startup paths for code execution)'. Packet vector: the traversal 'allows an attacker to access sensitive files outside the intended directory, potentially leading to configuration leakage or arbitrary file access'. active_exploitation 'confirmed'; cisa_kev true with kev_date 2025-05-19; poc_available true; patch_available true — the fix exists, which is what makes 'patched' the verdict this control overrides; live_patch_available false with live_patch_notes 'Vendor patch typically requires service restart or system reboot per the KEV requiredAction.' Cited as insufficient on this entry: NIST SP 800-53 Rev 5 SI-2 'Flaw Remediation' and EU NIS2 Directive Art. 21 'Security of network and information systems'.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-network-security"
+        ]
+      }
+    ]
   },
   "CVE-2024-11182": {
     "name": "MDaemon Email Server Cross-Site Scripting (XSS) Vulnerability",
@@ -33136,7 +33378,41 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "human_researcher",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-134",
+        "name": "DEVICE-MANAGEMENT-GATEWAY-CONFIG-ENDPOINT-AUTH-INPUT-NEUTRALIZATION",
+        "description": "UniFi OS is the management plane running on the device, and this defect is its authorization decision and its routing decision being taken from two different forms of the same request: the packet has the auth-gateway classifying the raw, percent-encoded URI as a public endpoint while the normalized/decoded request is routed to an authenticated internal service, so encoded '..%2f' sequences smuggle the request out of its intended directory into arbitrary file read and write on the underlying OS. Bound to this product the control has two halves. The endpoint half is that the nginx/unifi-core flow canonicalizes the URI once and takes both the public-versus-authenticated classification and the routing decision from that same normalized form, and that any resolved filesystem path is verified to remain inside the directory the handler is entitled to serve before a file handle is opened — not by stripping traversal sequences from the request string, since the whole defect is that one layer acts on what another has already decoded differently. That half is what the vendor update establishes; this control states the property to verify, it does not implement it. The operator half is that no UniFi OS device is left with its management interface answering from a network with no management role, because the packet's only stated access requirement is network access to the device with no authentication — reachability is the entire exploit precondition. Precondition on that half, which is where it gets over-claimed: restricting reachability bounds who can send the request, it does not repair the URI-handling split, and it is unavailable where the management interface must answer the network the device administers — any host already inside the permitted segment satisfies the packet's access requirement in full. Distinguishing test: from a segment with no management role, send a public UniFi OS endpoint prefix carrying encoded traversal at a staging device and confirm it is refused before any file is read or written; an estate that passes device-account and administrator-MFA attestations still has this path open, because no credential is ever presented on it.",
+        "evidence": "Packet: 'Ubiquiti UniFi OS Path Traversal Vulnerability', CWE-22. Vector: 'The UniFi OS web management interface is reachable by any actor with network access to the device, with no authentication required... the auth-gateway classifies the raw, percent-encoded URI as a public endpoint while the normalized/decoded request is routed to an authenticated internal service, so encoded \"..%2f\" sequences smuggle the request out of its intended directory and yield arbitrary file read/write on the underlying OS', with the KEV description noting accessed files can be 'manipulated to access an underlying account'. cvss 10; rwep_score 79; poc_available true; active_exploitation 'confirmed'; CISA KEV-listed 2026-06-23. patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' Citing gaps include NIST-800-53-SC-7 (Boundary Protection), UK-CAF-B4 and NIS2-Art21-network-security.",
+        "gap_closes": [
+          "NIST-800-53-SC-7",
+          "NIS2-Art21-network-security",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-030",
+        "name": "PERIMETER-DEVICE-ZERODAY-SLA-TIER",
+        "description": "The packet describes a management interface that any actor with network access reaches without authenticating, on a device that fronts the network it administers — so the ordinary appliance-patch window does not apply, and the SLA tier for this CVE is vendor firmware deployed within hours of the 2026-06-23 KEV listing, or the management interface isolated from every segment with no management role until it is. The measurement has to be the firmware each unit is actually running, not an update queued, scheduled or downloaded to it, because a device that has taken no restart into the fixed image is still serving the vulnerable request flow. Two things set the priority above the generic patch queue. The packet records CVSS 10, a public PoC and confirmed exploitation with no authentication required, and it names this flaw being combined in the wild with the access-control bypass CVE-2026-34908 and the package-update command injection CVE-2026-34910 to reach an unauthenticated root RCE chain used to drop a Mirai/Gafgyt-derived botnet — so a unit is not remediated by anything that closes this traversal alone; the remediation state to track is the device reaching a firmware level on which all three named defects are closed, tracked as one device outcome rather than three separate vulnerability tickets that can each be marked done at different times. Precondition: the packet registers no live-patch path, so there is no in-place mitigation that removes the defect ahead of the firmware — isolating the management interface is the interim state and it holds only for the segments it actually excludes, which on a device whose management interface must remain reachable to administer the network may be none.",
+        "evidence": "Packet: CISA KEV-listed 2026-06-23; active_exploitation 'confirmed'; cvss 10; rwep_score 79; poc_available true; privileges required none — 'reachable by any actor with network access to the device, with no authentication required'. 'In the wild it is combined with the access-control bypass (CVE-2026-34908) and the package-update command injection (CVE-2026-34910) to reach an unauthenticated root RCE chain used to drop a Mirai/Gafgyt-derived botnet.' patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' Citing gaps include AU-Essential-8-Patch, ISO-27001-2022-A.8.8 and NIST-800-53-SI-2, all cadence-scored.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "A UniFi OS device whose management interface answered untrusted requests while on a vulnerable build has to be dispositioned as compromised rather than simply updated. The packet gives the two reasons directly. The primitive is arbitrary file read and write on the underlying OS reached with no credential, so a successful exploit leaves no authentication event to find later, and the firmware update restores the request-handling behaviour while removing nothing that was already written through the write primitive. And the KEV description records that the accessed files can be 'manipulated to access an underlying account', so account access obtained through the read/write path stays valid across the update unless the credential is rotated — while the in-the-wild chain the packet names ends in a Mirai/Gafgyt-derived botnet dropped on the device itself. The response default for an exposed unit is therefore to rebuild it from vendor firmware to a factory-clean state rather than updating in place, re-provision its configuration from a known-good copy rather than preserving the running one, rotate every credential the device held or fronted — its local administrative accounts and anything recoverable from files on it — and review the device's outbound traffic for the botnet activity the packet describes. The distinguishing test is whether the remediation record shows a rebuild and credential rotation alongside the firmware version; a version check reporting the fixed build is exactly the evidence that reads clean while attacker-written content and attacker-held accounts persist, because neither is versioned.",
+        "evidence": "Packet: CWE-22 path traversal yielding 'arbitrary file read/write on the underlying OS' with 'no authentication required'; 'The KEV description notes the accessed files can be \"manipulated to access an underlying account,\" giving credential/account access'; 'In the wild it is combined with the access-control bypass (CVE-2026-34908) and the package-update command injection (CVE-2026-34910) to reach an unauthenticated root RCE chain used to drop a Mirai/Gafgyt-derived botnet.' active_exploitation 'confirmed'; poc_available true; CISA KEV-listed 2026-06-23; cvss 10; rwep_score 79. patch_available true with live_patch_available false — the vendor update is the remediation this control constrains from being applied in place on an already-exploited unit.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2"
+        ]
+      }
+    ]
   },
   "CVE-2026-34908": {
     "name": "Ubiquiti UniFi OS Improper Access Control Vulnerability",
@@ -33280,7 +33556,39 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-129",
+        "name": "ENTERPRISE-ERP-MANAGEMENT-FUNCTION-AUTHENTICATION",
+        "description": "The defect the packet records is CWE-306 on Splunk Enterprise's PostgreSQL sidecar recovery endpoints: the web application on port 8000 proxies backup and restore file operations to the local PostgreSQL service on port 5435, and those endpoints perform no authentication at all, so any network-reachable caller invokes them. Bound to this product, the control means every recovery, backup and restore function behind the port-8000 web tier authorizes its own caller before it runs, rather than being reachable because it sits behind a front end assumed to have authenticated somebody; and the caller-supplied 'database' parameter is validated as a database identifier instead of being passed through to pg_dump and pg_restore, where the packet shows it becomes a PostgreSQL connection string the attacker can override with his own hostaddr to make the server restore an attacker-supplied dump. This is why the least-privilege control cited as insufficient on this entry never touches the path: the attacker holds no Splunk account, so per-account privilege scoping is never consulted and an access-control attestation covering Splunk roles and capabilities passes cleanly while the endpoints stay open to an unauthenticated caller. Distinguishing test: from a segment with no operational need for Splunk administration, send unauthenticated requests to each recovery endpoint on a staging instance and confirm each is refused before any pg_dump or pg_restore process starts. Precondition, and this is where the control is normally over-claimed: endpoint-side authorization and parameter validation are properties the vendor update establishes — this control states what to verify, it does not implement it. Until that update lands the operator's only lever is limiting who can reach port 8000, and that is a weak lever here, because port 8000 is the product's own web application: restricting it to an administrative segment also removes it from the users the deployment exists to serve, and it does nothing against a caller already inside the permitted segment.",
+        "evidence": "Packet vector: the Splunk Enterprise web application (port 8000) exposes PostgreSQL sidecar recovery endpoints proxied to the local PostgreSQL service on port 5435 that 'perform no authentication, so any network-reachable, unauthenticated attacker can invoke backup/restore file operations (CWE-306)'; 'The attacker controls the database parameter passed to pg_dump/pg_restore, enabling PostgreSQL connection-string injection (e.g. overriding hostaddr) to force the server to restore an attacker-supplied database dump.' CVSS 9.8, RWEP 87, poc_available true, active_exploitation 'confirmed', cisa_kev true with kev_date 2026-06-18. NIST-800-53-AC-6 (Least Privilege) and UK-CAF-B4 (System security) are both recorded among the framework gaps citing this CVE.",
+        "gap_closes": [
+          "NIST-800-53-AC-6",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The packet supplies every input this clock keys on at once: CVSS 9.8, RWEP 87, a public PoC, confirmed in-the-wild exploitation, and a KEV listing dated 2026-06-18 against a path the packet itself calls pre-auth remote code execution. For this CVE the requirement is that the Splunk Enterprise vendor update run on the KEV clock from that listing rather than being folded into the estate's normal application-patching cadence, with completion measured per instance by the version actually in service rather than by an approval recorded in a change queue. That is the half the cited framework controls do not carry — a patching programme scored on cadence and a technical-vulnerability clause that leaves timescales to the organization both pass their attestations on their own schedule while an unauthenticated network caller still reaches the recovery endpoints. Preconditions the packet sets on the interim window, which is where this control gets over-stated: live_patch_available is false and the packet states remediation is the vendor update plus the named compensating controls until it lands, so there is no mechanism that closes this without taking each instance through the update, and every interim measure is a holding action rather than a fix. Restricting who can reach the port-8000 web tier bounds the population that can send the request; it does not close the endpoint, it does nothing against a caller inside the permitted segment, and it is unavailable wherever that tier has to stay reachable for the deployment's normal use.",
+        "evidence": "Packet: CVSS 9.8, RWEP 87, poc_available true, active_exploitation 'confirmed', cisa_kev true with kev_date 2026-06-18. Vector describes an unauthenticated attacker invoking backup/restore operations and the chain escalating 'to pre-auth remote code execution'. patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' Citing gaps include AU-Essential-8-Patch (Patch operating systems), ISO-27001-2022-A.8.8 (Management of technical vulnerabilities) and NIS2-Art21-vulnerability-management (Vulnerability handling).",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-078",
+        "name": "ENDPOINT-MGMT-DEPLOYMENT-CHANNEL-INTEGRITY",
+        "description": "The primitive the packet describes is an arbitrary file create and truncate as the splunk user, reached through lo_export inside an attacker-supplied database dump, and the packet names what the chain does with it: overwriting executable Python scripts in Splunk's application directory, which run as the splunk user when subsequently invoked. That makes Splunk's application directory a code-execution channel rather than application data, and it is the part of this CVE the vendor update does not address — the update closes the unauthenticated endpoint, but a Python file already overwritten through it survives the upgrade and still runs at its next invocation. The requirement is file-integrity monitoring over Splunk's application directory and every path the instance executes from, alerting on writes that do not correspond to a sanctioned administrator action or a vendor update, plus triage of any instance that was network-reachable during the exposure window against a known-good copy rather than closing it on the patch record. The alert has to key on the behaviour the packet documents — a write or truncate to an executable file under the Splunk application directory, performed by the splunk user, outside a change window — because nothing else in this chain emits a signal: the write arrives through the product's own PostgreSQL restore path, so there is no crash, no new binary dropped by an unusual parent, and no exploit-tool signature; a rule watching for process crashes or named tooling would miss an attempt behaving exactly as the packet describes. Preconditions: this depends on a baseline of the application directory captured before the write and on the monitoring records living off-host, because the attacker holds write access as the splunk user — the same identity a locally-stored baseline sits under. An instance with no prior baseline has to be compared against a clean installation of the same version, not against itself. And detection does not prevent the write; it bounds the window to alert-and-response time.",
+        "evidence": "Packet vector: the malicious SQL uses 'lo_export to write arbitrary files as the splunk user, giving an arbitrary file create/truncate primitive. The chain escalates to pre-auth remote code execution by overwriting executable Python scripts in Splunk's application directory, which run as the splunk user when subsequently invoked.' active_exploitation 'confirmed', poc_available true, cisa_kev true with kev_date 2026-06-18. patch_available true; live_patch_available false. NIST-800-53-SI-2 (Flaw Remediation) is recorded among the framework gaps citing this CVE.",
+        "gap_closes": [
+          "NIST-800-53-SI-2"
+        ]
+      }
+    ]
   },
   "CVE-2026-48907": {
     "name": "Widget Factory Joomla Content Editor Improper Access Control Vulnerability",
@@ -33493,7 +33801,31 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-134",
+        "name": "DEVICE-MANAGEMENT-GATEWAY-CONFIG-ENDPOINT-AUTH-INPUT-NEUTRALIZATION",
+        "description": "Catalyst SD-WAN Manager (formerly SD-WAN vManage) is the management plane the packet names, and the defect sits in its upload handling: the network-reachable web UI/API exposes a file-upload handler that does not validate user-supplied file paths during upload processing, so ../ sequences in a crafted HTTP request select where a privileged write lands. Bound to this product, the control means every file-accepting endpoint on SD-WAN Manager normalizes the caller-supplied destination to an absolute canonical path and rejects it when the resolved result leaves the intended upload directory, with that check performed before the write executes rather than by filtering the request string; and that the endpoint authorizes the caller against what that particular account is entitled to write instead of treating an authenticated session as sufficient. The packet's access requirement is what makes the second half load-bearing on this entry rather than the first alone: the attacker holds only a low-privileged, single-task user account, so on this appliance every account that can authenticate to the web UI/API is in exploit terms a root account on the control plane, and an authorization decision that ends at 'the session is valid' never consults that difference. Distinguishing test: on a staging SD-WAN Manager, authenticate as a single-task, low-privileged user and send each file-accepting endpoint an upload whose destination path resolves outside the intended upload directory; confirm it is refused before anything is written — an instance whose user-role definitions audit cleanly still performs the write if the handler resolves the caller-supplied path. Precondition, and this is where the control is usually over-claimed: the endpoint-side path validation is a property the vendor update establishes, so this control states what to verify and does not implement it. The packet registers no live-patch path for this product class, so until that update and its restart land the only operator-side lever is restricting which segments can reach the web UI/API — which bounds who can present the upload, leaves the handler fully exploitable to anything inside the permitted segment, and is unavailable wherever the interface must stay reachable for operators to manage the fabric.",
+        "evidence": "Packet vector: the web UI/API of Cisco Catalyst SD-WAN Manager (formerly SD-WAN vManage) is network-reachable and exposes a file-upload handler that fails to validate user-supplied file paths during upload processing; an authenticated attacker holding only a low-privileged, single-task user account sends a crafted HTTP request with path-traversal (../) sequences and obtains an arbitrary file create/overwrite primitive anywhere on the underlying operating system. CWE-22. CISA KEV-listed 2026-06-15 with active_exploitation confirmed and poc_available true; RWEP 77 against CVSS 6.5. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.'",
+        "gap_closes": [
+          "NIST-800-53-SC-7",
+          "NIS2-Art21-network-security",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-037",
+        "name": "FLEET-COMPROMISE-IR-PLAYBOOK",
+        "description": "This appliance is a fleet control plane in the sense the control governs — the packet states that compromising it yields full compromise of the SD-WAN control plane that pushes policy and routing to every managed edge device — and the packet's escalation path is an attacker overwriting a privileged configuration, cron, or startup file to reach root. That is exactly the case where the patch record and the incident are different questions: the vendor update closes the traversal but reverts nothing already written through it, and a cron or startup file planted before remediation survives the upgrade and keeps running as root. Scoped to what this packet establishes, the playbook for any SD-WAN Manager that was reachable and unpatched during the window opened by the 2026-06-15 KEV listing is: compare the appliance's configuration, cron, and startup content against a known-good baseline instead of accepting the upgraded build as closure; treat every credential and key the controller holds or distributes as exposed and rotate it, since root on the appliance reaches all of it; review the policy and routing pushed to managed edge devices during that window for changes with no corresponding operator action; and set in advance the quarantine criteria for an edge device that accepted configuration from a suspect controller. Precondition: this is post-compromise triage and not prevention — it does nothing to stop the traversal, and it depends on a baseline captured before the exposure window, which is the part most estates do not hold. Where no such baseline exists the comparison cannot be made and the terminal action is rebuilding the controller from vendor media and re-pushing known-good policy, not inspecting it in place.",
+        "evidence": "Packet vector: the arbitrary file create/overwrite primitive is chained by overwriting a privileged configuration, cron, or startup file, or planting executable content, to elevate to root, 'yielding full compromise of the SD-WAN control plane that pushes policy and routing to every managed edge device'. active_exploitation confirmed, CISA KEV-listed 2026-06-15, poc_available true, RWEP 77. patch_available true with live_patch_available false; live_patch_notes state remediation is the vendor update plus the named compensating controls until it lands.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8"
+        ]
+      }
+    ]
   },
   "CVE-2025-34028": {
     "name": "Commvault Command Center Path Traversal Vulnerability",
@@ -34508,7 +34840,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-145",
+        "name": "ACTIVELY-EXPLOITED-OS-LPE-REMEDIATION-WINDOW-ENFORCEMENT",
+        "description": "clfs.sys ships with Windows and the packet's path into it runs through standard CLFS APIs operating on a crafted base-log (.blf) file, available to any local user — so the affected population is every Windows host in the estate, not a role-scoped subset, and there is no feature to switch off or service to stop that takes the driver out of a normal build. For this CVE the control means driving the Microsoft update that carries the fix across that whole population on the clock that opened with the 2025-04-08 KEV listing rather than folding it into the next monthly rollup, with completion measured by each host's installed build against the fixed build for its SKU rather than by 'approved' or 'downloaded' in the management console. The packet registers no live-patch path for this product class and states that remediation is the vendor update plus compensating controls until it lands, so a host that has taken the update but not restarted still runs the vulnerable driver and must be counted as exposed. Priority follows the packet rather than the 7.8 CVSS: from the SYSTEM token this flaw produces, the packet's chain injects into winlogon.exe, dumps LSASS for credentials and launches RansomEXX from dllhost.exe, so a host where the escalation succeeds is a credential-theft and domain-wide ransomware event rather than an endpoint finding. Enumerate multi-user hosts first — session hosts, shared workstations, build agents, jump boxes — because the precondition the flaw needs, a local low-privileged authenticated user running code, is their normal operating state; they are also where the restart is most likely to be deferred because taking it evicts logged-on users, and a deferral recorded as patched is the specific way this remediation goes wrong. The privilege half of the estate's control set cannot substitute here: the attacker already holds a legitimate low-privileged account and the boundary that fails is inside the kernel driver, so tightening account privilege leaves the path intact while its attestation reads clean.",
+        "evidence": "Packet fields for this entry: name 'Microsoft Windows Common Log File System (CLFS) Driver Use-After-Free Vulnerability', CWE-416. Vector and attack_vector: 'A local, low-privileged authenticated user reaches the Common Log File System (clfs.sys) kernel driver through standard CLFS APIs operating on a crafted base-log (.blf) file, triggering a use-after-free (CWE-416) on a freed CLFS structure. The actor leaks kernel addresses to user mode via NtQuerySystemInformation, then weaponizes the dangling pointer together with the RtlSetAllBits primitive to overwrite the exploiting process's token privileges field with 0xFFFFFFFF, granting all privileges and effectively SYSTEM. With SYSTEM, the chain injects into winlogon.exe, dumps LSASS for credentials, and launches RansomEXX from dllhost.exe — turning a post-compromise foothold into domain-wide ransomware impact.' cisa_kev true with kev_date 2025-04-08, active_exploitation confirmed, poc_available true, cvss 7.8, rwep_score 81. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' AU-Essential-8-Patch (Patch operating systems), ISO-27001-2022-A.8.8 (Management of technical vulnerabilities), NIST-800-53-SI-2 (Flaw Remediation) and NIS2-Art21-vulnerability-management (Vulnerability handling) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-003",
+        "name": "KERNEL-EXPLOITATION-DETECTION",
+        "description": "On a Windows estate this control's telemetry is ETW and endpoint-agent event data rather than the auditd or eBPF form it takes on Linux hosts, and the packet describes the exploit precisely enough to key on it rather than on a stand-in. The signature is a same-process privilege transition: a low-privileged process creates or writes a CLFS base-log (.blf) file and drives the CLFS APIs against it, calls NtQuerySystemInformation to leak kernel addresses into user mode, and then keeps running under the same PID with its token privileges field set to all bits. Alert on that transition — a process whose token acquires SYSTEM-equivalent privileges with no corresponding logon, service start, or impersonation event — and correlate it with the follow-on the packet names: code injection into winlogon.exe, a handle opened to lsass.exe with memory-read access, and dllhost.exe spawning a child that is not a COM surrogate workload. The correlation is what makes the rule usable: .blf files are written by legitimate software and NtQuerySystemInformation is called by ordinary tooling, so either signal alone is noise, while the sequence inside a short window is not. Note what would miss an attempt behaving exactly as the packet describes: nothing crashes, no driver is loaded, no service is created, and no new SYSTEM-owned process appears — the escalation happens inside an already-running ordinary process — so rules keyed on crashes, driver-load events, new elevated processes, or named exploit-tool signatures see nothing. Preconditions. This requires endpoint telemetry that was already being collected and forwarded off-host before the attempt, because the moment the token flips the actor holds SYSTEM on that host and can stop or blind a locally-buffered agent, leaving an on-host-only trail under the attacker's control at exactly the point it matters. And detection prevents nothing: it bounds the interval between the escalation and the LSASS dump and ransomware launch to alert-and-response time, during the window before the vendor update and its restart land. It is the interim posture, not the remediation.",
+        "evidence": "Packet fields for this entry: the vector describes the full chain — a crafted base-log (.blf) file driving a use-after-free in clfs.sys, kernel addresses leaked to user mode via NtQuerySystemInformation, the RtlSetAllBits primitive used to overwrite the exploiting process's token privileges field with 0xFFFFFFFF for effective SYSTEM, then injection into winlogon.exe, an LSASS credential dump, and RansomEXX launched from dllhost.exe. cisa_kev true, kev_date 2025-04-08, active_exploitation confirmed, poc_available true, cvss 7.8, rwep_score 81. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands' — the packet itself names compensating controls as what holds until the update. UK-CAF-B4 (System security) is recorded as a citing gap against this entry, and its stated shortfall is precisely this: an interim compensating-control posture for an actively-exploited local privilege escalation with a public PoC is not enumerated as distinct from applying the vendor patch.",
+        "gap_closes": [
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2025-30406": {
     "name": "Gladinet CentreStack and Triofox Use of Hard-coded Cryptographic Key Vulnerability",
@@ -34901,7 +35256,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "human_researcher",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "Sitecore CMS and Experience Platform (XP) is a content-management web tier, and the packet's path into it is one HTTP POST: a ysoserial.net TypeConfuseDelegate gadget supplied as the __CSRFTOKEN parameter to any page the Sitecore.Security.AntiCsrf module guards, executing inside ObjectStateFormatter before the module ever compares that value to __CSRFCOOKIE. With a public PoC and confirmed exploitation there is no attack complexity buying time, so the clock runs from the 2025-03-26 KEV listing rather than from the next CMS release train. The packet records a vendor patch with no live-patch path and states that until it lands the remaining measures are this entry's named compensating controls — so remediation means getting every Sitecore instance onto the vendor's fixed build, with completion measured by reading the running build off each instance rather than from a deployment ticket. Two preconditions bound what meeting this SLA actually buys. First, this is the 9.x post-authentication variant: the packet records that exploitation requires a valid authenticated session, so restricting who can reach the CMS narrows the caller population during the window before the fix but closes nothing against anyone already holding a Sitecore credential, including a low-privilege content account — network restriction is a holding measure, not a substitute for the build. Second, an SLA met from the listing forward settles nothing about instances that were already reachable on an affected build, and the packet's chain ends in a reverse shell, which is a triage question rather than a patching one.",
+        "evidence": "Packet fields for CVE-2019-9875 ('Sitecore CMS and Experience Platform (XP) Deserialization Vulnerability (post-authentication)'): cwe_refs CWE-502; cisa_kev true; kev_date 2025-03-26; active_exploitation 'confirmed'; cvss 8.8; rwep_score 68; poc_available true; patch_available true; live_patch_available false; live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.'; ai_discovered false. Packet attack_vector: the Sitecore.Security.AntiCsrf module 'deserializes the attacker-supplied __CSRFTOKEN HTTP POST parameter via ASP.NET's ObjectStateFormatter BEFORE comparing it to the __CSRFCOOKIE value', the formatter 'is instantiated with a null _page' so 'it performs no MAC/signature validation on the LOSFormatter stream', an attacker 'supplies a ysoserial.net TypeConfuseDelegate gadget encoded for the ObjectStateFormatter', and 'For the 9.x branch this CVE requires a valid authenticated session (PR:L)'. Cited as insufficient on this entry: ASD Essential Eight 'Patch operating systems', ISO/IEC 27001:2022 A.8.8 'Management of technical vulnerabilities', UK NCSC CAF B4 'System security'.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet's path does not stop at deserialization: code runs in-process as the IIS application-pool identity, the gadget's PowerShell stager pulls a second stage and opens a reverse shell, giving arbitrary command execution on the web server and access to the underlying filesystem. Sitecore is an application tier rather than an appliance, but it occupies the position this control governs, and with exploitation confirmed a Sitecore instance that was reachable by a credential holder on an affected build during the exposure window is a triage subject rather than a patch ticket. The vendor build removes the deserialization sink and removes nothing the second stage wrote, and it revokes nothing the application-pool identity could read — connection strings and configuration secrets held by that instance stay valid across the upgrade. The requirement for this CVE is therefore: export configuration and content for review, rebuild the web tier from a known-good image on the fixed build instead of upgrading the live instance, and rotate every secret reachable from the application-pool identity. The packet supplies one triage anchor directly, and it is the one most likely to be misread: because the gadget executes 'regardless of the subsequent cookie-mismatch error', a successful exploitation leaves a CSRF cookie-versus-parameter mismatch entry in the application's own logs — a line that reads as the security control working, written after the code has already run. Precondition and honest limit: the packet records the end state as a reverse shell and names no implant, tooling or artifact, so the trigger for this runbook is reachability during the window, not an observed indicator; and if the triage is deferred, a later rebuild taken from a baseline captured after the exposure reproduces whatever was written rather than removing it.",
+        "evidence": "Packet attack_vector for CVE-2019-9875: the gadget 'executes during deserialization regardless of the subsequent cookie-mismatch error'; 'code runs in-process as the IIS app-pool identity, after which the gadget's PowerShell stager pulls a second stage and opens a reverse shell, giving arbitrary command execution on the web server and access to the underlying filesystem'; an attacker must 'reach a guarded Sitecore endpoint (e.g. CreateNewUser.aspx)'. active_exploitation 'confirmed'; cisa_kev true with kev_date 2025-03-26; poc_available true; patch_available true — the fix exists, which is what makes 'patched' the compliance verdict this control has to override; live_patch_available false. Cited as insufficient on this entry: NIST SP 800-53 Rev 5 SI-2 'Flaw Remediation' and EU NIS2 Directive (2022/2555) 'Vulnerability handling'.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      }
+    ]
   },
   "CVE-2019-9874": {
     "name": "Sitecore CMS and Experience Platform (XP) Deserialization Vulnerability (unauthenticated)",
@@ -34956,7 +35334,42 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "human_researcher",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-125",
+        "name": "SERVICE-INTERNAL-PROTOCOL-DESERIALIZATION-HARDENING",
+        "description": "For Sitecore the trust boundary is inside the AntiCSRF module's own request handling: the packet has Sitecore.Security.AntiCSRF taking the value of the __CSRFTOKEN HTTP POST parameter and deserializing it with .NET BinaryFormatter without type restriction, and on affected 8.x and earlier builds doing so before authentication is enforced — so an anonymous request rebuilds an arbitrary object graph in the IIS worker process. Two properties have to hold for this product. The parameter is a CSRF token, so what arrives in it must be constrained to the concrete, primitive shape that flow legitimately carries and must never be permitted to instantiate types the module does not need — an unrestricted formatter is not a deserializer with a weak filter, it is no filter at all, which is why a published gadget chain is sufficient. And the authentication decision must precede the deserialization, because ordering is the second half of the defect: reversing it removes the anonymous reach even where the sink remains. Both are properties the vendor update establishes; this control states what to verify, it does not implement it. Precondition, and this is where the control is normally over-claimed: its network-restriction half is largely unavailable here. The packet says the sink is reachable 'on any AntiCSRF-protected page', so binding the surface to a management segment only helps an instance whose AntiCSRF-protected pages are all administrative — where such a page is part of the publicly served site, no segmentation removes the path and the update is the only closure. Constraining the IIS application-pool identity's rights and what it can reach bounds what a successful gadget chain does next, but it does not stop the deserialization, and the packet records the outcome as command execution in that identity's context with no separate privilege-escalation step needed.",
+        "evidence": "Packet: 'Sitecore CMS and Experience Platform (XP) Deserialization Vulnerability (unauthenticated)', CWE-502. Vector: 'The Sitecore.Security.AntiCSRF module accepts the value of the HTTP POST parameter __CSRFTOKEN and deserializes it with .NET BinaryFormatter without type restriction, reachable over the network on any AntiCSRF-protected page. On affected 8.x and earlier builds the deserialization occurs before authentication is enforced, so an unauthenticated remote attacker (CVSS 9.8, AV:N/PR:N) can submit a crafted serialized object... arbitrary command execution in the context of the IIS application pool identity hosting Sitecore, yielding full server compromise without needing a separate privilege-escalation step.' CISA KEV-listed 2025-03-26; active_exploitation 'confirmed'; cvss 9.8; rwep_score 68; poc_available true; patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.'",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "A Sitecore instance that served AntiCSRF-protected pages to an untrusted network on an affected 8.x-or-earlier build has to be dispositioned as compromised rather than upgraded in place. Two facts from the packet make patch-in-place the wrong default. First, the deserialization occurs before authentication is enforced, so a successful exploit generates no authentication event to find afterwards — the site's login records will look clean for the intrusion. Second, the outcome is arbitrary command execution as the IIS application-pool identity with access to the underlying filesystem, and an .aspx web shell or scheduled task placed through that primitive survives the vendor upgrade untouched and keeps answering after it. The response default is therefore to treat the deployed web root and every writable media and upload directory as attacker-modifiable: diff the live tree against the known-good deployment artefact, rebuild the server from source of truth at the fixed version instead of upgrading in place, and rotate what the box held — the application-pool and service accounts, the Sitecore administrator accounts, the database connection strings in the instance configuration, and any key or certificate material stored on it. The distinguishing test is whether the remediation record carries a file-integrity comparison of the deployed tree alongside the version change; an upgrade ticket on its own cannot show that a file written through this path was removed, and a version scan reporting the fixed build is exactly the evidence that passes while an implant stays resident. Scope the exposure window honestly: the packet pairs a public PoC and confirmed exploitation with a 2019 CVE identifier carried into a 2025 KEV listing, so an instance left on an affected build was reachable across that whole interval — 'we upgraded when it reached KEV' bounds the remediation date, not the intrusion window.",
+        "evidence": "Packet: unauthenticated CWE-502 deserialization where 'the deserialization occurs before authentication is enforced', giving 'arbitrary command execution in the context of the IIS application pool identity hosting Sitecore, yielding full server compromise without needing a separate privilege-escalation step', with the 9.x-branch companion entry noting access to the underlying filesystem. active_exploitation 'confirmed'; poc_available true; CISA KEV-listed 2025-03-26 against a CVE identifier dated 2019; cvss 9.8; rwep_score 68; patch_available true, and live_patch_available false — remediation is the vendor update, which is precisely the patch-in-place action this control constrains.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "This entry is the case where a cadence-driven vulnerability programme has no event to fire on at all: the vendor fix long predates the listing, so nothing in a monthly or quarterly cycle distinguishes this from any other aged advisory, and the 2025-03-26 KEV listing is the only trigger that reflects the real risk state. The requirement for this product is that Sitecore instances on affected 8.x-and-earlier builds move to the vendor-fixed release on a clock started by that listing rather than on a CMS-upgrade roadmap, and that the SLA is measured by the instance actually serving requests on the fixed build — not by an approved change ticket or a scheduled upgrade window, which for a major Sitecore version step is where the delay accumulates. Precondition, stated plainly because this is where the SLA is usually reported as met: the packet records no live-patch path, so there is no in-product interim fix that removes the sink — the compensating controls named on the entry are what hold until the upgrade lands, and where the upgrade cannot be completed inside the clock, the only remaining lever is removing the affected instance's AntiCSRF-protected pages from untrusted network reach. That lever is unavailable for an instance whose AntiCSRF-protected pages are part of the public site, and for those an unmet clock is exposure to an unauthenticated, publicly-exploited pre-auth RCE rather than an accepted risk with a compensating control behind it.",
+        "evidence": "Packet: CISA KEV-listed 2025-03-26 with active_exploitation 'confirmed' and poc_available true on a CVE identifier dated 2019; cvss 9.8, rwep_score 68; privileges required none — 'an unauthenticated remote attacker (CVSS 9.8, AV:N/PR:N) can submit a crafted serialized object'. patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' The entry cites AU-Essential-8-Patch, ISO-27001-2022-A.8.8, NIST-800-53-SI-2 and NIS2-Art21-vulnerability-management — all cadence-based — as insufficient.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      }
+    ]
   },
   "CVE-2017-12637": {
     "name": "SAP NetWeaver Directory Traversal Vulnerability",
@@ -35682,7 +36095,30 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_research",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-145",
+        "name": "ACTIVELY-EXPLOITED-OS-LPE-REMEDIATION-WINDOW-ENFORCEMENT",
+        "description": "The packet places this in the Win32k kernel subsystem and describes an attacker who already holds low-privilege local code execution winning a race to turn a use-after-free on the W32PROCESS structure into a controlled kernel write/execute primitive and a SYSTEM token — the escalation executes in kernel code, below every account boundary an endpoint estate is audited on. For this CVE the control means the Windows update carrying the fix is driven on the clock the 2025-03-11 KEV listing opened rather than folded into the next monthly rollup, with completion measured by the build each host is actually running rather than by 'approved' or 'downloaded' in the management console. The population to enumerate first is the one the packet names — hosts on affected builds predating 1809 — not the Windows estate at large, which the packet does not implicate. The prioritisation half is the load-bearing one here, because every routine escalation trigger points the wrong way: CVSS 7.0, a vendor rating of Important rather than Critical that the packet attributes to the race that must be won, and poc_available false. A program that accelerates on severity band or on public-exploit availability leaves this on the ordinary cycle while the packet records exploitation as confirmed and the flaw already wired into a delivery chain behind the PipeMagic backdoor. Precondition: the packet registers no live-patch path, so nothing removes the defect before the update lands — the compensating controls named on the entry are a holding measure for that window, not a substitute for the update. And because this is a second-stage escalation reached only after an initial foothold, updating a host that was already exploited closes the escalation path without removing the foothold that reached it or undoing what was done with SYSTEM; such a host belongs on the incident path, not on the patch record.",
+        "evidence": "Packet: 'Microsoft Windows Win32k Use-After-Free Vulnerability', CWE-416. Vector: 'An attacker who already has low-privilege local code execution on an affected, pre-1809 Windows build triggers a use-after-free in the Win32k kernel subsystem: by winning a race condition, the W32PROCESS structure is dereferenced one more time than its reference count allows... escalating the process token to SYSTEM', the bug 'reached only after initial foothold (here, via the PipeMagic backdoor)', a 'second-stage local privilege escalation' with follow-on 'credential theft, persistence, and ransomware staging', and 'The high attack complexity reflects the race that must be won, which is why Microsoft rated it Important rather than Critical.' CISA KEV-listed 2025-03-11; active_exploitation 'confirmed'; cvss 7; rwep_score 59; poc_available false; patch_available true; live_patch_available false with live_patch_notes 'No live-patch path for this product class; remediation is the vendor update plus the named compensating controls until it lands.' The entry cites AU-Essential-8-Patch, ISO-27001-2022-A.8.8, NIST-800-53-SI-2, NIS2-Art21-vulnerability-management and UK-CAF-B4 as insufficient.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-003",
+        "name": "KERNEL-EXPLOITATION-DETECTION",
+        "description": "This is a Windows kernel-mode flaw, so the rule is built on host kernel/process telemetry rather than on auditd or eBPF — and the packet describes the behaviour precisely enough to key on it. The exploit has to win a race, and a race is repetitive: a low-privileged user-mode process drives the same Win32k path in a tight loop, many attempts in a short window from one process and its threads, before the timing lands. The rule therefore keys on that shape, paired with the outcome the packet names: that same already-running low-privileged process afterwards holding a SYSTEM token — a token and integrity-level transition inside a live process with no service start, no consented elevation and no scheduled task to account for it — followed by SYSTEM-context activity from a process whose parent chain is an ordinary user session. Both halves are needed: repeated Win32k calls alone are noisy, and a SYSTEM-context process alone is normal; it is the pairing inside a short window that separates this from either. Note what will not see it. Do not key on a crash or bugcheck: the packet's chain grooms the pool to reclaim the freed allocation with attacker-controlled data and continues into credential theft, persistence and ransomware staging, which requires the host to keep running. Do not key on a driver load or a new signed binary — nothing is loaded, the vulnerable code is the operating system's own Win32k, and the exploit runs from a foothold already on the host. Do not key on a public exploit's signature: the packet records poc_available false, so there is no public tool artefact to match. Precondition: this needs kernel-level process and token telemetry already being collected and shipped off-host before the attempt; a rule written afterwards against telemetry nobody was gathering produces nothing. Detection does not prevent the escalation and does not remediate the flaw — it bounds the window between the KEV listing and the update, and because the packet places the bug after an initial foothold, an alert here is an incident already in progress: the response is host isolation and hunting the delivery-stage implant, not only applying the update.",
+        "evidence": "Packet: CWE-416 use-after-free in the Win32k kernel subsystem, exploited 'by winning a race condition' and turned into 'a controlled kernel write/execute primitive, escalating the process token to SYSTEM'; reached 'only after initial foothold (here, via the PipeMagic backdoor)' and used for 'credential theft, persistence, and ransomware staging'. active_exploitation 'confirmed'; CISA KEV-listed 2025-03-11; poc_available false; patch_available true; live_patch_available false, so every affected host carries an exposure window until the vendor update lands.",
+        "gap_closes": [
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2026-45659": {
     "name": "Microsoft SharePoint Server Deserialization of Untrusted Data Vulnerability",
@@ -35737,7 +36173,22 @@
     },
     "ai_discovered_zeroday": false,
     "ai_discovery_source": "vendor_disclosure",
-    "ai_assist_factor": "none"
+    "ai_assist_factor": "none",
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The packet's timing is the whole argument on this entry: KEV listing 2026-07-01 with a due date of 2026-07-04 — three days — against SharePoint Server, which in most estates is patched inside a scheduled farm-maintenance window measured in weeks. Applied to this CVE the control means the SharePoint update is driven on the KEV clock rather than folded into the next farm window, and completion is measured per server against the fixed build, not per farm: a farm is remediated only when every server running the SharePoint Server product has taken the update, and one that has updated some of its servers still answers requests on the rest. The population to enumerate first is any farm whose authenticated surface is open to a broad user base, because the packet describes the attacker as authorized — the account is legitimate, so the deserialization path is reachable by whoever already holds a SharePoint account rather than by an attacker who must first defeat authentication, and tightening per-account privilege does not close it. Precondition: the packet records no live-patch path for this product class and gives the vendor update as the remediation, so there is no interim measure here that removes the deserialization sink — the clock is met by the update landing on every affected server, not by a compensating rule, and a server that has taken the update but not been through the restart the product class requires has not met it. Priority follows the packet rather than the CVSS band or PoC availability: no public PoC is recorded, but exploitation is confirmed in the wild and the KEV due date is three days after listing, so the absence of a published exploit is not a reason to run the standard window. Because exploitation is confirmed, a farm reachable by any account holder during the exposure window needs triage of what ran on it rather than being closed on the patch record.",
+        "evidence": "Packet name: 'Microsoft SharePoint Server Deserialization of Untrusted Data Vulnerability' (CWE-502). Packet vector: 'Deserialization of untrusted data in Microsoft Office SharePoint allows an authorized attacker to execute code over a network.' Packet attack_vector adds: 'CISA KEV-listed 2026-07-01 (due 2026-07-04) with confirmed in-the-wild exploitation.' active_exploitation confirmed; poc_available false; CVSS 8.8; RWEP 59. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch path for this product class; remediation is the vendor update (or, for end-of-life products, decommissioning) plus the named compensating controls until it lands.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2026-48558": {
     "name": "SimpleHelp Authentication Bypass Vulnerability",
@@ -37469,7 +37920,32 @@
         "adequate": false,
         "gap": "Technical vulnerability management requires confirming remediation efficacy, not just patch application — relevant here given the reported incomplete fix."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-018",
+        "name": "SCANNER-PAPER-COMPLIANCE-TEST",
+        "description": "Every framework control cited on this entry closes on a version number, and this is the packet where that is not evidence of remediation: the live-patch note records community reports on the JoomShaper forum, post-6.6.2, that the IconsTrait.php code path may remain exploitable after the 6.6.2 patch, and states that operators should verify remediation rather than treat the version bump alone as sufficient. An extension inventory that reads SP Page Builder's reported version and marks the site patched is therefore paper compliance on this specific defect. The operational test for this product is a request rather than a version read: against a staging copy of the site, from an unauthenticated client carrying no session and no CSRF token, POST a PHP payload to SP Page Builder's asset.uploadCustomIcon task and then attempt to fetch the resulting file. The site is remediated when the upload is refused before anything is written — not when the extension reports 6.6.2. Re-run the test after every subsequent SP Page Builder update, because the packet's point is that a version bump on this exact code path has already once been reported as not closing it. Precondition: this control verifies, it does not repair. The missing session, permission and CSRF checks on the upload task are the vendor's to restore, so a site that fails the test has only the vendor's next release, or removal of the extension, as an actual fix — the test result is not itself a mitigation. And because the packet records confirmed exploitation with a public PoC, a site passing the test today establishes nothing about whether it was already exploited before the test was run.",
+        "evidence": "Packet live_patch_notes: 'Community reports (JoomShaper forum, post-6.6.2) indicate the IconsTrait.php code path may remain exploitable after the 6.6.2 patch; operators should verify remediation rather than treat the version bump alone as sufficient.' patch_available true, live_patch_available false. attack_vector: 'An unauthenticated attacker POSTs directly to SP Page Builder's asset.uploadCustomIcon task, which accepts any file type with no session, permission, or CSRF check, then requests the uploaded PHP file to execute code and create a rogue Super User account.' Vector: 'A vulnerability in SP Page Builder for Joomla allows unauthenticated users to upload arbitrary files, ultimately resulting in the upload and execution of PHP code.' cisa_kev true, kev_date 2026-07-07, active_exploitation 'confirmed', poc_available true, cvss 9.8, rwep_score 68, cwe_refs CWE-434. Citing gaps: ISO-27001-2022-A.8.8 Management of technical vulnerabilities, NIST-800-53-SI-2 Flaw Remediation, NIS2-Art21-vulnerability-management and AU-ISM-1546 Patch operating systems and applications.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management",
+          "AU-ISM-1546"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet's exploitation path ends in two artifacts that no extension update removes: a PHP file the attacker uploaded through the asset.uploadCustomIcon task and then requested in order to execute it, and a rogue Super User account created on the Joomla site. Updating SP Page Builder replaces extension code — it does not delete a file already written into the site's upload directories, and it does not touch the Joomla user table, so a site recorded as patched can still be holding an administrator account the attacker controls. Applied to this deployment the requirement is that a publicly reachable Joomla site running SP Page Builder during the exposure window is handled as compromised rather than as patched: enumerate Super User and administrator accounts against a known-good list and remove any not provisioned by the site's own administrators, rotate the credentials of the accounts that remain, search the extension's upload directories and the web root for PHP files that belong to no installed package, and restore from a known-good backup in preference to cleaning in place where the site's content allows it. This entry gives the control more weight than usual, because the packet also records that the 6.6.2 patch may not have closed the upload path — so the update can be treated as ending neither the exposure nor the persistence. Precondition: this is the incident-response default, not a preventive control. It does not close the upload endpoint — that is the vendor fix, subject to the verification requirement recorded separately on this entry — and a backup taken during the exposure window may itself already contain the rogue account or the uploaded file, so the restore point must be chosen against that window rather than by recency.",
+        "evidence": "Packet attack_vector: 'An unauthenticated attacker POSTs directly to SP Page Builder's asset.uploadCustomIcon task, which accepts any file type with no session, permission, or CSRF check, then requests the uploaded PHP file to execute code and create a rogue Super User account.' Vector: 'A vulnerability in SP Page Builder for Joomla allows unauthenticated users to upload arbitrary files, ultimately resulting in the upload and execution of PHP code.' live_patch_notes: 'Community reports (JoomShaper forum, post-6.6.2) indicate the IconsTrait.php code path may remain exploitable after the 6.6.2 patch; operators should verify remediation rather than treat the version bump alone as sufficient.' patch_available true, live_patch_available false, poc_available true, active_exploitation 'confirmed', cisa_kev true, kev_date 2026-07-07, cvss 9.8, rwep_score 68. Citing gaps include NIST-800-53-SI-2 Flaw Remediation, NIS2-Art21-vulnerability-management and UK-CAF-B4 System security.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2026-48282": {
     "name": "Adobe ColdFusion Path Traversal Vulnerability",
@@ -38642,7 +39118,30 @@
         "adequate": false,
         "gap": "Identity-and-access controls were structurally bypassed since the vulnerability itself manufactures a valid administrator identity without authentication."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-134",
+        "name": "DEVICE-MANAGEMENT-GATEWAY-CONFIG-ENDPOINT-AUTH-INPUT-NEUTRALIZATION",
+        "description": "PRTG Network Monitor's web interface is the management plane for everything it polls, and the packet places the defect at two points on it at once. /public/login.htm is a page that by design answers unauthenticated callers, and a crafted HTTP request can override the attributes of its include directive; /api/addusers, the account-creation function that request is then made to reach, performs no authentication of its own — which is why the entry is classed CWE-306, missing authentication for a critical function. Bound to this product, the control means every PRTG API function that changes state, account creation first among them, makes its own authorization decision before it executes rather than inheriting one from the authenticated pages that normally invoke it; and the attribute the pre-authentication login page uses to select what it includes is validated against a fixed allow-list rather than taken from the request. It also means no PRTG web interface is left reachable from a segment with no operational need to reach it. This is why the identity-and-access controls cited on this entry do not touch the path: the attacker never holds a PRTG account, so no credential is presented and no access decision is consulted — an attestation that every PRTG operator authenticates at login passes cleanly while the endpoint creates administrators for callers who never logged in. Distinguishing test: from an unauthenticated client on a staging instance, send the crafted request to /public/login.htm naming /api/addusers with id and users parameters, then enumerate the instance's user list and confirm it is unchanged — testing that the login page rejects bad credentials proves nothing here, because the exploit never attempts a login. Preconditions: the endpoint-side authorization and the include-attribute validation are properties the vendor fix establishes — the packet records PRTG before 18.2.40.1683 as affected with a patch available, so this control states what to verify, not what an operator can implement. Restricting which segments can reach the web interface bounds who can send the request but leaves the path fully exploitable to anything inside the permitted segment, and it is unavailable where the console must stay broadly reachable for operational use. And because exploitation is confirmed and the outcome is a persistent read-write account, upgrading closes the inclusion path but does not delete an account created through it: an instance that was reachable during the exposure window needs its user list compared against a known-good baseline rather than being closed on the version number.",
+        "evidence": "Packet vector: 'PRTG Network Monitor before 18.2.40.1683 allows remote unauthenticated attackers to create users with read-write privileges (including administrator). A remote unauthenticated user can craft an HTTP request and override attributes of the include directive in /public/login.htm and perform a Local File Inclusion attack, by including /api/addusers and executing it. By providing the id and users parameters, an unauthenticated attacker can create a user with read-write privileges (including administrator).' cwe_refs CWE-306; cisa_kev true with kev_date 2025-02-04; active_exploitation confirmed; poc_available true; cvss 9.8; rwep_score 67; patch_available true; live_patch_available false.",
+        "gap_closes": [
+          "UK-CAF-B2",
+          "NIST-800-53-IA-2"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The clock on this entry is unusual and that is the point: the packet places the fix at PRTG 18.2.40.1683 while CISA listed the CVE on 2025-02-04, so under this control's 'KEV listing or patch availability, whichever is later' rule the SLA runs from the listing, against a defect whose fix had been shipping for years already. The population still exposed at listing is therefore not one waiting on a vendor — it is instances that no update process has touched — which makes meeting the SLA a discovery problem before it is a patching problem. The requirement here is to find every PRTG installation, including the ones standing on a departmental server, a lab host, or a machine inherited from a team that no longer exists, and confirm each reports a version at or above 18.2.40.1683 rather than confirming that the inventoried instances were updated. Track and document mitigation state per instance, not in aggregate: with a public PoC, confirmed in-the-wild exploitation and an unauthenticated path that mints an administrator on a monitoring server, one missed install is not a percentage point on a compliance figure. Precondition: the packet records a vendor patch and no live-patch path, so there is nothing that removes this flaw short of reaching the fixed version — where an instance genuinely cannot take the upgrade inside the window, the only remaining lever is removing its reachability from segments that have no need to reach it, and that is a holding measure with a review date, not a remediation. Because exploitation is confirmed, an instance found late is a triage item as well as a patch item: the flaw's product is a persistent privileged account, and the upgrade does not remove one already created.",
+        "evidence": "Packet: cisa_kev true with kev_date 2025-02-04; active_exploitation confirmed; poc_available true; cvss 9.8; rwep_score 67; patch_available true; live_patch_available false. Vector states the affected range as 'PRTG Network Monitor before 18.2.40.1683' and the outcome as remote unauthenticated attackers creating 'users with read-write privileges (including administrator)'.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIS2-Art21-vulnerability-management",
+          "ISO-27001-2022-A.8.9"
+        ]
+      }
+    ]
   },
   "CVE-2025-24085": {
     "name": "Apple Multiple Products Use-After-Free Vulnerability",
@@ -38684,7 +39183,33 @@
         "adequate": false,
         "gap": "Essential Eight's patch-OS timeline (48 hours for extreme-risk vulnerabilities) is difficult to meet for a distributed device fleet without forced MDM update enforcement."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-056",
+        "name": "MOBILE-ENDPOINT-MDM-ENFORCED-KEV-SLA",
+        "description": "The packet names the fixed build per platform — iOS 18.3 and iPadOS 18.3, iPadOS 17.7.6, macOS Sequoia 15.3, macOS Sonoma 14.7.5, macOS Ventura 13.7.5, tvOS 18.3, visionOS 2.3, watchOS 11.3 — and that list is what makes this a fleet-enforcement problem rather than a single-update problem: a real Apple estate spans several of those trains at once, and a device held on an older train is remediated only by the build named for its own train, not by the newest one. Applied to this CVE the control means the update is driven from the management plane on the KEV clock that opened 2025-01-29 with user deferral disallowed, and completion is measured per device against the fixed build for the train it is actually on — not against 'update available', 'assigned' or 'downloaded' in the management console. Deferral is the load-bearing part here because of how short the exploitation path is: once the malicious application is on the device it reaches the flaw with a small number of local XPC calls into the mediaplaybackd daemon's CoreMedia Remaker subsystem, needing no network position and no further user step, so the exposure window is exactly the time the device spends below its fixed build and a user postponing the update for a week is a week in which any application on that device holds a path to privileged code execution. Precondition: the packet records no live-patch path, so nothing short of the device reaching its fixed build removes the flaw — a device that cannot reach it is not covered by this SLA at all and belongs on the access-condition and application-restriction levers instead of being carried as an exception. Because exploitation is confirmed, a device that ran an untrusted application while below its fixed build belongs on the incident path rather than being closed on the update record.",
+        "evidence": "Packet vector: 'A use after free issue was addressed with improved memory management. This issue is fixed in iOS 18.3 and iPadOS 18.3, iPadOS 17.7.6, macOS Sequoia 15.3, macOS Sonoma 14.7.5, macOS Ventura 13.7.5, tvOS 18.3, visionOS 2.3, watchOS 11.3. A malicious application may be able to elevate privileges. Apple is aware of a report that this issue may have been actively exploited against versions of iOS before iOS 17.2.' Packet attack_vector: 'A malicious application on the device makes a small number of XPC calls to the mediaplaybackd daemon's CoreMedia Remaker subsystem, triggering a use-after-free in FigRemakerTrack object handling that yields code execution in that privileged process and, ultimately, privilege escalation.' CWE-416; CISA KEV listed 2025-01-29; active_exploitation confirmed; poc_available true; CVSS 10; RWEP 83. patch_available true, live_patch_available false (live_patch_notes null — no interim mechanism is recorded).",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "Because the packet's exploitation path starts with a malicious application resident on the device, this control's two halves split cleanly for this CVE. First, the fixed build named for each train must operate as an access condition — organizational mail, VPN and document access denied to any device below iOS/iPadOS 18.3, iPadOS 17.7.6, macOS Sequoia 15.3, macOS Sonoma 14.7.5, macOS Ventura 13.7.5, tvOS 18.3, visionOS 2.3 or watchOS 11.3 — rather than a row on a patch-compliance report; an estate that surfaces the stale build on a dashboard while the device keeps its access has recorded the exposure rather than removed it. The distinguishing test is to enrol a device pinned below the fixed build for its train and confirm the policy actually denies it protected resources. Second, since the attacker's foothold is an installed application making XPC calls into the CoreMedia Remaker subsystem of mediaplaybackd, constraining what code is permitted to install and run at all is the only lever the operator holds on a device that cannot yet take its fixed build. Precondition, and this is where the second half gets over-claimed: restricting untrusted or side-loaded application installation raises the bar for getting the attacker's application onto the device, but it does not evict an application already installed, and it does not cover one that arrived through the normal store channel — a device suspected of already running the malicious application belongs on the incident path, not the install-policy path. The whole control is a holding measure for the window before the fixed build lands, not a substitute for it: the packet records no live-patch path, so only the build itself removes the use-after-free.",
+        "evidence": "Packet vector names the fixed builds ('iOS 18.3 and iPadOS 18.3, iPadOS 17.7.6, macOS Sequoia 15.3, macOS Sonoma 14.7.5, macOS Ventura 13.7.5, tvOS 18.3, visionOS 2.3, watchOS 11.3') and states 'A malicious application may be able to elevate privileges.' Packet attack_vector places the trigger in a malicious application on the device making a small number of XPC calls to the mediaplaybackd daemon's CoreMedia Remaker subsystem, triggering a use-after-free (CWE-416) in FigRemakerTrack object handling. CISA KEV listed 2025-01-29; active_exploitation confirmed; CVSS 10; RWEP 83; poc_available true. patch_available true; live_patch_available false. Citing gap NIST-800-53-CM-7 (Least Functionality) is recorded against this entry.",
+        "gap_closes": [
+          "NIST-800-53-CM-7",
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2025-23006": {
     "name": "SonicWall SMA1000 Appliances Deserialization Vulnerability",
@@ -38758,7 +39283,30 @@
         "adequate": false,
         "gap": "A five-year-old client-side library flaw remained unpatched in production long enough to appear in attacker C2 infrastructure, showing flaw-remediation tracking rarely extends to bundled front-end JS dependencies."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-021",
+        "name": "TIER-3-DEPENDENCY-INVENTORY",
+        "description": "jQuery is not installed software on an estate — it is a file that ships inside applications, and the packet's affected range (greater than or equal to 1.0.3 and before 3.5.0) covers effectively every copy ever vendored into a web root, emitted by a build step, or embedded in a third-party product's web interface. That is why a flaw-remediation programme keyed to installed packages reports nothing here: the vulnerable code is a dependency no manifest on the host declares, so the scan result is silence rather than a finding. The requirement is that the software inventory reach those copies — enumerate jQuery across the served content and build outputs of every web application in the estate, not only the declared direct dependencies, and record each copy's version against the packet's fixed version of 3.5.0. Two populations fall out of that scan and they have different terminal states, which is the part an inventory row alone hides. A copy the organisation builds and ships is a code change, not a patch: the packet fixes the flaw in 3.5.0, and a copy on 1.x or 2.x reaches that only through a major-version upgrade of the library, which is application work with regression risk and has to be scheduled as such. A copy bundled inside a third-party product is not substitutable by the operator at all — that item belongs on the vendor's fixed-build clock, and where the vendor ships no build carrying jQuery 3.5.0 or later, the honest terminal state is replacing or removing that product, because an inventory row left open indefinitely marks a KEV-listed flaw with a public PoC as managed while it stays exploitable. Scope the sweep to what the packet establishes — jQuery below 3.5.0 — rather than to front-end libraries in general, which the packet gives no basis for.",
+        "evidence": "Packet vector: 'In jQuery versions greater than or equal to 1.0.3 and before 3.5.0, passing HTML containing <option> elements from untrusted sources - even after sanitizing it - to one of jQuery's DOM manipulation methods (i.e. .html(), .append(), and others) may execute untrusted code. This problem is patched in jQuery 3.5.0.' CWE-79, CVSS 6.9, RWEP 65, poc_available true, active_exploitation 'confirmed', cisa_kev true with kev_date 2025-01-23. patch_available true; live_patch_available false with live_patch_notes null. AU-Essential-8-Patch (Patch operating systems), ISO-27001-2022-A.8.8 and NIST-800-53-SI-2 (Flaw Remediation) are recorded among the framework gaps citing this CVE.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-018",
+        "name": "SCANNER-PAPER-COMPLIANCE-TEST",
+        "description": "Two distinct measurements report clean on this CVE while it stays exploitable, and the control's job here is to name both. The first is version-only scanning: an inventory that reads installed operating-system packages has no row for a jQuery file served out of a web root, so the result says nothing about the flaw rather than saying it is absent — silence read as a pass. The second is specific to this defect and is the one that catches operators out. The packet states the untrusted HTML executes 'even after sanitizing it', and that the crafted <option> content survives common sanitization, so an application recording an HTML sanitizer as its compensating control has recorded something the packet says does not hold on this path. The operational test therefore has to exercise the documented behaviour rather than a version string: on a staging copy of each affected application, pass HTML containing crafted <option> elements through whatever sanitization that application applies and on into the jQuery DOM manipulation methods the packet names — .html(), .append() and the others — and confirm nothing executes in the resulting page. A copy that executes the payload is exploitable regardless of what the sanitizer is configured to strip, and an application whose primary bundle reports jQuery 3.5.0 while a second bundle on the same page still loads an older copy fails this test even though the version report reads clean. Precondition: this is a test, not a mitigation. A failing result identifies an application that must take the library upgrade to 3.5.0 or later; passing it on one application says nothing about the others, so it runs per application rather than once for the estate, and no result of it removes the need for the upgrade.",
+        "evidence": "Packet vector: untrusted HTML containing <option> elements passed to jQuery's DOM manipulation methods '(i.e. .html(), .append(), and others) may execute untrusted code', and the packet states this holds 'even after sanitizing it'; 'This problem is patched in jQuery 3.5.0.' Attack vector: the crafted <option> content 'survives common sanitization', 'causing the browser to execute attacker-controlled JavaScript in the victim's session'. CWE-79, poc_available true, active_exploitation 'confirmed', cisa_kev true with kev_date 2025-01-23. NIS2-Art21-vulnerability-management and UK-CAF-B4 (System security) are recorded among the framework gaps citing this CVE.",
+        "gap_closes": [
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2024-50603": {
     "name": "Aviatrix Controllers OS Command Injection Vulnerability",
@@ -39490,7 +40038,31 @@
         "adequate": false,
         "gap": "Technical vulnerability management assumes an eventual vendor patch — NUUO has not responded to disclosure and the flaw persists in the latest firmware, so remediation-SLA-based controls cannot be satisfied and must fall back to decommissioning."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-127",
+        "name": "UNSUPPORTED-NETWORK-DEVICE-ENUMERATION-AND-RETIREMENT",
+        "description": "The NVRmini2 is a network video recorder, and this packet resolves the patch-versus-replace question in one direction: patch_available is false and no live-patch path is recorded, so there is no fixed build for an operator to reach. That makes each unit in service a replacement item rather than a remediation-clock item, and the requirement is an inventory naming every NVRmini2 with its firmware level - the packet scopes the defect to builds 'through 3.11' - together with a vendor-obtained answer on whether a fixed firmware exists for that hardware, rather than an assumption in either direction. Any unit for which the vendor produces a fix belongs on a remediation clock; every unit for which it does not belongs on a dated removal or replacement schedule, because a risk acceptance with no removal date leaves a KEV-listed device carrying a public PoC and confirmed exploitation in service indefinitely. Scope the inventory to what the packet names - NUUO NVRmini2 - not to every recorder or camera on the estate; the packet ties the missing authentication check to handle_import_user.php on this product and provides no mapping into other video hardware, so treating all surveillance appliances as instances of this CVE manufactures removal work against devices no evidence implicates. The distinguishing test is per unit rather than per estate: produce, for each NVRmini2 in service, its firmware level and the vendor's answer on a fix - an asset register that records the device as 'segmented' or 'monitored' with no removal date has recorded the exposure rather than removed it. Precondition, and it is why this cannot end at a segmentation entry in the risk register: the exposure being scored is an unauthenticated archive upload that creates an arbitrary account, so restricting which segments can reach the device bounds who can send that upload but repairs nothing, and with no vendor fix recorded there is no update behind which that interim state resolves. Because active_exploitation is confirmed and the primitive is account creation, a unit that was reachable during the exposure window may already carry an attacker-created account and, through the CVE-2011-5325 chain the packet names, attacker-written files under the web root; isolating that unit afterwards does not clear it, and its accounts and configuration must be treated as attacker-controlled until it is removed.",
+        "evidence": "Packet records patch_available: false, live_patch_available: false, live_patch_notes: null - no fixed build is recorded for this entry. CISA KEV-listed 2024-12-18, active_exploitation confirmed, poc_available true, CVSS 9.8, RWEP 79. Vector: 'NUUO NVRmini2 through 3.11 allows an unauthenticated attacker to upload an encrypted TAR archive, which can be abused to add arbitrary users because of the lack of handle_import_user.php authentication. When combined with another flaw (CVE-2011-5325), it is possible to overwrite arbitrary files under the web root and achieve code execution as root.' Citing gaps include AU-Essential-8-Patch, ISO-27001-2022-A.8.8 and NIS2-Art21-vulnerability-management.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-134",
+        "name": "DEVICE-MANAGEMENT-GATEWAY-CONFIG-ENDPOINT-AUTH-INPUT-NEUTRALIZATION",
+        "description": "The NVRmini2's web interface is the device's configuration plane, and the defect is that plane's file-accepting endpoint making no authorization decision at all: the packet places the missing check in handle_import_user.php, which takes an encrypted TAR archive from an unauthenticated caller and imports users from it, so an attacker creates an arbitrary account without ever presenting a credential. Bound to this product, the control means a configuration endpoint that accepts an uploaded archive authorizes its caller before the archive is processed, and resolves and validates every destination path the import writes before the write executes - the chained CVE-2011-5325 step the packet names is exactly that second half failing, with archive content overwriting files under the web root and ending in code execution as root. This is also why the identification-and-authentication and identity-and-access gaps cited on this entry pass their attestations while the path stays open: the attacker never authenticates as any NVR user, so per-account authentication and privilege scoping are never consulted - the account model is bypassed rather than abused, and an attestation that every recorder administrator authenticates at login reads clean. Distinguishing test: from a segment with no operational need to administer the recorder, send an unauthenticated upload to the user-import handler on a bench unit and confirm it is refused before anything is imported or written to disk. Precondition, and it is decisive on this entry: endpoint-side authorization and path validation are properties a vendor fix would have to establish, and this packet records patch_available false with no live-patch path, so there is no update behind which this control resolves. The only operator-side lever left is reachability - restricting which segments can reach the management interface bounds the population that can send the upload, but leaves the endpoint fully exploitable to anything inside the permitted segment, and it is unavailable wherever the recorder's web interface must stay reachable for normal viewing and administration. Treat reachability restriction as a holding measure attached to a removal date, never as closure of the endpoint.",
+        "evidence": "Packet vector: 'NUUO NVRmini2 through 3.11 allows an unauthenticated attacker to upload an encrypted TAR archive, which can be abused to add arbitrary users because of the lack of handle_import_user.php authentication. When combined with another flaw (CVE-2011-5325), it is possible to overwrite arbitrary files under the web root and achieve code execution as root.' CWE-306 (missing authentication), CVSS 9.8, RWEP 79, poc_available true, active_exploitation confirmed, CISA KEV-listed 2024-12-18. patch_available: false and live_patch_available: false, with live_patch_notes null. Citing gaps include NIST-800-53-IA-2 (Identification and Authentication), UK-CAF-B2 (Identity and access control) and NIST-800-53-CM-7 (Least Functionality).",
+        "gap_closes": [
+          "NIST-800-53-IA-2",
+          "UK-CAF-B2",
+          "NIST-800-53-CM-7"
+        ]
+      }
+    ]
   },
   "CVE-2021-40407": {
     "name": "Reolink RLC-410W IP Camera OS Command Injection Vulnerability",
@@ -40160,7 +40732,30 @@
         "adequate": false,
         "gap": "Apple shipped a fix in Safari 18.1.1/iOS 17.7.2/18.1.1/macOS 15.1.1, but the flaw was exploited as a zero-day before any patch existed, so flaw remediation alone could not have prevented initial exploitation."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-056",
+        "name": "MOBILE-ENDPOINT-MDM-ENFORCED-KEV-SLA",
+        "description": "The packet's fix list is what makes this a fleet-wide push rather than one update: Safari 18.1.1, iOS 17.7.2 and iPadOS 17.7.2, iOS 18.1.1 and iPadOS 18.1.1, macOS Sequoia 15.1.1 and visionOS 2.1.1 all carry the same cookie-state-management fix, so remediation means every Apple device class the organization manages moves together on the clock that opened with the 2024-11-21 KEV listing — not on the next OS-upgrade window, and not in the order a severity-ranked queue would produce. That ranking is the specific failure mode here: the packet records CVSS 6.3 and poc_available false, which is precisely the profile a monthly patch queue defers, while the same packet records active exploitation as confirmed, notes Apple is aware of a report of exploitation on Intel-based Mac systems, and describes this cross-site-scripting flaw being chained with a separate code-execution flaw. The user-deferral half of the control is the load-bearing half on this fix, because on iOS, iPadOS and visionOS it arrives as an OS update whose prompt the person holding the device can dismiss indefinitely; the packet records live_patch_available false, so the fixed build is the only remediation and a dismissed prompt is untreated exposure. Distinguishing test: take an enrolled device one build below the fixed build for its track and confirm the management system installs the update inside the SLA without the holder being able to postpone it — an estate whose compliance report shows the update as available and pending user action has recorded the exposure rather than removed it. Precondition: this reaches enrolled devices only. Personally-owned Macs, iPhones and iPads used for work, and any unenrolled device, sit outside the push entirely, and the delivery path the packet describes — a user lured to attacker-controlled web content in Safari — needs nothing from the organization to work on them.",
+        "evidence": "Packet: CWE-79 cross-site scripting, with the vector stating a cookie management issue addressed with improved state management, fixed in Safari 18.1.1, iOS 17.7.2 and iPadOS 17.7.2, iOS 18.1.1 and iPadOS 18.1.1, macOS Sequoia 15.1.1, visionOS 2.1.1, and that Apple is aware of a report that this issue may have been actively exploited on Intel-based Mac systems. cisa_kev true, kev_date 2024-11-21, active_exploitation confirmed, poc_available false, cvss 6.3, rwep_score 55. patch_available true, live_patch_available false, live_patch_notes null. The attack_vector records a targeted campaign against Intel-based Macs, chained with a separate code-execution flaw. AU-Essential-8-Patch (Patch operating systems), NIST-800-53-SI-2 (Flaw Remediation) and NIS2-Art21-vulnerability-management (Vulnerability handling) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "The packet names two concurrently fixed iOS/iPadOS tracks — 17.7.2 and 18.1.1 — and that is what a single numeric minimum-version floor cannot express. Set the floor at 18.1.1 and every device correctly remediated on 17.7.2 fails the policy; set it at 17.7.2 and a device on 18.0 or 18.1 passes it, because those builds compare as newer while still carrying the unfixed cookie-state handling. For this entry the enforced floor has to be stated per track: at or above 17.7.2 on the 17.x line, at or above 18.1.1 on the 18.x line. The Mac side has the same shape for a different reason — the packet lists Safari 18.1.1 as a fixed build distinct from macOS Sequoia 15.1.1, so a check keyed only on the macOS build cannot see a Mac whose remediation is the Safari update, and the packet places the reported exploitation on Intel-based Mac systems. visionOS 2.1.1 sits on the same fix list and is the device class conditional-access policies most often do not enumerate at all. The requirement is that the per-track fixed build function as an access condition — a device below it is refused mail, VPN and document access until it is at or above it — rather than appearing as a row on a patch-compliance report. Distinguishing test: enrol one device pinned at 18.0 and one at 17.7.2 and confirm the policy denies the first and admits the second; a floor expressed as a single version number gets exactly this pair wrong, in the direction that admits the unfixed device. Precondition: withholding access bounds what a below-fix device can reach, it does not protect the device. The packet's exploitation path is a user loading attacker-controlled web content in Safari, which requires no organizational resource, so a blocked device remains fully exploitable and its user's own sessions and credentials stay in scope. Where a device cannot be brought to the fixed build, blocking it is the holding measure for the window, not the remediation.",
+        "evidence": "Packet: CWE-79 cross-site scripting reached by processing maliciously crafted web content; the vector names the fix as Safari 18.1.1, iOS 17.7.2 and iPadOS 17.7.2, iOS 18.1.1 and iPadOS 18.1.1, macOS Sequoia 15.1.1, visionOS 2.1.1 — two iOS/iPadOS tracks fixed concurrently, and Safari listed separately from macOS Sequoia. cisa_kev true, kev_date 2024-11-21, active_exploitation confirmed, poc_available false, cvss 6.3, rwep_score 55. patch_available true, live_patch_available false, live_patch_notes null. The vector states Apple is aware of a report that this issue may have been actively exploited on Intel-based Mac systems. ISO-27001-2022-A.8.8 (Management of technical vulnerabilities) and UK-CAF-B4 (System security) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2024-44308": {
     "name": "Apple Multiple Products Code Execution Vulnerability",
@@ -41599,7 +42194,39 @@
         "adequate": false,
         "gap": "Patch Applications guidance assumes a supported product; CSA 4.6.x's EOL status means the compliant action is decommission, not patch."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-122",
+        "name": "EOL-ASSET-DECOMMISSION",
+        "description": "This entry is a split estate and the split has to be resolved per appliance before any remediation clock means anything. The packet places the flaw in the admin web console of Ivanti CSA before version 5.0.2, so units on the 5.0.x branch have a build to reach — 5.0.2 or later — and belong on the KEV clock that opened 2024-10-09. The lesson entry's framework_coverage records the other population: CSA 4.6.x is past end-of-life with no vendor patch for that branch, so for those units there is no fixed build at all and the compliant action it names is decommission, not patch. The requirement in operational terms: enumerate every Ivanti CSA in service with its branch, put the 5.0.x units on the interim upgrade clock, and put every 4.6.x unit on a dated replacement or removal schedule — a risk acceptance with no removal date leaves a KEV-listed appliance with confirmed in-the-wild exploitation in service indefinitely. Scope it to what the packet names, Ivanti CSA units; nothing here implicates other Ivanti products, and inventorying them as instances of this CVE manufactures replacement work against software no evidence in this entry touches. Precondition, and this is where the interim half is normally over-claimed: reaching 5.0.2 closes this defect only and says nothing about anything found in the appliance since that build, which is exactly why the requirement cannot end at 'every CSA reports 5.0.2 or later'. The packet records patch_available true with live_patch_available false and no live-patch note, so each unit has to be taken through the vendor update — there is no in-place mitigation that leaves the appliance running unmodified. And because active exploitation is confirmed, a unit that was reachable during the exposure window is an incident item first: an appliance that already executed an attacker's commands is not remediated by the upgrade it later receives, and a 4.6.x unit being replaced is not remediated by the replacement either — what it held has to be treated as read.",
+        "evidence": "Packet: 'An OS command injection vulnerability in the admin web console of Ivanti CSA before version 5.0.2 allows a remote authenticated attacker with admin privileges to obtain remote code execution.' CISA KEV listing 2024-10-09; active_exploitation confirmed; poc_available false (so the absence of a public exploit is not grounds to defer — exploitation is already observed); RWEP 55; CVSS 7.2; patch_available true; live_patch_available false with live_patch_notes null. The repo lesson entry's framework_coverage for NIST-800-53-SI-2 records that 'CSA 4.6.x is past end-of-life with no vendor patch available for this branch, so flaw-remediation controls have no compliant remediation path short of migration', and its AU-Essential-8-Patch entry records that 'CSA 4.6.x's EOL status means the compliant action is decommission, not patch.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "For this appliance the control's trigger conditions are met literally rather than by analogy: the packet's outcome is full remote code execution on the CSA, the admin position the flaw needs can be obtained by chaining a separate CSA authentication-bypass flaw so the effective path requires no credential, and exploitation is confirmed in the wild. What that means here is that the vendor upgrade is not the remediation decision for any CSA that was reachable while unpatched — the upgrade closes the injection sink and removes nothing an attacker executed through it. The default for those units is extraction of the running configuration for reference, rebuild from vendor media onto a build at or above 5.0.2, and rotation of every credential the appliance held or could authenticate to, before it is returned to service. This is precisely what the two vulnerability-management gaps cited here cannot express: A.8.8 and NIS2 Art 21 handling both terminate at a binary open/remediated verdict, so an appliance that was compromised and then upgraded closes as remediated on both attestations while the attacker's access persists through whatever was left on it. The distinguishing test is a question asked per unit rather than a scan: for each CSA in service before the upgrade, what evidence exists that it was not compromised? The lesson entry records that appliance-level logging on CSA is limited, so for most estates that evidence does not exist, and absence of an alert is not it. Precondition: rebuilding is only remediation if the rebuilt unit lands at or above 5.0.2 and the credentials it held are rotated — a 4.6.x unit rebuilt from its own image is reinstated with the vulnerability intact, since the lesson records no patch exists for that branch, and a rebuild without credential rotation returns a clean appliance to an attacker who already holds what it stored.",
+        "evidence": "Packet attack_vector: an attacker with admin access to the CSA administrative web console — 'obtained directly or by chaining a separate CSA authentication-bypass flaw' — injects OS commands 'passed unsanitized to the underlying operating system, yielding full remote code execution on the appliance'. CISA KEV 2024-10-09; active_exploitation confirmed; CWE-77 and CWE-78; patch_available true; live_patch_available false, live_patch_notes null. The repo lesson entry's defense_chain response section states that 'response must assume complete appliance compromise and treat any credentials it held as burned' and names isolating the appliance, hunting for implanted webshells and harvested credentials, and rotating all credentials the appliance had access to; its detection section records that 'appliance-level logging on CSA is limited'.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-134",
+        "name": "DEVICE-MANAGEMENT-GATEWAY-CONFIG-ENDPOINT-AUTH-INPUT-NEUTRALIZATION",
+        "description": "The CSA's administrative web console is the management surface this control governs, and the defect sits on it: operator-supplied input reaches the underlying operating system unsanitized (CWE-77, CWE-78) and executes. Bound to this appliance the control means two properties. First, the console's command-invoking functions neutralize their input at the point it reaches the operating system — argument-array invocation or validation at the sink — rather than relying on the caller having authenticated as an administrator, because the packet's own path has that authentication reached by chaining a separate CSA authentication-bypass flaw, at which point the privilege verdict the sink is trusting is worthless. Second, no CSA is left with its administrative console reachable from a segment with no operational need to administer it. This is why the least-privilege gap cited on this entry does not close the path: the admin role is correctly assigned and correctly scoped, and the exploit trades that bounded application-admin position for unbounded OS execution on the appliance — an application-admin-to-host-OS bridge that a user-to-role privilege review never examines, so an AC-6 attestation passes cleanly while the console stays wired to a shell. Distinguishing test: enumerate, per CSA, which networks can reach the administrative console, and from a segment with no administrative role confirm the console does not answer — 'the appliance is behind the firewall' is a statement about topology, not a demonstration that the admin surface is unreachable. Precondition, and this is the half that gets over-claimed: the input neutralization is a property the vendor update at 5.0.2 or later establishes; this control states what to verify, it does not implement it, and the lesson records no such update exists for the end-of-life 4.6.x branch. Until the update lands, restricting reachability bounds who can present the request but leaves the console fully exploitable to anything inside the permitted segment — a compromised administrator workstation or jump host satisfies the precondition in full — and the lever is unavailable entirely where the administrative interface must stay reachable for the appliance's normal operation. It also does nothing about an appliance that already holds an implant.",
+        "evidence": "Packet: the flaw is in 'the admin web console of Ivanti CSA before version 5.0.2'; CWE-77 and CWE-78; the attack_vector describes injected OS commands 'passed unsanitized to the underlying operating system, yielding full remote code execution on the appliance', with the admin position reachable 'by chaining a separate CSA authentication-bypass flaw'. The entry cites NIST-800-53-AC-6 (Least Privilege) and UK-CAF-B4 (System security) as insufficient. patch_available true with the packet's affected range ending before 5.0.2; live_patch_available false with no live-patch note recorded, so no in-place mitigation is available from the vendor.",
+        "gap_closes": [
+          "NIST-800-53-AC-6",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2024-9379": {
     "name": "Ivanti Cloud Services Appliance (CSA) SQL Injection Vulnerability",
@@ -43000,7 +43627,31 @@
         "adequate": false,
         "gap": "Least-functionality/removal of the end-of-life Flash Player plugin is the only durable fix; a control that merely inventories software does not force removal of an unsupported client component."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-122",
+        "name": "EOL-ASSET-DECOMMISSION",
+        "description": "This packet carries both halves of the control. A vendor fix is recorded — the vector names Flash Player 10.3.183.67 and, on the 11.x branch, 11.6.602.171 on Windows and Mac OS X and 11.2.202.273 on Linux — and the live-patch notes state the vendor update requires no reboot. Against that, the lesson's framework coverage records that removal of the end-of-life Flash Player plugin is the only durable fix, and that a control which merely inventories software does not force removal of an unsupported client component. Together those set the requirement: reaching a fixed Flash Player build is an interim state, and the terminal state is removing Flash Player from the host, or replacing the workload where something still depends on it. A KEV listing dated 2024-09-17 for a flaw the vector records as exploited in the wild in February 2013 is the evidence that the long tail is real — a host still exposed eleven years after the fix shipped is a host with no functioning update path for the product. Scope it to what the packet names: Adobe Flash Player, and specifically the Firefox plugin sandbox the vector identifies, on Windows, Mac OS X and Linux. The packet provides no mapping from this defect into other SWF-capable or media-runtime software, so treating every plug-in or media player in the estate as an instance of this CVE manufactures removal work against software no evidence implicates. Operationally: enumerate every host with Flash Player installed, record for each whether it can take the fixed build for its branch and platform, put those on the interim clock, and put every host on a dated removal or replacement schedule — a risk acceptance with no removal date leaves a KEV-listed flaw with confirmed exploitation in service indefinitely. Precondition on the interim half, which is where this is usually over-claimed: applying the 2013 update closes this defect only and says nothing about anything found in Flash Player since, and because the product's patch path has ended there is no later update to take, which is exactly why the requirement cannot end at every install reporting a fixed build. And because exploitation is confirmed and the delivery path is crafted SWF content the victim's browser loads, a host that browsed untrusted web content while exposed belongs on the incident path — hunted for second-stage payloads and rebuilt if confirmed — rather than being closed on the patch record.",
+        "evidence": "Packet: CWE-269 incorrect default permissions — the vector states the Firefox sandbox in Adobe Flash Player before 10.3.183.67 and 11.x before 11.6.602.171 on Windows and Mac OS X, and before 10.3.183.67 and 11.x before 11.2.202.273 on Linux, does not properly restrict privileges, making it easier for remote attackers to execute arbitrary code via crafted SWF content, as exploited in the wild in February 2013. cisa_kev true, kev_date 2024-09-17, active_exploitation confirmed, poc_available false, cvss 8.8, rwep_score 48. patch_available true, live_patch_available false, live_patch_notes: 'No live-patching primitive for this product; the vendor update (no reboot required) is the remediation.' The lesson's framework_coverage for NIST-800-53-CM-7 records that least-functionality/removal of the end-of-life Flash Player plugin is the only durable fix, and that a control which merely inventories software does not force removal of an unsupported client component; its response guidance names purging Flash from managed endpoints, hunting for post-exploitation second-stage payloads, and reimaging confirmed-compromised hosts. AU-Essential-8-App-Hardening (User application hardening), NIST-800-53-CM-7 (Least Functionality), NIST-800-53-SI-2 (Flaw Remediation) and ISO-27001-2022-A.8.8 (Management of technical vulnerabilities) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "NIST-800-53-CM-7",
+          "AU-Essential-8-App-Hardening",
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The clock is the one thing a vulnerability-management program will get wrong on this entry. The packet records the fix as shipping against exploitation the vector places in February 2013, and the KEV listing as 2024-09-17, so under 'whichever is later' the deadline runs from the 2024 listing rather than from the original disclosure — an eleven-year-old finding becomes a current obligation the day CISA lists it. Two of the packet's own fields are what keep it off the queue otherwise: poc_available is false, so a triage rule gated on public exploit code never raises it, and a program that ages or deprioritizes findings by CVE year treats a 2013 identifier as historic while the packet records active_exploitation as confirmed. For this entry the documented-compensating-controls branch is the one most operators will need, because live_patch_available is false and the durable answer for an unsupported plugin is removal, which usually needs a change window longer than the deadline: block SWF content at the web gateway for the hosts that cannot be reached inside the window, and record that as a time-bound compensating state carrying a removal date, not as remediation. Distinguishing test: query the vulnerability register for items whose KEV listing date postdates their CVE year by more than a year, and confirm each carries a deadline derived from the listing date — a register whose Flash Player finding is dated 2013 and sorted to the bottom is measuring disclosure age rather than exposure. Precondition: a gateway block on SWF content bounds delivery only over the paths the gateway sees; it does nothing for content that reaches the browser without traversing it, and it does not remove the plugin, so it holds only until the removal schedule lands.",
+        "evidence": "Packet: cisa_kev true with kev_date 2024-09-17, against a vector that records exploitation in the wild in February 2013 and names the fixed builds (10.3.183.67; 11.x fixed at 11.6.602.171 on Windows and Mac OS X and 11.2.202.273 on Linux). active_exploitation confirmed, poc_available false, cvss 8.8, rwep_score 48. patch_available true, live_patch_available false, live_patch_notes: 'No live-patching primitive for this product; the vendor update (no reboot required) is the remediation.' The lesson's prevention guidance names uninstalling or disabling the Flash Player plugin and blocking SWF content at the web gateway. NIS2-Art21-vulnerability-management (Vulnerability handling) and UK-CAF-B4 (System security) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "NIS2-Art21-vulnerability-management",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2014-0497": {
     "name": "Adobe Flash Player Integer Underflow Vulnerablity",
@@ -43334,7 +43985,22 @@
         "adequate": false,
         "gap": "Least-privilege alone does not help — the flaw lets an already-present low-privileged user escalate to SYSTEM, defeating the privilege boundary the control assumes."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-145",
+        "name": "ACTIVELY-EXPLOITED-OS-LPE-REMEDIATION-WINDOW-ENFORCEMENT",
+        "description": "The packet puts this in Windows Installer and describes a local low-privileged user triggering a repair, which briefly runs an elevated process, then either hijacking that elevated context or racing the rollback temp files with symlinks and oplocks to obtain a SYSTEM shell. Every account in that sequence is already legitimate, which is what makes the least-privilege gap cited on this entry unclosable from the account side: nothing over-broad is being abused, so an attestation showing ordinary users hold ordinary rights passes cleanly while any interactive user on the host can take SYSTEM. The remediation is therefore the Windows update itself, driven across every affected host on the clock that opened with the 2024-09-10 KEV listing rather than folded into the next monthly rollup, with completion measured by each host's installed build against the fixed build for its SKU - never by 'approved', 'downloaded' or 'installed' in the management console. The packet's live-patch notes are decisive on that measurement: there is no live-patching primitive for this product and the vendor update requires a reboot to be the remediation, so a host that has taken the update and not restarted is still running the vulnerable Windows Installer path and must be counted as exposed rather than compliant. Enumerate hosts where many non-administrative users hold interactive sessions first - shared workstations, terminal and jump hosts - because on those the precondition this flaw needs, an ordinary local user able to trigger an installer repair, is the normal operating state rather than an anomaly, and those are also the hosts where the required restart is most likely to be deferred because taking it evicts logged-on users. The distinguishing test pairs the build with the boot: read each host's installed build and its last restart time together, and treat any host whose last restart predates the update as unremediated - a patch report showing the update installed marks that machine compliant while the vulnerable repair path is still live in memory. Priority follows the packet rather than the CVSS band: 7.8 reads as a routine endpoint item, while confirmed exploitation plus a public PoC on a user-triggerable path to SYSTEM makes this the escalation half of a chain whose initial-access half arrives separately.",
+        "evidence": "Packet: CISA KEV-listed 2024-09-10, active_exploitation confirmed, poc_available true, CVSS 7.8, RWEP 75, CWE-269, patch_available true, live_patch_available false, live_patch_notes 'No live-patching primitive for this product; the vendor update requires a reboot and is the remediation.' Attack vector: 'A local low-privileged user triggers a Windows Installer repair, which briefly runs an elevated process; by hijacking that elevated context (or racing the rollback temp files with symlinks/oplocks) the attacker obtains a SYSTEM shell.' Citing gaps include NIST-800-53-AC-6 (Least Privilege), NIST-800-53-SI-2 (Flaw Remediation), AU-Essential-8-Patch, ISO-27001-2022-A.8.8 and NIS2-Art21-patch-management.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2",
+          "NIST-800-53-AC-6"
+        ]
+      }
+    ]
   },
   "CVE-2024-38226": {
     "name": "Microsoft Publisher Protection Mechanism Failure Vulnerability",
@@ -43739,7 +44405,30 @@
         "adequate": false,
         "gap": "Even with an available patch, the interval between the emergency Chrome release and enterprise-wide browser update rollout leaves a window that observed in-the-wild exploitation can hit."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-057",
+        "name": "BROWSER-MANAGED-UPDATE-NO-DEFERRAL",
+        "description": "The delivery path the packet describes is a user visiting a crafted HTML page, so for this CVE the browser's own update channel is the entire remediation and the enterprise update ring is the only thing that can delay it. Bound to this product, the control means the Chrome security channel is not held in a staged or piloted ring for this release: the packet records a vendor fix (builds at or above 128.0.6613.84 no longer carry the V8 defect) with no live-patching primitive and no host reboot requirement — but a browser already running when the update lands keeps the pre-fix V8 resident, because the replaced binary is inert until the browser process itself restarts. So there is no maintenance window to schedule and no host to reboot, and there is still a relaunch to force: completion is the version V8 is actually executing, not the version deployed, and a managed record reading 128.0.6613.84 against a session nobody has restarted is a host still running the vulnerable renderer. Two things therefore keep an estate exposed — a deferral the operator configured, and a long-lived browser session the policy never forced to relaunch, which is the one most likely to have met a crafted page in the interim. The clock runs from the 2024-08-28 KEV listing, not from the next scheduled ring promotion, because exploitation is already confirmed. Note what the packet does and does not support about priority: poc_available is false, so an estate that ranks work by public exploit availability will place this below flaws with published code while it is being used in the wild — the ranking input here is the confirmed-exploitation flag, not PoC presence. Precondition: removing the ring deferral reaches only browser instances a managed update policy actually governs. A personally-installed Chrome, or one whose update policy has been overridden locally, is not covered by shortening the ring; those instances are remediated by bringing them under policy or removing them. And because the packet places the heap corruption in the renderer and describes it as typically chained with a sandbox escape, the update closes the V8 defect but says nothing about a host that was already reached through a chain built on it.",
+        "evidence": "Packet: CWE-787 and CWE-358, CVSS 8.8, RWEP 54, cisa_kev true with kev_date 2024-08-28, active_exploitation 'confirmed', poc_available false. Vector: 'Inappropriate implementation in V8 in Google Chrome prior to 128.0.6613.84 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.' Attack vector: 'A user visits a crafted HTML page that exploits an inappropriate implementation in V8 to corrupt the heap in the renderer, typically chained with a sandbox escape for full code execution.' patch_available true, patch_required_reboot false; live_patch_available false with live_patch_notes 'No live-patching primitive for this product; the vendor update (no reboot required) is the remediation.' The entry's framework_control_gaps state the relaunch requirement directly — NIS2-Art21-patch-management: 'Standard patch management must be augmented with forced browser relaunch, since an updated binary is inert until the browser restarts.'",
+        "gap_closes": [
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-018",
+        "name": "SCANNER-PAPER-COMPLIANCE-TEST",
+        "description": "Chrome takes this fix through its own security channel rather than through the operating system's package inventory, so on this CVE an operating-system patching attestation is measuring something that never carried the fix — which is exactly why an OS-patching control is recorded as insufficient here. The requirement is that browser currency be measured from the version each managed Chrome instance actually reports, checked against the packet's fixed level of 128.0.6613.84, and never from a management console's 'approved' or 'pushed' state or from an OS patch report that has no row for the browser at all. Distinguishing test: pull the running browser version from every managed endpoint — Chrome against 128.0.6613.84, and each other Chromium-based product against its own vendor fixed build — and show none is below it — an estate whose OS patch compliance reads clean while its fleet report carries no browser-version column has recorded nothing whatsoever about this flaw, and that is the specific way this one is marked compliant while it stays exploitable. Scope the sweep to what the entry names, which is wider than Chrome: the defect is in V8, and the affected products are recorded as Google Chrome and other Chromium-based browsers, Edge and Opera among them, on affected V8 builds. A sweep limited to Chrome therefore reports clean across an estate standardised on Edge, and those products are checked against their own vendor fixed builds rather than against the Chrome version string, which they do not report. The Chromium lineage is also the bound: treating every browser or every JavaScript engine in the estate as an instance of this CVE manufactures remediation work against software no evidence implicates. Precondition: this check sees only instances enrolled in the reporting channel. A browser installed outside managed software distribution produces no row, and an absent row is not a pass — it is an unmeasured instance that has to be enrolled or removed before the estate figure means anything.",
+        "evidence": "Packet vector names the affected component and fixed level: 'Inappropriate implementation in V8 in Google Chrome prior to 128.0.6613.84'. The entry's affected field reads 'Google Chrome and other Chromium-based browsers: an inappropriate implementation in the V8 JavaScript engine permits heap corruption via a crafted HTML page', and affected_versions lists both 'Google Chrome prior to 128.0.6613.84' and 'Chromium-based browsers (Edge, Opera) on affected V8 builds'. patch_available true; live_patch_notes state 'No live-patching primitive for this product; the vendor update (no reboot required) is the remediation.' active_exploitation 'confirmed', cisa_kev true, kev_date 2024-08-28. AU-Essential-8-Patch (Patch operating systems) and ISO-27001-2022-A.8.8 (Management of technical vulnerabilities) are both recorded among the framework gaps citing this CVE.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8"
+        ]
+      }
+    ]
   },
   "CVE-2024-38856": {
     "name": "Apache OFBiz Incorrect Authorization Vulnerability",
@@ -46346,7 +47035,30 @@
         "adequate": false,
         "gap": "Malicious-code protection that relies on OLE/MSHTML mitigations is exactly what this security-feature bypass defeats, so signature/heuristic AV on the mitigation layer is insufficient."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-120",
+        "name": "DOWNLOAD-PROVENANCE-ENFORCEMENT",
+        "description": "The packet's delivery path is an attacker-supplied document the victim opens, and what the flaw defeats is the prompt: MSHTML bypasses the OLE/mitigation prompts, loads malicious content, and executes in the user's context. That is precisely why provenance has to drive an isolated render rather than a warning on this entry — the mechanism this CVE breaks is the one that asks the user, so a policy that tightens prompt settings is hardening the thing the packet says stops firing. Bound to this entry the requirement is that every document arriving by mail, web download or untrusted file share be marked untrusted at the ingress boundary; that a marked document open in a reduced-privilege isolated view instead of the full handler path that reaches the MSHTML platform; and that the mark survive its container — extracted from an archive, mounted from a disk image, or renamed — because a document that loses it is indistinguishable from a locally authored one and takes the full path with the user's privileges. The distinguishing test is a delivery test, not a settings export: send a document through each ingress path to a managed workstation, extract it from whatever container it arrived in, and confirm it still opens isolated. An estate whose user-application-hardening and malware-protection attestations show the OLE and macro settings configured passes cleanly while the packet's exact scenario — a crafted document that bypasses those prompts — still executes, and signature-based malware protection has no artifact to match on a bespoke document. Precondition: an isolated render bounds this path, it does not remove it. A document that reaches the full handler anyway — provenance stripped in transit, delivered from an internal share the boundary does not mark, or opened out of the isolated view by the user — reaches the same code. The packet registers no live-patch path and states the vendor update requires a reboot and is the remediation, so a host that installed the update without restarting still runs the vulnerable code, and a host that already opened such a document belongs on the incident path rather than being closed on the delivery control.",
+        "evidence": "Packet: Windows MSHTML Platform Security Feature Bypass Vulnerability (CWE-20). Attack path per the packet: \"An attacker delivers a crafted document that, when opened, causes the MSHTML platform to bypass OLE/mitigation prompts and load malicious content, leading to code execution in the user's context.\" CISA KEV-listed 2024-05-14 with confirmed in-the-wild exploitation; CVSS 8.8, RWEP 57; poc_available false; patch_available true, live_patch_available false, live_patch_notes: \"No live-patching primitive for this product; the vendor update requires a reboot and is the remediation.\" AU-Essential-8-App-Hardening (User application hardening), NIST-800-53-SI-3 (Malicious Code Protection) and ISO-27001-2022-A.8.7 (Protection against malware) are cited as insufficient on this entry.",
+        "gap_closes": [
+          "AU-Essential-8-App-Hardening",
+          "NIST-800-53-SI-3",
+          "ISO-27001-2022-A.8.7"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-041",
+        "name": "PROTECTION-MECHANISM-FAILURE-CLASS-REGRESSION",
+        "description": "The packet classifies this as a security feature bypass and names what fails: the OLE/mitigation prompts do not fire on the crafted document. The estate's evidence that its hardening works is evidence about settings, while the thing this CVE changes is whether the mechanism behind those settings still triggers — and nothing in a configuration report distinguishes the two. Bound to this entry the requirement is that the prompt-and-provenance class be re-exercised as a class after each patch deployment on this platform: the OLE/mitigation prompt path, the untrusted-origin mark, and whatever mail detonation and endpoint rules the estate relies on to catch a document that gets past them, driven with a document that takes the MSHTML platform down the path the packet describes — rather than tested once against this CVE and retired. The distinguishing test is a detonation result, not a policy export: a document exercising the mitigation-prompt path must be shown to be stopped on the build currently deployed; a report showing the prompt enabled is exactly the artifact that stayed green while this bypass was exploited in the wild, which is what makes the flaw-remediation and vulnerability-handling controls cited here insufficient — they record that an update was applied, not that the mechanism the update was supposed to restore now fires. Precondition: this is verification, not prevention. It tells the operator the mechanism has stopped working; it removes nothing, and the packet records the vendor update — which requires a reboot — as the remediation. A battery can also only cover primitives already known to it, so a bypass discovered after the battery was written passes it; its value is that it catches this class of mechanism failing without any setting changing or any signature firing, which is what the packet establishes happened here.",
+        "evidence": "Packet: the entry is a Windows MSHTML Platform Security Feature Bypass (CWE-20) in which the crafted document causes the MSHTML platform to bypass OLE/mitigation prompts and load malicious content. CISA KEV-listed 2024-05-14 with confirmed in-the-wild exploitation; CVSS 8.8, RWEP 57; poc_available false; patch_available true with live_patch_available false and the packet stating the vendor update requires a reboot and is the remediation. NIST-800-53-SI-2 (Flaw Remediation) and NIS2-Art21-vulnerability-handling are cited as insufficient on this entry.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-handling"
+        ]
+      }
+    ]
   },
   "CVE-2024-30051": {
     "name": "Microsoft DWM Core Library Privilege Escalation Vulnerability",
@@ -46785,7 +47497,42 @@
         "adequate": false,
         "gap": "The compromised device IS the network boundary; boundary protection cannot compensate when the firewall's own preauth GlobalProtect endpoint yields root RCE."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-030",
+        "name": "PERIMETER-DEVICE-ZERODAY-SLA-TIER",
+        "description": "PAN-OS running GlobalProtect is the perimeter-trust-boundary case this tier exists for, and the packet states why a generic patch SLA cannot govern it: an unauthenticated attacker sends GlobalProtect requests with a crafted SESSID cookie and ends with arbitrary code execution at root privilege on the firewall itself. The lesson's framework coverage puts it plainly — the compromised device IS the network boundary, so boundary protection cannot compensate when the firewall's pre-auth GlobalProtect endpoint yields root RCE. For this CVE the tier means the clock starts at the 2024-04-12 KEV listing, and the alternative to meeting it is taking the GlobalProtect portal and gateway interface out of service, not accepting a 14- or 30-day window. Scope the enumeration to what the packet names — PAN-OS in the affected versions with the GlobalProtect feature configured — and explicitly do not sweep Cloud NGFW, Panorama appliances or Prisma Access, which the vector states are not impacted; counting those produces emergency change work against products no evidence implicates. Precondition on the isolation half, which is where this control is most often over-claimed: the GlobalProtect portal and gateway exist to answer unauthenticated requests from the internet so remote users can connect, so isolating that interface means withdrawing remote access for the duration, and it is simply unavailable at a site where remote access is the business function. Where isolation is not available, the vendor fix on the accelerated clock is the only remaining lever. The packet records live_patch_available false and states the vendor update requires a reboot, so a firewall that has taken the fix but has not been rebooted is still running the vulnerable code and must be counted as exposed — on a device carrying production traffic that reboot is the step most likely to be deferred, because taking it interrupts the traffic the firewall is passing, and a deferral recorded as patched is the specific way this remediation goes wrong.",
+        "evidence": "Packet: CWE-20 / CWE-77 command injection arising from arbitrary file creation in the GlobalProtect feature of PAN-OS; the vector states it may enable an unauthenticated attacker to execute arbitrary code with root privileges on the firewall, for specific PAN-OS versions and distinct feature configurations, and that Cloud NGFW, Panorama appliances and Prisma Access are not impacted. cisa_kev true, kev_date 2024-04-12, active_exploitation confirmed, poc_available true, cvss 10, rwep_score 84. patch_available true, live_patch_available false, live_patch_notes: 'No live-patching primitive for this product; the vendor update requires a reboot and is the remediation.' The lesson's framework_coverage for NIST-800-53-SC-7 records that the compromised device IS the network boundary and that boundary protection cannot compensate when the firewall's pre-auth GlobalProtect endpoint yields root RCE. AU-Essential-8-Patch, ISO-27001-2022-A.8.8, NIST-800-53-SI-2, NIST-800-53-SC-7 and UK-CAF-B4 are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIST-800-53-SC-7",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet records what survives the fix here: observed chains deploy the UPSTYLE Python backdoor, which reads the device error log for further commands. That implant is a file and a running process on the firewall, placed by code that executed as root — the update closes the SESSID file-creation path but removes nothing already written through it, so a device exploited before remediation comes out of the update still implanted and still reachable by its operator. For a PAN-OS firewall the default response therefore has to be: capture the configuration and support data for analysis, rebuild the device from known-good firmware, and treat every secret the firewall held as exposed — GlobalProtect and management credentials, the directory or RADIUS service accounts it authenticated with, its certificates and private keys, and any pre-shared key on a tunnel it terminated — because the packet's stated outcome is root on the device that stored all of them. This is the state a flaw-remediation or technical-vulnerability-management attestation cannot express: both close when the fixed version is installed, and an implanted-then-patched firewall satisfies them exactly while remaining under attacker control. Distinguishing test: for any device that was internet-reachable and unpatched during the exposure window, require evidence that the rebuild-and-rotation path was taken rather than the version record — an estate that can produce a fixed version for every firewall but cannot say which of them were reachable before the fix landed has not answered the question. Precondition: a rebuild removes what is on the device, it does not reach what left it. Credentials read off the firewall before the rebuild stay valid until they are rotated, and sessions or tunnels established with them stay live — so the rotation is the part that must complete, not the reimage.",
+        "evidence": "Packet: the attack_vector states that an unauthenticated attacker sends GlobalProtect requests with a crafted SESSID cookie whose value is written to disk as an attacker-controlled filename, that a later process executes that content as an OS command with root privileges, and that observed chains deploy the UPSTYLE Python backdoor which reads the device error log for further commands. cisa_kev true, kev_date 2024-04-12, active_exploitation confirmed, poc_available true, cvss 10, rwep_score 84. patch_available true, live_patch_available false, live_patch_notes: 'No live-patching primitive for this product; the vendor update requires a reboot and is the remediation.' The lesson's response guidance records assuming full device compromise, rotating all secrets and certificates on the firewall, and rebuilding from known-good firmware, on the basis that root RCE on the perimeter device means every credential and key it held must be considered exposed. NIS2-Art21-incident-handling (Incident handling), NIST-800-53-SI-2 (Flaw Remediation) and ISO-27001-2022-A.8.8 (Management of technical vulnerabilities) are recorded as citing gaps against this entry.",
+        "gap_closes": [
+          "NIS2-Art21-incident-handling",
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-031",
+        "name": "FIREWALL-EGRESS-TELEMETRY-SEPARATE-TRUST-ZONE",
+        "description": "The packet makes the firewall's own log a component of the exploit rather than a record of it: the UPSTYLE backdoor reads the device error log for further commands, and the attacker holds root, so every log the device writes sits in the attacker's read-and-write path. Anything an investigator later asks that firewall about the incident is answered by a file the attacker could edit. For this deployment the control means the GlobalProtect authentication logs, the system and configuration logs and the traffic logs are streamed as they are written to a collector in a separate trust zone — different management plane, different credentials, different authentication path from the firewall's own — so a pre-tamper copy exists off the device. What the rules key on has to be what this chain actually emits: GlobalProtect requests carrying anomalous SESSID cookie values, files appearing under the GlobalProtect and device-telemetry temp paths that the request handling writes to, and — because the backdoor's command channel is the device's own error log rather than a network callback — outbound connections originating from the firewall's management plane to destinations it has no operational reason to reach. A rule written against appliance crashes or named exploit tooling would see none of this: the request is a well-formed GlobalProtect request, the write is performed by the device's own service, and the implant runs as a Python process on the appliance. Precondition, and it is the whole control: the off-device stream must already have been running before the compromise. A collector stood up during the investigation holds nothing from the exposure window, and root on the firewall can stop the forwarder or feed it falsified entries from that point on — so the separate-zone copy is authoritative for what it received before the attacker acted, and no further. Constrain the management plane's egress at the same time, or the same root access that reaches the log reaches the internet.",
+        "evidence": "Packet: the attack_vector states the crafted SESSID cookie value is written to disk as an attacker-controlled filename and later executed as an OS command with root privileges, and that observed chains deploy the UPSTYLE Python backdoor which reads the device error log for further commands. cisa_kev true, kev_date 2024-04-12, active_exploitation confirmed, poc_available true, cvss 10, rwep_score 84. The lesson's detection guidance records that the backdoor manipulates on-box logs, so external/network telemetry is more trustworthy than the appliance's own logs, and names anomalous files under GlobalProtect/device-telemetry temp paths, crafted SESSID cookie values, UPSTYLE artifacts and outbound connections from the management plane as the hunt indicators. NIS2-Art21-incident-handling (Incident handling) is recorded as a citing gap against this entry.",
+        "gap_closes": [
+          "NIS2-Art21-incident-handling"
+        ]
+      }
+    ]
   },
   "CVE-2024-3273": {
     "name": "D-Link Multiple NAS Devices Command Injection Vulnerability",
@@ -46822,7 +47569,31 @@
         "adequate": false,
         "gap": "No patch exists for these EoL NAS devices, so flaw remediation cannot apply; the only fix is retirement, which SI-2's patch framing does not compel."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-122",
+        "name": "EOL-ASSET-DECOMMISSION",
+        "description": "The packet closes off every disposition except removal and states it in the entry's own words: the vulnerability only affects products no longer supported by the maintainer, the vendor was contacted early and confirmed immediately that the product is end-of-life, and it should be retired and replaced. patch_available is false and the live-patch note reads 'End-of-life or unpatched product with no vendor fix and no live-patch path; isolate or decommission affected systems.' So for the DNS-320L, DNS-325, DNS-327L and DNS-340L there is no fixed build to reach at all — this is not a product where removal becomes the terminal state after an interim patch, it is one where removal is the only state, and any requirement phrased as 'every unit reports the fixed firmware' is unsatisfiable by construction. Requirement in operational terms: enumerate every unit of these four models in service, give each a dated removal or replacement schedule, and settle the migration target for the stored data first — a NAS pulled with nowhere for its contents to go is exactly how a removal date decays into an open-ended risk acceptance. Hold the scope to the four models the packet names; it ties the command-injection sink to /cgi-bin/nas_sharing.cgi on those devices and provides no mapping into other D-Link hardware, so treating the wider storage estate as an instance of this CVE manufactures replacement work against products no evidence implicates. Precondition, which is where this is routinely over-claimed on cheap appliances: because exploitation is confirmed, the exploit has been publicly disclosed, and the packet records botnets weaponizing it to drop Mirai, a unit reachable during the exposure window may already be running attacker code as root. Removing the hardware ends the exposure but does not undo it — the data the unit held and any credential stored on it are to be treated as exposed regardless of the decommission date, and a unit suspected of having been reached belongs on the incident path rather than only on the replacement schedule. Distinguishing test: for each unit, produce a removal date and the migration target for its data. A vulnerability-management record carrying these models as 'no patch available, risk accepted' with no removal date is the outcome this control exists to catch, and it is a record every patch-framed framework control on this entry will accept as complete.",
+        "evidence": "Packet: patch_available false, live_patch_available false, live_patch_notes 'End-of-life or unpatched product with no vendor fix and no live-patch path; isolate or decommission affected systems.' Vector: '** UNSUPPORTED WHEN ASSIGNED ** ... D-Link DNS-320L, DNS-325, DNS-327L and DNS-340L up to 20240403. Affected is an unknown function of the file /cgi-bin/nas_sharing.cgi of the component HTTP GET Request Handler. The manipulation of the argument system leads to command injection ... The exploit has been disclosed to the public and may be used ... NOTE: This vulnerability only affects products that are no longer supported by the maintainer. NOTE: Vendor was contacted early and confirmed immediately that the product is end-of-life. It should be retired and replaced.' attack_vector: the base64-encoded command in the 'system' parameter 'is passed to a shell and executed as root, and botnets weaponized it to drop Mirai.' cisa_kev true, kev_date 2024-04-11, active_exploitation 'confirmed', poc_available true, cvss 9.8, rwep_score 79, cwe_refs CWE-77. Citing gaps include AU-Essential-8-Patch, ISO-27001-2022-A.8.8, NIST-800-53-SI-2 Flaw Remediation and UK-CAF-B4 System security.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-054",
+        "name": "BACKUP-TIER-NETWORK-ISOLATION",
+        "description": "These are network-attached storage appliances whose purpose is holding the primary and backup copy of a household's, remote worker's or branch site's files, and the packet reaches root on them in a single unauthenticated HTTP GET: a base64-encoded command in the 'system' parameter of /cgi-bin/nas_sharing.cgi is passed to a shell and executed as root. Since the packet records no vendor fix and names isolation as one of only two available dispositions, reachability is the only lever the operator holds for as long as a unit remains in service. For the DNS-320L, DNS-325, DNS-327L and DNS-340L that means the device's HTTP interface answers only from an operator subnet or an authenticated VPN — not from the general user VLAN, and above all not from the internet through a router port-forward or a UPnP mapping, which is how NAS hardware at homes, remote-worker locations and small branch sites normally acquires its exposure. Constrain the outbound path too: the packet records botnets weaponizing this to drop Mirai, and a Mirai-class implant on a rooted NAS needs egress both to reach its controller and to ship what the device stores. Precondition, and it must never be recorded as closure: this bounds who can send the GET, it does not close the path. The request carries no credential the site issued, so every host inside the permitted segment already satisfies the attacker's only requirement, and the appliance exists to serve the client network it sits on — for a unit serving a general user VLAN there is no segment that removes the path, only isolation of that VLAN from anything that matters. It also does nothing for a unit already reached: root execution has already happened, no firmware update exists to displace what was left behind, and re-segmenting a compromised NAS leaves the implant resident and the stored data already exposed. Isolation here is a holding measure for the window before removal, not an alternative to it. Distinguishing test: from a general user VLAN and from an external address, request /cgi-bin/nas_sharing.cgi on the unit; anything that answers is within reach of the publicly disclosed exploit. 'The NAS is on the internal network' is a claim about topology, not a demonstration that the interface is unreachable from untrusted segments.",
+        "evidence": "Packet attack_vector: 'An unauthenticated attacker sends a crafted HTTP GET to /cgi-bin/nas_sharing.cgi using the hardcoded messagebus backdoor account (CVE-2024-3272) with a base64-encoded command in the system parameter; the value is passed to a shell and executed as root, and botnets weaponized it to drop Mirai.' Vector names 'D-Link DNS-320L, DNS-325, DNS-327L and DNS-340L up to 20240403', the '/cgi-bin/nas_sharing.cgi of the component HTTP GET Request Handler', and states 'It is possible to launch the attack remotely.' patch_available false, live_patch_available false, live_patch_notes 'End-of-life or unpatched product with no vendor fix and no live-patch path; isolate or decommission affected systems.' poc_available true, active_exploitation 'confirmed', cisa_kev true, kev_date 2024-04-11, cvss 9.8. Citing gaps include NIST-800-53-SC-7 Boundary Protection and NIS2-Art21-network-security (Security of network and information systems).",
+        "gap_closes": [
+          "NIST-800-53-SC-7",
+          "NIS2-Art21-network-security"
+        ]
+      }
+    ]
   },
   "CVE-2024-3272": {
     "name": "D-Link Multiple NAS Devices Use of Hard-Coded Credentials Vulnerability",
@@ -47332,7 +48103,30 @@
         "adequate": false,
         "gap": "Essential-Eight OS-patching maturity is scoped to workstations/servers and does not compel the same-day mobile-device update discipline this actively-exploited iOS/iPadOS flaw requires."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "The remediation this packet records is not one build but a set of per-train builds — iOS 16.7.6 and iPadOS 16.7.6, iOS 17.4 and iPadOS 17.4, macOS Monterey 12.7.4, macOS Sonoma 14.4, macOS Ventura 13.6.5, tvOS 17.4, visionOS 1.1 and watchOS 10.4 — so for this CVE a single minimum-version rule is the wrong instrument and 'on the latest major release' is not the test. Each device must be at or above the fixed build for the train it is actually running: a Mac held on Monterey or Ventura for application-compatibility reasons has a fix available on that train and stays exposed until it takes it, and a policy expressed only as 'Sonoma 14.4 or later' either misses those devices or reports them non-compliant for the wrong reason. Make the per-train fixed build an access condition — mail, VPN and document access denied to a device below it — rather than a stale-build row on a patch-compliance dashboard. Enumerate every train the packet names, including tvOS, visionOS and watchOS, because those are the ones that sit outside most managed-device inventories, so their exposure never reaches the report the estate reads. Precondition: the packet records no vendor live-patch mechanism and states that remediation requires applying the fixed release and rebooting, so a device that has installed the update but not restarted still runs the vulnerable kernel and must be counted as exposed, not compliant. And the access condition bounds what the estate exposes to a device below the fixed build; it does nothing for a device already compromised — this bug is reached by an attacker who already holds arbitrary kernel read and write, so a device plausibly inside the targeted chain the packet describes belongs on the incident path rather than the update path. Distinguishing test: enrol a device pinned below its own train's fixed build and confirm the policy denies it protected-resource access — an estate that surfaces the stale build on a report while the device keeps its mail and VPN access has recorded the exposure rather than removed it.",
+        "evidence": "Packet vector: 'A memory corruption issue was addressed with improved validation. This issue is fixed in iOS 16.7.6 and iPadOS 16.7.6, iOS 17.4 and iPadOS 17.4, macOS Monterey 12.7.4, macOS Sonoma 14.4, macOS Ventura 13.6.5, tvOS 17.4, visionOS 1.1, watchOS 10.4. An attacker with arbitrary kernel read and write capability may be able to bypass kernel memory protections. Apple is aware of a report that this issue may have been exploited.' attack_vector: 'As a late stage of a targeted exploit chain, an attacker who already holds arbitrary kernel read/write uses this memory-corruption flaw to defeat kernel memory protections and gain durable kernel control.' cwe_refs CWE-787; cisa_kev true with kev_date 2024-03-06; active_exploitation confirmed; cvss 7.8; rwep_score 61; poc_available false; patch_available true; live_patch_available false with live_patch_notes 'No vendor live-patch mechanism; remediation requires applying the fixed release and rebooting.'",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-056",
+        "name": "MOBILE-ENDPOINT-MDM-ENFORCED-KEV-SLA",
+        "description": "For this entry the enforcement mechanism is the load-bearing part, because the packet's remediation is a reboot-gated OS update and nothing else: no live-patch mechanism is registered, and applying the fixed release and rebooting is the whole of it. On an Apple estate that update is normally offered to the user, and the restart is the step users postpone — so an estate that 'deployed' the update but permits deferral has recorded a deployment that has not happened, on a defect KEV-listed 2024-03-06 with confirmed exploitation and Apple's own note that it may have been exploited. Drive it instead through managed device configuration with a hard deadline and an enforced restart, on a KEV-tied clock rather than the estate's routine monthly ring, and measure completion by devices restarted onto the fixed build for their train rather than by updates approved, downloaded or 'available'. Precondition: this reaches only devices actually enrolled in management. Personally-owned devices carrying organizational data, and the tvOS, visionOS and watchOS units the packet names — which are frequently enrolled in nothing — are outside the enforcement path entirely, and for those the remaining lever is withholding organizational access from a device below the fixed build, not a push the estate cannot make. Enforcement also cannot help a device that is already in an attacker's hands: this bug is used by an attacker who already holds arbitrary kernel read and write, so the enforced update closes the exposure going forward and says nothing about a device that was already carried through the chain.",
+        "evidence": "Packet: cisa_kev true with kev_date 2024-03-06; active_exploitation confirmed; vector states 'Apple is aware of a report that this issue may have been exploited' and names the fixed builds across iOS/iPadOS, macOS Monterey/Sonoma/Ventura, tvOS, visionOS and watchOS; patch_available true; live_patch_available false with live_patch_notes 'No vendor live-patch mechanism; remediation requires applying the fixed release and rebooting.'; cvss 7.8; rwep_score 61; poc_available false.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIS2-Art21-patch-management"
+        ]
+      }
+    ]
   },
   "CVE-2024-23296": {
     "name": "Apple Multiple Products Memory Corruption Vulnerability",
@@ -47389,7 +48183,39 @@
         "adequate": false,
         "gap": "Essential-Eight patch maturity is workstation/server-scoped and does not compel same-day mobile OS updates for an active mobile zero-day."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-056",
+        "name": "MOBILE-ENDPOINT-MDM-ENFORCED-KEV-SLA",
+        "description": "The packet names the fixed releases across the iOS, iPadOS, macOS, tvOS, visionOS and watchOS families — iOS 16.7.8 and iPadOS 16.7.8, iOS 17.4 and iPadOS 17.4, macOS Monterey 12.7.6, macOS Sonoma 14.4, macOS Ventura 13.6.7, tvOS 17.4, visionOS 1.1, watchOS 10.4 — and that list is this control's scope statement: an estate that drives the KEV clock opened 2024-03-06 across managed iPhones and Macs alone still leaves the tvOS, visionOS and watchOS builds the packet names running the vulnerable code, and those device classes typically sit outside the management channel that enforces the phone and Mac deadline. For this CVE the control means the update is pushed on that clock with user deferral disallowed, and completion is measured per device against the specific build the packet names for that device's OS family and track — not against an 'available' or 'downloaded' state in the management console. The packet records no live-patch mechanism and states that remediation requires applying the fixed release and rebooting, so a device that has staged the update but not restarted is still running the vulnerable code and must be counted as exposed; on a phone or a watch the restart is precisely the step a user defers, which is how this remediation gets recorded as complete while the exposure stands. Distinguishing test: enumerate the estate's Apple devices by OS family and installed build and confirm each is at or above the build the packet names for it, including the device classes the standard patch report does not cover. Precondition: this reaches only devices the management channel can see and compel — a personally owned or unenrolled device carrying organizational data is remediated by removing that access or that data, not by a deadline it never receives.",
+        "evidence": "Packet vector: 'A memory corruption issue was addressed with improved validation. This issue is fixed in iOS 16.7.8 and iPadOS 16.7.8, iOS 17.4 and iPadOS 17.4, macOS Monterey 12.7.6, macOS Sonoma 14.4, macOS Ventura 13.6.7, tvOS 17.4, visionOS 1.1, watchOS 10.4. An attacker with arbitrary kernel read and write capability may be able to bypass kernel memory protections. Apple is aware of a report that this issue may have been exploited.' CWE-787, CISA KEV-listed 2024-03-06, active_exploitation confirmed, poc_available false, CVSS 7.8, RWEP 60. patch_available true, live_patch_available false, live_patch_notes: 'No vendor live-patch mechanism; remediation requires applying the fixed release and rebooting.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-patch-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "The packet fixes this on two parallel iOS/iPadOS tracks — 16.7.8 and 17.4 — and on three separate macOS releases, Monterey 12.7.6, Sonoma 14.4 and Ventura 13.6.7. That is the condition a single minimum-version threshold cannot express, and it fails in the dangerous direction: a policy admitting anything at or above 16.7.8 admits a device running 17.0, which is numerically higher and still carries the flaw, while a policy pinned at 17.4 marks every device that correctly took 16.7.8 as non-compliant. For this CVE the control therefore means the minimum build is stated per OS family and per track, and enforced as an access condition — mail, VPN and document access withheld from a device below the build named for its own track — rather than surfaced as a row on a patch-compliance report. The distinguishing test is to enrol a device sitting above one track's fix but below its own, a 17.x device on a build under 17.4, and confirm the policy denies it access to protected resources; an estate whose compliance rule is a single version comparison passes that device while it stays exposed. Precondition: withholding access bounds what the device can reach while it is exposed, it does not remediate the device, and the packet's attack description is of a device whose attacker already holds arbitrary kernel read and write — so a device suspected of having been in that state belongs on the incident path and is not returned to service by clearing a build check. This is a holding measure for the window before the fixed release and its required reboot land, not a substitute for them.",
+        "evidence": "Packet vector names the fixed builds on parallel tracks: 'iOS 16.7.8 and iPadOS 16.7.8, iOS 17.4 and iPadOS 17.4, macOS Monterey 12.7.6, macOS Sonoma 14.4, macOS Ventura 13.6.7, tvOS 17.4, visionOS 1.1, watchOS 10.4', and states 'An attacker with arbitrary kernel read and write capability may be able to bypass kernel memory protections.' attack_vector: the attacker already holds arbitrary kernel read/write and uses this RTKit memory-corruption flaw to bypass kernel memory protections, a late stage of a targeted exploit chain. CWE-787, KEV 2024-03-06, active_exploitation confirmed. patch_available true with live_patch_available false; remediation requires applying the fixed release and rebooting.",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-121",
+        "name": "MOBILE-ZERO-CLICK-HARDENING",
+        "description": "The packet places this flaw at a late stage of a targeted exploit chain: the attacker already holds arbitrary kernel read and write, and uses this RTKit memory-corruption bug to bypass kernel memory protections. Nothing an operator configures reduces that primitive once the chain has reached this point, and the control must not be recorded as if it did. What it means for this CVE is the earlier stage — for the cohort plausibly inside the targeting set chains of this class serve, place the device in a reduced-attack-surface posture so untrusted web content, message attachments, fonts and link previews are not processed automatically, narrowing the delivery of the stages that produce the kernel read/write this bug consumes. Two conditions have to be stated or this becomes a mitigation on paper only. First, the posture has to already be in force: assigned in response to a disclosure it gives nothing for the window that mattered, and the packet's own record is Apple aware of a report that the issue may have been exploited — that is after the fact by construction. Second, it does not evict an attacker already resident; a device believed to have carried the chain goes to the incident path rather than into a hardening rollout, and the packet's description of an attacker with arbitrary kernel read and write is the state in which no device-side setting is trustworthy. Distinguishing test: name the high-risk cohort and show the reduced-attack-surface posture was applied to each of those devices before the disclosure date, not after — a policy that exists but was assigned reactively passes an attestation while providing nothing during the exposure.",
+        "evidence": "Packet attack_vector: 'An attacker already holding arbitrary kernel read/write uses this RTKit memory-corruption flaw to bypass kernel memory protections, a late stage of a targeted exploit chain.' Packet vector: 'An attacker with arbitrary kernel read and write capability may be able to bypass kernel memory protections. Apple is aware of a report that this issue may have been exploited.' CWE-787, CISA KEV-listed 2024-03-06, active_exploitation confirmed, poc_available false, CVSS 7.8, RWEP 60. patch_available true, live_patch_available false — remediation requires applying the fixed release and rebooting. The entry cites NIST SP 800-53 Rev 5 CM-7 (Least Functionality) among its insufficient framework controls.",
+        "gap_closes": [
+          "NIST-800-53-CM-7"
+        ]
+      }
+    ]
   },
   "CVE-2023-21237": {
     "name": "Android Pixel Information Disclosure Vulnerability",
@@ -49059,7 +49885,30 @@
         "adequate": false,
         "gap": "Technical-vulnerability management does not chain to secrets rotation, leaving leaked DB credentials valid after the Joomla update."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet establishes the fact this control exists for: the unauthenticated request returns the application configuration including the MySQL username, password and host, and those credentials are reused for deeper access. The vendor fixed release stops the endpoint from serving that configuration; it does not invalidate a credential that has already been served. With a public PoC and confirmed in-the-wild exploitation against a version range as wide as 4.0.0 through 4.2.7, any site that was reachable before it was upgraded has to be treated as having disclosed those secrets. Bound to this entry the requirement is that the response for an exposed site is upgrade plus rotation, in that order and both recorded: rotate the database credential and every other secret carried in the site configuration, then review what that database account could reach and what was done with it across the exposure window, scoping the review from the site's own request logs rather than from whether anything looks wrong in the CMS — nothing needs to look wrong, because the exploit is a well-formed GET that returns 200. Note which half of this control the packet supports: it establishes configuration disclosure and credential reuse, not code execution on the web server, so the rebuild default the control specifies for a pre-auth RCE is not what this entry calls for; the config-exfiltration and credential-rotation half is, and it is the half the cited patching and vulnerability-management controls miss entirely, because their completion criterion is the installed version. Precondition: rotation closes the disclosed credential and nothing else. If the deeper access the packet describes was already used to establish something durable — an application account, or a change to stored content — rotating the database password does not remove it, and a site that cannot account for its exposure window needs its administrative accounts and content compared against a known-good state rather than being closed on the upgrade record.",
+        "evidence": "Packet: Joomla! 4.0.0 through 4.2.7; \"An improper access check allows unauthorized access to webservice endpoints\" (CWE-284). Attack path per the packet: \"An unauthenticated attacker appends ?public=true to Joomla 4.x REST config/user endpoints; the improper access check returns the application configuration, including MySQL username, password, and host, which is reused for deeper access.\" CISA KEV-listed 2024-01-08 with confirmed in-the-wild exploitation; poc_available true; CVSS 5.3 against RWEP 68; patch_available true, live_patch_available false, live_patch_notes: \"No vendor live-patch mechanism; remediation requires applying the vendor fixed release.\"",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-vulnerability-management",
+          "AU-Essential-8-Patch"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-129",
+        "name": "ENTERPRISE-ERP-MANAGEMENT-FUNCTION-AUTHENTICATION",
+        "description": "Read past the product class in this control's name to what it requires — a management function that authorizes its own caller instead of inheriting a verdict — because the Joomla webservice layer is where this product takes its access decision and this CVE is that decision being made from the request. The packet has an unauthenticated caller appending ?public=true to the 4.x REST config and user endpoints, and the improper access check handing back the application configuration. Bound to this product, the control means each webservice endpoint authorizes its caller itself before the handler runs and before any configuration value is serialized into a response, rather than honouring a flag the caller supplies; and the REST API surface answers only from networks with an operational need for it, so an arbitrary internet client cannot present the request at all. The access-enforcement and identity-and-access controls cited on this entry are cited precisely because they do not reach this path: the attacker never holds a Joomla account, so per-account privilege scoping is never consulted, and an attestation showing every administrator authenticates with correctly scoped roles passes cleanly while an unauthenticated GET returns the database password. Distinguishing test: from a network with no operational need for the API, issue unauthenticated requests to the config and user webservice endpoints on a staging site with the caller-supplied public flag set, and confirm each is refused before the handler runs and before anything is emitted — testing that the administrator login requires authentication demonstrates nothing about this path. Precondition: the endpoint-side check is a property the vendor fixed release establishes; this control states what to verify, it does not implement it. On a site still in the 4.0.0-4.2.7 range, restricting reachability bounds who can send the request but closes nothing — anything inside the permitted set still receives the configuration — and that restriction is unavailable where the site's own integrations consume those endpoints from outside a trusted segment.",
+        "evidence": "Packet: Joomla! 4.0.0 through 4.2.7, improper access check allowing unauthorized access to webservice endpoints (CWE-284); the unauthenticated request path is ?public=true against the Joomla 4.x REST config/user endpoints, returning the application configuration. CISA KEV-listed 2024-01-08 with confirmed in-the-wild exploitation; poc_available true; CVSS 5.3, RWEP 68; patch_available true with live_patch_available false and remediation stated as applying the vendor fixed release. NIST-800-53-AC-3 (Access Enforcement) and UK-CAF-B2 (Identity and access control) are cited as insufficient on this entry.",
+        "gap_closes": [
+          "NIST-800-53-AC-3",
+          "UK-CAF-B2"
+        ]
+      }
+    ]
   },
   "CVE-2016-20017": {
     "name": "D-Link DSL-2750B Devices Command Injection Vulnerability",
@@ -52231,7 +53080,21 @@
         "adequate": false,
         "gap": "System-security controls do not constrain outbound requests the collaboration server is allowed to make, so the SSRF-driven internal scanning is not prevented by the existing boundary posture."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "This entry is the case the control exists for, because the two numbers on it point in opposite directions: CVSS 5.3 puts it in the band most vulnerability-management programs park on a routine medium-severity clock, while the packet records it KEV-listed 2023-10-10 with confirmed in-the-wild exploitation, a public PoC, and RWEP 66. For this CVE the control means the KEV listing sets the clock on every Skype for Business Server the organization runs, not the severity band — and the reason the band misleads is specific: the impact the 5.3 reflects is disclosure, but the packet describes an unauthenticated request forcing the server to make an outbound HTTP request that reveals internal IP and port information from the perimeter. That is reconnaissance feeding a later intrusion, not the end of one, and it is available continuously because the server is internet-facing by design.\n\nRemediation, per the packet: patch_available is true, live_patch_available is false, and the recorded note is that remediation requires applying the vendor fixed release. There is nothing to run in place while the change waits for the next server-maintenance window, so the interval between the KEV listing and the fixed release actually being applied is exposure rather than planning time.\n\nDistinguishing test: produce, per Skype for Business Server instance, the date the fixed release was applied measured against 2023-10-10 — not the date the finding was closed in the tracker. A program that triages by CVSS records this as within SLA on a medium-severity clock while a KEV-listed, actively exploited flaw stays live on an internet-facing server for the whole of that window, and the attestation reads clean throughout.\n\nPrecondition, and it is why this is a scheduling control rather than a closure: an expedited clock governs when the fix lands; it does not constrain what the server is permitted to request outbound, so for the whole pre-patch window the forced-request path stays fully open and no amount of prioritization narrows it. And because exploitation is confirmed and the flaw's output is internal topology rather than an artifact left on the host, applying the fixed release stops future probes but does not retract addressing and port information an attacker already collected — a server that was internet-reachable during the window should have that reconnaissance treated as attacker-held, rather than the entry being closed on the patch record.",
+        "evidence": "Packet facts only. CISA KEV-listed 2023-10-10; active_exploitation 'confirmed'; poc_available true; CVSS 5.3; RWEP 66; CWE-918; ai_discovered false. Vector: 'Skype for Business Elevation of Privilege Vulnerability'. Attack vector: 'An unauthenticated attacker sends a crafted request to an internet-facing Skype for Business Server that forces the server to make an outbound HTTP request, disclosing internal IP/port information and enabling internal-network reconnaissance from the perimeter.' patch_available true; live_patch_available false; live_patch_notes: 'No vendor live-patch mechanism; remediation requires applying the vendor fixed release.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-vulnerability-management"
+        ]
+      }
+    ]
   },
   "CVE-2023-36563": {
     "name": "Microsoft WordPad Information Disclosure Vulnerability",
@@ -53582,7 +54445,28 @@
         "adequate": false,
         "gap": "Identity and access control expects strong authentication yet does not compel the tunnel-group hardening (no-MFA default group disabled, rate-limiting) needed to stop credential brute forcing on the appliance."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-131",
+        "name": "PERIMETER-VPN-GATEWAY-AUTH-BYPASS-EXPEDITED-REMEDIATION",
+        "description": "ASA and FTD are the perimeter's remote-access authentication enforcement point, and the packet places this defect inside that enforcement rather than around it: AAA is not properly separated between the RA VPN feature and the HTTPS-management and site-to-site VPN features, so specifying a default connection profile/tunnel group lets an unauthenticated attacker run credential guessing directly against the gateway, and on ASA 9.16 and earlier lets a clientless SSL VPN session be established with an unauthorized user. Be precise about which half is which, because the packet is: it states the flaw does not allow bypassing authentication, and that valid credentials — including a valid second factor where MFA is configured — are still required to establish a remote-access VPN session. The expedited clock is therefore warranted not by a full authentication bypass but by what the gateway gives an unauthenticated attacker for free, which is unlimited credential discovery against the enforcement point itself, plus the authorization failure on 9.16 and earlier.\n\nFor this CVE the clock runs from the 2023-09-13 KEV listing through the reload of every affected ASA and FTD unit. The packet records patch_available true, live_patch_available false, and remediation as applying the fixed release and rebooting — so a unit that has taken the release but has not been reloaded still runs the vulnerable code and must be counted as exposed, not as patched. On a production VPN concentrator that reload is the step most likely to be deferred, because taking it drops live tunnels, and a deferral recorded as 'patched' is the specific way this remediation goes wrong. Enumerate units on ASA 9.16 or earlier first: those carry the clientless-session path on top of the credential-discovery path.\n\nDistinguishing test: report, per ASA and FTD unit, the software version the device is actually running and the timestamp of its last reload, rather than the version staged from the management console — a fleet where the console shows the fixed release pushed everywhere passes a patch attestation while unreloaded devices keep serving the vulnerable code.\n\nPrecondition on interim exposure, which is where this is normally over-claimed: the packet's stated exploitation requirement is reaching a default connection profile/tunnel group on the RA VPN feature. Restricting who can reach that profile bounds the population that can attempt it, but a remote-access VPN exists to answer from untrusted networks, so on any gateway serving general remote access there is no source restriction that removes the path — only the fixed release and the reload do. Interim measures narrow the attempt rate; they do not close it.\n\nAnd what the reload does not do: the flaw's product is valid username and password combinations identified by brute force, and the packet records Akira ransomware using this for initial access. Credentials harvested during the exposure window remain valid after the fixed release is applied — the update repairs the AAA separation, it does not invalidate what was already taken — so a gateway whose RA VPN was reachable during the window needs those accounts rotated and established sessions reviewed, not closed on the patch record.",
+        "evidence": "Packet facts only. CISA KEV-listed 2023-09-13; active_exploitation 'confirmed'; CVSS 9.1; RWEP 62; poc_available false; CWE-288 and CWE-863; ai_discovered false. Vector: the RA VPN feature of Cisco ASA and FTD 'could allow an unauthenticated, remote attacker to conduct a brute force attack in an attempt to identify valid username and password combinations or an authenticated, remote attacker to establish a clientless SSL VPN session with an unauthorized user'; 'due to improper separation of authentication, authorization, and accounting (AAA) between the remote access VPN feature and the HTTPS management and site-to-site VPN features'; exploited 'by specifying a default connection profile/tunnel group'; clientless SSL VPN session 'only when running Cisco ASA Software Release 9.16 or earlier'; and the packet's own notes that 'This vulnerability does not allow an attacker to bypass authentication. To successfully establish a remote access VPN session, valid credentials are required, including a valid second factor if multi-factor authentication (MFA) is configured.' Attack vector records 'Akira ransomware used this for initial access'. patch_available true; live_patch_available false; live_patch_notes: 'No vendor live-patch mechanism; remediation requires applying the fixed release and rebooting.'",
+        "gap_closes": [
+          "AU-ISM-1546",
+          "NIS2-Art21-network-security"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-031",
+        "name": "FIREWALL-EGRESS-TELEMETRY-SEPARATE-TRUST-ZONE",
+        "description": "ASA and FTD are perimeter firewalls, and for this CVE the gateway is also the only place the attack is visible before it succeeds. The packet gives the exploitation shape precisely enough to key on: an unauthenticated attacker specifies a default connection profile/tunnel group and conducts a brute-force attack to identify valid username and password combinations, then uses those credentials to establish a remote-access VPN session. The successful login is not the signal — the packet states the flaw does not bypass authentication and that valid credentials are required, so that session looks exactly like a legitimate one on the wire and in the logs. The distinguishing signal is the failed-authentication volume against the default connection profile on the RA VPN feature immediately preceding it, from a source that then authenticates successfully. That correlation — burst of failed RA VPN authentications against a default tunnel group, followed by a successful RA VPN authentication for one of the guessed accounts — is what the rule keys on.\n\nApplied to this device, the control means the gateway's authentication logs, not only its traffic and threat logs, are forwarded to a SIEM in a separate trust zone: the correlation is over time and across attempts and across units, which is not something a single appliance's local view supports, and the appliance is itself the asset under attack rather than a sensor watching one.\n\nAsk what an exploit doing exactly what the packet describes emits, because it is very little: no malware, no file write, no process crash, no exploit tooling on any host — only VPN authentication records on the gateway, then a valid VPN session. A rule keyed on endpoint telemetry, malware signatures, or appliance instability sees none of it, and an authentication-monitoring posture that only counts successes sees the intrusion but not the attack that produced it.\n\nPreconditions, and they are why this is a compensating measure rather than a fix. The logs must already be shipping and the volume rule must already exist — telemetry configured after the intrusion produces nothing, and on a perimeter gateway authentication-log forwarding is exactly what is most often left at defaults. Detection does not prevent the brute force from succeeding; it bounds the exposure to alert-and-response time during the window before the fixed release and the reload land, and it supplies the account list to rotate afterwards. It also does not cover the clientless SSL VPN path the packet attributes to ASA 9.16 and earlier, where a session is established with an unauthorized user using valid credentials — no failed-attempt burst necessarily precedes that, so this rule would not distinguish it from a normal session.",
+        "evidence": "Packet facts only. Vector and attack_vector state that an unauthenticated remote attacker can 'conduct a brute force attack in an attempt to identify valid username and password combinations' by 'specifying a default connection profile/tunnel group', that the identified credentials 'could then be used to establish an unauthorized remote access VPN session', and that on ASA Software Release 9.16 or earlier a clientless SSL VPN session can be established with an unauthorized user 'using valid credentials'; the packet's notes state the flaw 'does not allow an attacker to bypass authentication' and that valid credentials remain required. CISA KEV-listed 2023-09-13 with active_exploitation 'confirmed'; the attack_vector records Akira ransomware using this for initial access. patch_available true; live_patch_available false; live_patch_notes: 'No vendor live-patch mechanism; remediation requires applying the fixed release and rebooting.'",
+        "gap_closes": [
+          "NIS2-Art21-network-security"
+        ]
+      }
+    ]
   },
   "CVE-2023-4863": {
     "name": "Google Chromium WebP Heap-Based Buffer Overflow Vulnerability",
@@ -55004,7 +55888,33 @@
         "adequate": false,
         "gap": "A.8.8 technical-vulnerability management would flag CVE-2023-24489 only once tracked as KEV; the low-key May release meant many asset owners had no ticket until August, so the padding-oracle-to-web-shell chain stayed live on unpatched controllers."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The packet's exploitation path does not end at code execution, it ends at a file on disk: an unauthenticated attacker exploits the AES-CBC padding oracle in the StorageZones Controller, forges valid-padding encrypted parameters, chains a path traversal, and uploads an .aspx web shell into the IIS webroot via /documentum/upload.aspx. The remediation the packet records — upgrading the ShareFile StorageZones Controller to 5.11.24 or later and recycling the IIS application pool — replaces product code and restarts the worker process. Neither step deletes an .aspx file an attacker already wrote into the webroot, and neither invalidates any credential the application-pool identity holds, so the upgraded controller keeps serving the shell through the patched application. Applied to this product the requirement is that any customer-managed StorageZones Controller reachable during the exposure window is handled as compromised rather than as patched: export and preserve its configuration, rebuild the host at the fixed build from a known-good image instead of upgrading in place, rotate the credentials the controller's service identity holds, and enumerate the IIS webroot and the directories it serves against a known-good installation for .aspx files that are not part of the product. Distinguishing test: on a controller that has already been upgraded, list the webroot's .aspx files and account for each one — an attestation that every StorageZones Controller reports 5.11.24 or later passes cleanly with a web shell dropped before the upgrade still present. Precondition, and this is where the control gets over-claimed: it is an incident-response default, not a preventive measure. It does nothing for a controller still running a vulnerable build, since only the upgrade closes the padding oracle, and a rebuild helps only if the restore point predates the exposure window — where the stored content forces a later restore point, the artifact hunt above is the remaining lever, not the rebuild.",
+        "evidence": "Packet attack_vector: 'Unauthenticated attacker exploits an unauthenticated-AES-CBC padding oracle in the StorageZones Controller, forges valid-padding encrypted parameters, and chains a path traversal to upload an .aspx web shell to the IIS webroot via /documentum/upload.aspx, achieving remote code execution.' Packet live_patch_notes: 'No live-patch mechanism; remediation is upgrading the ShareFile StorageZones Controller to version 5.11.24 or later and recycling the IIS application pool.' patch_available true, live_patch_available false. Vector: 'A vulnerability has been discovered in the customer-managed ShareFile storage zones controller which, if exploited, could allow an unauthenticated attacker to remotely compromise the customer-managed ShareFile storage zones controller.' cisa_kev true, kev_date 2023-08-16, active_exploitation 'confirmed', poc_available true, cvss 9.8, rwep_score 74, cwe_refs CWE-284. Citing gaps include AU-Essential-8-Patch, ISO-27001-2022-A.8.8 Management of technical vulnerabilities, NIST-800-53-SI-2 Flaw Remediation and NIS2-Art21-patch-management.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-patch-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "The packet's vector names the customer-managed ShareFile storage zones controller, which is the fact that decides who owns the clock: none of this fix arrives through the vendor's cloud service, so the operator takes the upgrade to 5.11.24 or later and the IIS application-pool recycle themselves, on every controller they run, and the packet records no live-patch path that would let them defer the restart. Applied to this CVE the control means that clock runs from the 2023-08-16 KEV listing rather than from the next scheduled application-patch window: the flaw needs no credential, the PoC is public, exploitation is confirmed, and /documentum/upload.aspx answers an unauthenticated request, so a controller left on a vulnerable build across even a standard two-week application SLA is exposed for the whole window with no authentication step in front of it. The control's 'whichever is later' clause is the part that matters on this entry — the vendor's fixed build and the KEV listing are separate events, and an SLA keyed only to a vendor release starts counting from a build an operator may never have registered as security-relevant, whereas the KEV date is an unambiguous trigger. Distinguishing test: produce, per controller, the installed StorageZones Controller version and the timestamp of the application-pool recycle that followed it, measured against 2023-08-16. A patch report showing the fixed build with no recorded recycle does not demonstrate that the patched code is the code running, because the packet makes the recycle part of the remediation and offers no live-patch alternative. Precondition: meeting this SLA closes the exploit path forward only. It says nothing about a controller that was already reached during the window, which is the separate incident-response requirement recorded against this entry.",
+        "evidence": "Packet: cisa_kev true, kev_date 2023-08-16, active_exploitation 'confirmed', poc_available true, cvss 9.8, rwep_score 74. patch_available true, live_patch_available false, live_patch_notes 'No live-patch mechanism; remediation is upgrading the ShareFile StorageZones Controller to version 5.11.24 or later and recycling the IIS application pool.' Vector: 'A vulnerability has been discovered in the customer-managed ShareFile storage zones controller which, if exploited, could allow an unauthenticated attacker to remotely compromise the customer-managed ShareFile storage zones controller.' attack_vector places the unauthenticated upload at '/documentum/upload.aspx'. Citing gaps include AU-Essential-8-Patch (ASD Essential Eight), NIS2-Art21-patch-management, NIST-800-53-SI-2 Flaw Remediation and UK-CAF-B4 System security.",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2023-38180": {
     "name": "Microsoft .NET Core and Visual Studio Denial-of-Service Vulnerability",
@@ -57336,7 +58246,38 @@
         "adequate": false,
         "gap": "A.8.8 technical-vulnerability management should have flagged the Roundcube advisory, but organisations running webmail as a secondary service frequently omit it from the asset/patch register, so CVE-2021-44026 stayed open on exactly the mail servers APT28 targeted."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-085",
+        "name": "DB-ABSTRACTION-LAYER-PARAMETERIZATION-VERIFICATION",
+        "description": "The packet puts the defect exactly where this control requires verification: crafted values submitted in Roundcube's mail search and search_params are concatenated into a SQL query without proper escaping. The property to establish is therefore at the layer that builds the search query — the search terms bound as parameters rather than interpolated into the statement — and it must be established by reading that layer's behaviour, not inferred from an input filter in front of it or from a WAF on the webmail host. That substitution is the specific failure the CAF B4 citation on this entry describes: securely configuring the host, its TLS and its service accounts leaves the injection reachable, because the malformed value is carried inside a request the application is supposed to accept. Remediation per the packet is upgrading Roundcube Webmail to 1.3.17 or 1.4.12 or later, and the packet records this as an application update with no host reboot required — so on this entry the reason a KEV item usually slips its window, a reboot that has to be scheduled against a mail outage, does not apply. Distinguishing test, run against the version actually in service rather than the version on the asset register: submit a search request whose search / search_params values carry SQL metacharacters to a staging instance and confirm the query layer binds them as values instead of concatenating them into the statement. An instance sitting behind a rule that blocks the obvious payload shapes still concatenates, and the next encoding the rule does not match reaches the same sink. Preconditions: parameterization is a property of the shipped code, so the upgrade is what establishes it — this control is the verification and the gate, and it gives nothing on its own to an instance that has not been upgraded. A WAF rule on the search endpoint is a holding measure bounded to the request shapes it recognizes and to traffic that traverses it, and it does nothing for a request reaching the application by another route. Neither the upgrade nor the rule undoes a read that already happened: the packet has the injection reading the webmail database including mail, contacts and credentials, and those credentials stay valid on every service they were reused against until they are rotated.",
+        "evidence": "Packet attack_vector: 'An attacker submits crafted values in Roundcube's mail search / search_params, which are concatenated into a SQL query without proper escaping, allowing injection that reads or manipulates the webmail database (mail, contacts, credentials).' CWE-89; CVSS 9.8; RWEP 66; poc_available true; active_exploitation confirmed; CISA KEV 2023-06-22. Packet live_patch_notes: 'No live-patch mechanism; remediation is upgrading Roundcube Webmail to 1.3.17 or 1.4.12 (or later), an application update with no host reboot required.' The repo lesson entry's framework_coverage for UK-CAF-B4 records that 'CAF B4 secure-configuration would not stop a code-level SQLi in Roundcube's search: the injection is in the application's query construction, so hardening the host or TLS does nothing against the malformed _search/search_params input.'",
+        "gap_closes": [
+          "UK-CAF-B4",
+          "NIST-800-53-SI-2"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-040",
+        "name": "OWA-PER-REQUEST-SIEM-INGESTION",
+        "description": "The requirement this control carries is webmail request-level telemetry rather than authentication telemetry, and Roundcube's search flow is where it has to be applied for this CVE. The packet's exploitation is a crafted search request: the attacker's content sits in the search / search_params values of a request the webmail application is built to serve, so nothing appears in the authentication record — no failed login, no privilege change, no new session anomaly — and the lesson records that the campaign reached the search flow through legitimate sessions. What must be forwarded off the webmail host, to a collector the webmail service cannot write to, is the per-request web-tier access record including the search parameter values, together with query-level logging from the Roundcube database, retained long enough to cover the interval between exposure and discovery rather than the days a default web-server log rotation keeps. Detection keys on what the packet documents and nothing else: search requests whose parameter values carry SQL syntax, and queries against the mail store whose scope exceeds the requesting session's own mailbox. Both halves matter — SQL-shaped text in a search box can be a user typing a quotation mark, and a broad query can be a maintenance job; it is the pairing inside one session that distinguishes the exploit. Note what will not see it: no file is written, no process is spawned, no authentication fails, and a successful injection returns an ordinary response — so file-integrity monitoring, endpoint anti-malware and failed-login alerting have no artifact to match, and a rule keyed on server errors would miss a read that succeeds. This closes the incident-handling gap directly, because the notification clock starts at awareness and a quiet database read produces no awareness at all. Preconditions: the telemetry has to be collected and shipped off-host before the attempt — a query written afterwards against logs nobody retained produces nothing, and self-hosted webmail is exactly where request-level logging is least often exported. And this detects; it does not prevent. Mail and contacts read during the window stay read, and credentials the packet places in that database stay usable until they are rotated.",
+        "evidence": "Packet attack_vector: the attacker 'submits crafted values in Roundcube's mail search / search_params', and the injection 'reads or manipulates the webmail database (mail, contacts, credentials)'. CISA KEV 2023-06-22; active_exploitation confirmed; poc_available true. The repo lesson entry records privileges_required as 'none per CVSS scoring (PR:N)' while noting the operators reached the search flow through spearphishing-driven sessions; its defense_chain detection section names 'Query-level logging/alerting on the Roundcube database and web-tier detection of SQL fragments in _search/search_params' as what would have worked, with the adequacy note that 'Detection is what most ... victims lacked, letting the espionage read go unnoticed', and its framework_coverage for NIS2-Art21-incident-handling records that 'a successful SQLi read of the Roundcube mail store is quiet exfiltration; without query-level logging on the webmail DB ... leaves little trace to trigger the incident process.'",
+        "gap_closes": [
+          "NIS2-Art21-incident-handling"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-001",
+        "name": "CISA-KEV-RESPONSE-SLA",
+        "description": "On this entry the KEV clock is both the cheapest control to satisfy and the one the estate most often fails, and the packet says why in its own remediation note. The fix is an application upgrade to 1.3.17 or 1.4.12 or later with no host reboot required, so the mitigation this control demands within the KEV window is an in-place application update on an internet-facing service — not a firmware cycle, not a maintenance window negotiated against a mail outage. That removes the usual justification for a 30-day interpretation of 'appropriate timescales', which is the undefined interval the ISO A.8.8 citation on this entry rests on, and it is why a 4-hour-class response is realistic here rather than aspirational. The Essential Eight patch citation fails for a different reason worth stating separately: Roundcube is an application running on the host, frequently maintained outside whatever pipeline pushes operating-system updates, so an attestation that measures OS patch level reads clean on a host whose webmail has been years behind since the fix shipped. Verified mitigation means the version read from the running instance is at or above 1.3.17 on the 1.3 branch or 1.4.12 on the 1.4 branch, or a later release — read from the instance, not from an inventory record, because the register omission is the documented failure mode for this class of asset. Precondition: reaching the minimum build named in the packet closes this defect only and says nothing about anything found in Roundcube since, so the requirement is a current supported release rather than the floor, and a site that was exposed while unpatched still needs its stored mail credentials rotated — the clock governs how fast the injection path closes, not what was read through it.",
+        "evidence": "Packet: CISA KEV listing 2023-06-22; active_exploitation confirmed; poc_available true; CVSS 9.8; RWEP 66; patch_available true; live_patch_available false. Packet live_patch_notes: 'No live-patch mechanism; remediation is upgrading Roundcube Webmail to 1.3.17 or 1.4.12 (or later), an application update with no host reboot required.' The repo lesson entry's framework_coverage records for ISO-27001-2022-A.8.8 that 'organisations running webmail as a secondary service frequently omit it from the asset/patch register', and for AU-Essential-8-Patch that 'Essential 8 patch cadence targets vendor-supplied fixes, but internet-facing webmail is often community-maintained and slips patch windows'; its defense_chain prevention adequacy note records that 'the recurring failure was operational — webmail left off the patch inventory.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8"
+        ]
+      }
+    ]
   },
   "CVE-2016-9079": {
     "name": "Mozilla Firefox, Firefox ESR, and Thunderbird Use-After-Free Vulnerability",
@@ -57616,7 +58557,21 @@
         "adequate": false,
         "gap": "A.8.8 technical-vulnerability management scores by CVSS/priority, but an 8.8 rating understates a confirmed zero-day under active exploitation; teams deprioritising 'High (not Critical)' browser CVEs left the type-confusion exploitable while it was already being used."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-057",
+        "name": "BROWSER-MANAGED-UPDATE-NO-DEFERRAL",
+        "description": "The exploitation path here is a crafted HTML/JavaScript page — no install, no attachment, no privilege — so every managed workstation is exposed the moment a user browses, and the only variable an operator controls is which build the browser process is executing. Bound to this CVE, the control means the enterprise update ring carries Chrome to 114.0.5735.110 on Windows and 114.0.5735.106 on macOS/Linux without sitting in a pilot or staging tier: the packet records patch_available true but live_patch_available false, so there is no interim protected state and no vendor mitigation rule to run while a ring soaks — a machine is either on the fixed build or fully exposed to the page.\n\nThe second half is where this remediation is routinely recorded as complete while the estate stays exposed, and the packet states it outright: remediation is the auto-update 'followed by a browser restart to load the patched build'. The update lands on disk without the running process taking it, so a software inventory that reads the installed version reports the fixed build while every already-running browser keeps executing the vulnerable V8. The machines this misses are precisely the worst ones — the long-lived sessions that are never closed and never rebooted. Completion must therefore be measured on the build the live browser process reports, and the relaunch has to be forced by policy rather than left to a dismissible prompt.\n\nDistinguishing test: on a workstation that has been running since before the update, read the version the live browser process reports rather than the version on disk, and confirm it is at or above the fixed build. A fleet whose software inventory shows 114.0.5735.110 everywhere while running processes report a pre-fix build satisfies a patch-cadence attestation and still executes the crafted page.\n\nScope: the packet ties the type confusion to V8 in Chrome/Chromium and names no other product, so this covers the Chrome and Chromium installs in the estate. Extending it to other browsers or to embedded renderers would manufacture removal and update work against software no evidence in this packet implicates.\n\nWhat it does not do: forcing the relaunch closes this defect only. It does not reduce the renderer attack surface the page reaches, and the packet describes the heap corruption as being chained with a renderer sandbox escape — so a workstation that browsed untrusted content while below the fixed build belongs on the incident path rather than being closed on the update record.",
+        "evidence": "Packet facts only. CISA KEV-listed 2023-06-07; active_exploitation 'confirmed'; poc_available true; CVSS 8.8; RWEP 66; ai_discovered false. Vector: 'Type confusion in V8 in Google Chrome prior to 114.0.5735.110 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page. (Chromium security severity: High)'. Attack vector: 'A crafted HTML/JavaScript page forces V8 to misinterpret an object's type, corrupting the heap into read/write primitives that, chained with a renderer sandbox escape, yield code execution.' patch_available true; live_patch_available false; live_patch_notes: 'No live-patch mechanism; remediation is Chrome/Chromium auto-update to 114.0.5735.110 (Windows) or 114.0.5735.106 (macOS/Linux) followed by a browser restart to load the patched build.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "NIST-800-53-SI-2"
+        ]
+      }
+    ]
   },
   "CVE-2023-33009": {
     "name": "Zyxel Multiple Firewalls Buffer Overflow Vulnerability (CVE-2023-33009)",
@@ -57872,7 +58827,32 @@
         "adequate": false,
         "gap": "A.8.8 technical-vulnerability management prioritizes by score, but the EPSS ~0.999 preauth SQLi with confirmed ransomware use and supply-chain reach means only removing internet exposure of MOVEit before the fix could have helped; risk-rated scheduling was too slow for a zero-day mass-exploitation event."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-032",
+        "name": "PERIMETER-COMPROMISE-REBUILD-NOT-PATCH",
+        "description": "The MOVEit Transfer web application answers unauthenticated callers over HTTP or HTTPS, and the packet's exploitation path does not stop at the SQL injection: the attacker writes the LEMURLOOT web shell (human2.aspx) into the application and uses it to enumerate and exfiltrate the files the transfer server holds together with its Azure storage keys. Bound to this product, the control means an instance that was network-reachable during the exploitation window is handled as a compromised host rather than as a patch ticket — hunt the web root for human2.aspx and any other file written through the injection, rebuild from a known-good baseline where one is found, and rotate everything the instance held or brokered: the Azure storage keys the packet names, the credentials the web application uses against its MySQL, Microsoft SQL Server or Azure SQL back end, and the accounts of counterparties whose files it stored. The packet's own remediation note states both halves in order — upgrade to a fixed release, hunt for the LEMURLOOT web shell, then restart the service — and it is the second half the cited flaw-remediation and technical-vulnerability-management controls do not carry: each of them closes when the installed version reaches the fixed build, so a server reporting a fixed release while human2.aspx still answers passes those attestations with the attacker's access intact. Distinguishing test: for each MOVEit Transfer instance produce the web-root file listing and its change history covering the exposure window and show no unaccounted .aspx file was written, instead of producing the upgrade record. Precondition, and this is where the control is usually over-claimed: it is a response requirement and removes nothing preventively — it does not close the injection, and it applies only to instances exposed before the fixed release landed. The packet places exploitation in May and June 2023, so an instance first stood up on a fixed release and never reachable before it is a patch case, not an incident case; conversely, an instance that took the upgrade without the hunt has not been cleared, because the fix removes the write path and not what was already written through it.",
+        "evidence": "Packet: an unauthenticated attacker sends crafted requests to the MOVEit Transfer web app to trigger SQL injection, then writes the LEMURLOOT (human2.aspx) web shell to enumerate and exfiltrate stored files and Azure storage keys; exploitation of unpatched systems can occur via HTTP or HTTPS; exploited in the wild in May and June 2023; depending on the database engine (MySQL, Microsoft SQL Server, or Azure SQL) an attacker may execute SQL statements that alter or delete database elements. cisa_kev true, kev_date 2023-06-02, active_exploitation confirmed, poc_available true, CVSS 9.8, RWEP 76. patch_available true; live_patch_available false with live_patch_notes: no vendor live-patch mechanism, remediation requires upgrading MOVEit Transfer to a fixed release (2021.0.6 / 2021.1.4 / 2022.0.4 / 2022.1.5 / 2023.0.1 or later) and hunting for the LEMURLOOT web shell, then restarting the service.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "AU-Essential-8-Patch"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-085",
+        "name": "DB-ABSTRACTION-LAYER-PARAMETERIZATION-VERIFICATION",
+        "description": "The defect the packet describes is SQL injection in the MOVEit Transfer web application reachable by an unauthenticated caller, with impact varying by the database engine behind it — the packet names MySQL, Microsoft SQL Server and Azure SQL — and extending past inference of structure and contents to executing SQL statements that alter or delete database elements. For a product the operator does not build, this control's requirement lands on what is accepted as evidence that the injection is closed: the parameterization must hold inside the shipped query layer, not in a filter placed in front of it. A WAF or request-inspection rule at the MOVEit boundary is therefore not a remediation state for this CVE and must not be recorded as one; the only state that closes it is the instance running one of the fixed releases the packet names — 2021.0.6, 2021.1.4, 2022.0.4, 2022.1.5 or 2023.0.1 or later — with the service restart the packet's remediation note requires, so an instance upgraded but not restarted is not yet remediated. Scope this to MOVEit Transfer, which is the product the packet names. The packet also states that all versions before those five are affected including older unsupported versions, naming 2020.0 and 2019x, so an instance on a branch with no fixed release in that list cannot reach a fixed build in place and has to be moved onto a supported branch — a migration decision that a patch ticket keyed to 'latest available for the installed branch' will never surface, and one that the cited flaw-remediation and system-security controls do not distinguish from routine patching. Precondition: reaching a fixed release closes this injection and says nothing about anything found in the product since, and it does not undo a database that was already read or altered through the injection during the exposure window — that remains an incident-response question, not a version question.",
+        "evidence": "Packet: a SQL injection vulnerability in the MOVEit Transfer web application could allow an unauthenticated attacker to gain access to MOVEit Transfer's database; depending on the database engine being used (MySQL, Microsoft SQL Server, or Azure SQL), an attacker may be able to infer information about the structure and contents of the database and execute SQL statements that alter or delete database elements; all versions before 2021.0.6, 2021.1.4, 2022.0.4, 2022.1.5 and 2023.0.1 are affected, including older unsupported versions (e.g. 2020.0 and 2019x); exploitation can occur via HTTP or HTTPS. patch_available true; live_patch_available false with live_patch_notes recording remediation as upgrading to a fixed release then restarting the service. CVSS 9.8, RWEP 76, poc_available true, active_exploitation confirmed, KEV-listed 2023-06-02.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2023-28771": {
     "name": "Zyxel Multiple Firewalls OS Command Injection Vulnerability",
@@ -59713,7 +60693,21 @@
         "adequate": false,
         "gap": "A.8.8 technical-vulnerability management ranking by CVSS treats 8.8 as High-not-Critical, understating a zero-day with a public exploit; deprioritising it left the type confusion open while it was actively used."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-057",
+        "name": "BROWSER-MANAGED-UPDATE-NO-DEFERRAL",
+        "description": "The packet records no live-patch mechanism for this flaw and states the remediation exactly: Chrome/Chromium auto-update to 112.0.5615.121 or later, followed by a browser restart to load the patched build. That places the whole residual exposure in the gap between the update arriving on disk and the user relaunching the browser, which is why this control has to bind both halves for this CVE. On a managed estate it means the security update channel is not deferred by the update ring at all for a KEV-listed renderer defect, and the relaunch is forced by policy rather than left to the user's convenience — the delivery path here is an ordinary page visit (the packet's path is a crafted JavaScript page passing a JSGlobalProxy to Error.captureStackTrace, confusing V8's object typing and corrupting the heap into read/write primitives usable for renderer code execution), and the browser session least likely to be restarted, an instance left open for weeks, is exactly the one most likely to meet such a page. Completion must be measured on the version V8 is actually executing, not the version deployed: a host whose management record shows 112.0.5615.121 while the running browser has never been relaunched is still executing the pre-fix renderer, and every framework control citing this CVE closes on the deployed version. Distinguishing test: collect the running browser's version from managed hosts and confirm it is at or above 112.0.5615.121 — an estate reporting package-level compliance while long-lived sessions keep the old build resident is exposed to a defect the packet records as KEV-listed since 2023-04-17, with a public exploit and confirmed in-the-wild use. Precondition: forcing the relaunch bounds the window, it does not undo a visit that already happened; a host that rendered untrusted web content while below the fixed build belongs on the incident path rather than being closed on the update record.",
+        "evidence": "Packet: type confusion in V8 in Google Chrome prior to 112.0.5615.121 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page; a crafted JavaScript page passes a JSGlobalProxy to Error.captureStackTrace, confusing V8's object typing and corrupting the heap into read/write primitives usable for renderer code execution. cisa_kev true, kev_date 2023-04-17, active_exploitation confirmed, poc_available true, CVSS 8.8, RWEP 66. patch_available true; live_patch_available false with live_patch_notes: no live-patch mechanism, remediation is Chrome/Chromium auto-update to 112.0.5615.121 or later followed by a browser restart to load the patched build.",
+        "gap_closes": [
+          "NIST-800-53-SI-2",
+          "ISO-27001-2022-A.8.8",
+          "NIS2-Art21-patch-management",
+          "AU-Essential-8-Patch"
+        ]
+      }
+    ]
   },
   "CVE-2023-20963": {
     "name": "Android Framework Privilege Escalation Vulnerability (CVE-2023-20963)",
@@ -61026,7 +62020,31 @@
         "adequate": false,
         "gap": "A.8.8 technical vulnerability management could not remediate a driver whose fix was withheld by the device supply chain; the kernel write primitive persisted on inventoried mobile assets despite being a known, KEV-listed exposure."
       }
-    }
+    },
+    "new_control_requirements": [
+      {
+        "id": "NEW-CTRL-126",
+        "name": "MOBILE-OS-MINIMUM-PATCH-LEVEL-ENFORCEMENT",
+        "description": "The packet's remediation is an OEM firmware/security update that ships the fixed Mali driver and reboots the device — the operator does not build that update and cannot accelerate it, so the only lever they actually hold is whether a device still carrying an affected driver keeps its access. For this CVE the enforced minimum has to be expressed as the OEM build that carries a Mali kernel driver outside the affected revision ranges (Midgard r26p0 through r31p0, Bifrost r0p0 through r35p0, Valhall r19p0 through r35p0), and it must function as an access condition — mail, VPN and document access denied to a device below it — not as a row on a patch-compliance dashboard. Distinguishing test: enrol a device whose Mali driver is still inside an affected range and confirm the policy actually denies it access to protected resources; an estate that surfaces the stale build on a report while the device keeps its access has recorded the exposure rather than removed it. Precondition, and this is where the control is usually over-claimed: the packet's path starts with an unprivileged local app, so restricting which applications may be installed raises the bar for getting that app onto the device but does not evict one already installed and does not cover one that arrived through the normal store channel — a device suspected of having run it belongs on the incident path, because the escalation ends at root and the packet records this as the local-escalation stage of a spyware chain. The update also reboots the device, so a device that has downloaded it but not restarted still runs the vulnerable driver and is not remediated.",
+        "evidence": "Packet vector: 'Arm Mali GPU Kernel Driver allows a non-privileged user to achieve write access to read-only memory pages. This affects Midgard r26p0 through r31p0, Bifrost r0p0 through r35p0, and Valhall r19p0 through r35p0.' Attack vector: 'An unprivileged local app abuses the Mali GPU kernel driver to obtain host-writable mappings of read-only imported pages, corrupting kernel memory to escalate privileges to root (used as the local-escalation stage of a spyware chain).' CISA KEV listed 2023-03-30; active_exploitation confirmed; poc_available true; CVSS 7.8; RWEP 72. patch_available true, live_patch_available false, live_patch_notes: 'No live-patch mechanism for the Mali GPU kernel driver on affected devices; remediation requires the OEM firmware/security update that ships the fixed driver, which reboots the device.'",
+        "gap_closes": [
+          "AU-Essential-8-Patch",
+          "NIST-800-53-SI-2",
+          "NIS2-Art21-patch-management"
+        ]
+      },
+      {
+        "id": "NEW-CTRL-018",
+        "name": "SCANNER-PAPER-COMPLIANCE-TEST",
+        "description": "The packet scopes this vulnerability by Mali kernel driver revision — Midgard r26p0 through r31p0, Bifrost r0p0 through r35p0, Valhall r19p0 through r35p0 — and not by an OS version or a reported patch-level string, while the fixed driver arrives inside an OEM firmware/security update rather than as a package the operator installs. So a fleet report that reads a device's OS build or patch-level date and marks it compliant has tested nothing about this CVE: the driver revision is chosen by the OEM's build, so two devices reporting the same patch level from different vendors can ship different Mali revisions, and one of them can still be inside an affected range. The operational test is to read the Mali driver revision actually loaded on a representative device of each OEM and model in the estate, check it against the three affected ranges, and re-run that check after every OEM update — because the update that moves the patch-level string is not necessarily the one that moves the driver, and that mismatch is exactly what a version-only scan cannot see. Precondition: this establishes exposure or its absence, it does not remediate. The packet records no live-patch mechanism, so a device found inside an affected range stays exposed until the OEM update lands and the device reboots, and a device the check clears today can regress if a later OEM build ships a driver revision back inside the range.",
+        "evidence": "Packet vector names the affected component by driver revision: 'This affects Midgard r26p0 through r31p0, Bifrost r0p0 through r35p0, and Valhall r19p0 through r35p0.' live_patch_available false with live_patch_notes: 'No live-patch mechanism for the Mali GPU kernel driver on affected devices; remediation requires the OEM firmware/security update that ships the fixed driver, which reboots the device.' CISA KEV listed 2023-03-30; active_exploitation confirmed; poc_available true; CVSS 7.8; RWEP 72; patch_available true. Citing gaps on this entry include ISO/IEC 27001:2022 A.8.8 (Management of technical vulnerabilities) and UK CAF B4 (System security).",
+        "gap_closes": [
+          "ISO-27001-2022-A.8.8",
+          "AU-Essential-8-Patch",
+          "UK-CAF-B4"
+        ]
+      }
+    ]
   },
   "CVE-2023-26360": {
     "name": "Adobe ColdFusion Deserialization of Untrusted Data Vulnerability (CVE-2023-26360)",