01Citrix confirms attacks on a new NetScaler SAML flaw that the September fixed builds do not close. Citrix's bulletin CTX697174 fixes CVE-2026-88779, a memory overflow in NetScaler ADC and Gateway that a network attacker needing no privileges reaches when the appliance is configured as a SAML service provider or identity provider; Citrix says it has observed targeted attacks and scopes the impact to denial of service. Appliances already on the 14.1-73.37 or 13.1-64.23 builds that fixed the September zero-days remain affected and must move to 14.1-73.41, 13.1-64.28 or the FIPS builds, and identity-provider appliances then need 14.1-73.46, 13.1-64.29 or the FIPS builds, which close the further SAML flaw CVE-2026-107406 that Citrix disclosed on 2026-10-08. Reports of code execution on patched honeypots are unconfirmed, and CISA added the flaw to its KEV catalog on 2026-10-04. Rapid7's incident analysis of 2026-10-19 shows the crash used as the reboot trigger of a root-execution chain with CVE-2026-88771 on an affected 14.1 appliance, so repeated crashes on an appliance not yet patched for that flaw call for a compromise check. →
02Huntress: a member account plus an .aspx upload gave an attacker webshells on three servers and a hunt for card data. Huntress observed from 2026-09-10 a threat actor compromising three web servers of a shared recreation-management platform by registering a member account and uploading .aspx webshells into the member-files directory, then searching for database credentials and cardholder data. After one server went back into production prematurely, the actor returned with the same account and planted a credential-harvesting script in the login page's jQuery file. Huntress names no vendor or CVE. →
03Huntress: a ChatGPT Custom GPT lure feeds a ClickFix chain that ends in a signed-host DLL sideload and a RAT. Huntress describes a September 2026 campaign in which, in some incidents, a sponsored Google result for "chatgpt" opens an attacker-built Custom GPT on chatgpt.com, which sends the visitor to a Google Sites page showing a fake CAPTCHA and a ClickFix paste-and-run command. The command installs an MSI that starts a signed Canon application sideloading a malicious DLL, which unpacks a RAT with remote desktop, camera and microphone capture and DNS-over-HTTPS command resolution; Huntress has responded to at least 40 incidents from the Google Sites domain, two confirmed through a Custom GPT. →
Huntress observed on 2026-09-10 a threat actor compromising multiple tenants of a shared recreation-management platform with one repeatable method: "register a member account, upload a malicious file, and turn it into a webshell" (Huntress, 2026-09-30). Huntress's summary calls the three compromised web servers municipal and says the actor aimed to steal payment data; the post names neither the platform vendor nor a CVE. On the first server the actor spent roughly six hours on failed unauthenticated attempts (login brute force, IIS 8.3 tilde enumeration, WebDAV verbs, upload-handler bypasses, forced browsing); what worked was registering an account and uploading 14 files into the member-files directory, the .aspx ones as webshells (Huntress, 2026-09-30).
Through the webshells, run from the IIS worker process, the actor enumerated the host and IIS sites, read configuration files for connection strings and keys, connected to the database with the harvested SQL credentials, searched files for card data and read a payment gateway's plain-text webhook logs to extract card numbers, expiry dates and card verification codes (Huntress, 2026-09-30). On a third server it uploaded five webshells, copied them under names resembling site assets, set their timestamps to match web.config and probed the payment-module folders (Huntress, 2026-09-30).
When that server was put back into production prematurely, the actor returned with the same registered account, re-uploaded webshells and appended an obfuscated loader to a jQuery file that the authentication page loads, intending to turn every browser that loads the page into an encrypted command client that evaluates pushed code and harvests credentials in real time (Huntress, 2026-09-30). Huntress deduces from a Simplified Chinese locale in a PowerShell user-agent that the actor is most likely based in China, and says script comments suggest AI-generated code. It does not state how many cards were exposed or whether any data left the servers.
Triage: legitimate member uploads of documents and images land in the same directory; the discriminator is a script-capable extension there, or a file there that is requested directly and then spawns child processes (Huntress, 2026-09-30).
register a member account, upload a malicious file, and turn it into a webshell.
When the third server was put back into production prematurely, the threat actor returned with a vengeance.
creating their own account on the platform and finding a flaw in the upload function
Huntress reports a campaign from late September 2026 in which, in some incidents, victims who searched Google for "chatgpt" got a sponsored result that opened an attacker-built Custom GPT on the genuine chatgpt.com domain. It answers any input with a fake service-availability notice pointing to a "backup domain" on Google Sites, which shows a CAPTCHA and tells the user to paste a command into a terminal. The command fetches a layered, obfuscated script from a host written as a single decimal number, so "rules and URL filters that look for a dotted IP address never see one"; the script installs a hidden MSI with msiexec and deletes itself (Huntress, 2026-09-28).
The MSI presents itself as a printer-configuration reader, launches a legitimately signed Canon application, which loads a patched Canon logging DLL whose imports pull in unsigned DLLs; one XOR-decodes shellcode hidden in a file posing as audio. The shellcode bypasses AMSI, unhooks ntdll, runs anti-VM checks and hosts the .NET runtime, then opens a custom encrypted archive holding a persistence script and the remote access trojan. The script re-creates an HKCU Run value every 150 seconds and a scheduled task of the same name every 875 seconds if either is removed. The RAT offers remote desktop, camera, microphone and audio capture, file search and follow-on payload execution, and resolves its command server through DNS-over-HTTPS to public resolvers, "so they never appear in local DNS logs" (Huntress, 2026-09-28).
Huntress has responded to at least 40 incidents from the Google Sites domain and confirmed two through a Custom GPT. OpenAI took the first GPT down as of 2026-09-25; on 2026-09-27 Huntress found a replacement that swaps the signed host for a Stardock application, hides the loader in a Microsoft NuGet package, strips Mark-of-the-Web from the MSI and pulls a second script that is freshly obfuscated on every request, while the RAT is byte-identical (Huntress, 2026-09-28).
Triage: the signed Canon and Stardock host binaries are legitimate (Huntress: do not block globally); the discriminator is the same binary launched by msiexec from a fake product folder under the user profile, beside a same-named Run value and scheduled task that return when deleted (Huntress, 2026-09-28).
has responded to at least 40 incidents stemming from the specific Google Sites domain involved in this attack, and confirmed that two of these incidents came through a Custom GPT instance
Most ClickFix chains we see are two or three hops: paste a command, download something, run it. This one has eight, and each hop exists to hide the next one.
rules and URL filters that look for a dotted IP address never see one
Detections tied to Canon or Stardock names will miss the next swap.
Citrix's bulletin CTX697174 describes CVE-2026-88779, a memory overflow in customer-managed NetScaler ADC and NetScaler Gateway reachable only when the appliance is configured as a SAML service provider or identity provider; the CVSS 4.0 vector is network, no privileges, no interaction, availability impact only (Citrix, 2026-10-03). Citrix says it "has observed targeted attacks on unmitigated NetScaler deployments which can lead to Denial of Service" and has not identified an integrity impact on customer data (Citrix, 2026-10-03). The flaw is independent of the vulnerabilities disclosed in the earlier bulletin CTX697096 (Citrix, 2026-10-02), so an appliance already on 14.1-73.37 or 13.1-64.23 is still affected, and Citrix tells operators who meet the SAML precondition to "upgrade your deployment again" (Citrix, 2026-10-03). The fixed builds are 14.1-73.41, 13.1-64.28, 14.1-73.41 FIPS and 13.1-37.282 (Citrix, 2026-10-03).
Until the upgrade, Citrix offers Global Deny List signatures through NetScaler Console with virtual patching enabled, for builds from 14.1-73.37 up to 14.1-73.41 and from 13.1-64.23 up to 13.1-64.28; they "can help to mitigate" the flaw, but the upgrade is the fix (Citrix, 2026-10-03).
Other reporting reads more into the flaw: administrators reported repeated crashes and reboots of appliances running patched releases, including 14.1-73.37, with the nsaaad authentication service failing after crafted SAML-related requests, one seeing a script-download command in the logs that did not succeed and another a payload in the username field (Cyber Press, 2026-10-03). heise relays researcher Kevin Beaumont's observation that one of two patched honeypots was running a downloaded binary, and a statement that watchTowr Labs reproduced the flaw (heise online, 2026-10-03); ASD's ACSC says an attacker "may induce system crashes, denial of service and potential exploitation" and that it is aware of impacts to Australian organisations (ASD's ACSC, 2026-10-03). None of these sources gives evidence of code execution beyond that observation and reproduction claim, and CISA added the flaw to its KEV catalog on 2026-10-04 (CISA KEV catalog, 2026-10-04).
Triage: reboots alone are not proof of compromise; crashes together with malformed authentication traffic and scanning warrant incident-response handling until forensics rules out intrusion (Cyber Press, 2026-10-03).
Citrix has observed targeted attacks on unmitigated NetScaler deployments which can lead to Denial of Service.
NetScaler ADC or NetScaler Gateway must be configured as a SAML SP OR SAML IdP
please upgrade your deployment again with the software released as part of the
This issue is independent of the vulnerabilities disclosed in
A two-vulnerability sequence in which the SAML parsing flaw supplies a crash/reboot condition while the log-poisoning flaw supplies attacker-controlled command text.
After reboot, the NetScaler call-home crash handler processed the retained attacker-controlled text and executed a payload as root.
Citrix provides an indicator-of-compromise script that teams can run to check exposed appliances for signs of compromise, though a clean result is not definitive proof
CISA added CVE-2026-88779 to its KEV catalog on 2026-10-04, after Citrix's bulletin (CISA, 2026-10-04); the catalog describes a memory-buffer flaw that "could allow for a denial of service" (CISA KEV catalog, 2026-10-04). That is the scope Citrix states, so the listing confirms exploitation without adding evidence of code execution, and the upgrade guidance above is unchanged.
Citrix published bulletin CTX697191 on 2026-10-08 for CVE-2026-107406, a memory overflow in customer-managed NetScaler ADC and Gateway that leads to remote code execution or denial of service (CVSS 4.0 9.5) and applies to the same SAML configurations (Citrix, 2026-10-08). For an appliance configured as a SAML identity provider the affected builds run through 14.1-73.41, 13.1-64.28 and the FIPS and NDcPP equivalents (Citrix, 2026-10-08), and those are the builds named above as the fix for CVE-2026-88779 (Citrix, 2026-10-03); the fixed builds for the new flaw are 14.1-73.46, 13.1-64.29, 14.1-73.46 FIPS and 13.1.37.283 and later (Citrix, 2026-10-08). An appliance that is only a SAML service provider is listed as affected only before 14.1-73.37 and 13.1-64.23, so the 14.1-73.41 and 13.1-64.28 builds remain enough for it (Citrix, 2026-10-08). Citrix's blog of the same day says that as of the bulletin's publication it is not aware of any unmitigated exploits of the new flaw (Citrix, 2026-10-08).
Rapid7's incident analysis of an early-October compromise of a NetScaler ADC on an affected 14.1 build shows CVE-2026-88779 used as the crash and reboot condition in a two-flaw sequence with CVE-2026-88771, the pre-authentication log-poisoning flaw that supplies attacker-controlled command text: repeated nsaaad crashes and a Pitboss reboot followed text inserted into an authentication username, and after the reboot the call-home crash handler processed the retained text and executed a payload as root (Rapid7, 2026-10-19). The root execution therefore came through the log-poisoning flaw and the crash handler, with this flaw supplying the reboot, and Rapid7 reports persistence, internal reconnaissance and exfiltration of the appliance configuration after it; the full account is in the CVE-2026-88771 analysis (Rapid7, 2026-10-19).
watchTowr's Jake Knott told The Register that "Citrix provides an indicator-of-compromise script that teams can run to check exposed appliances for signs of compromise, though a clean result is not definitive proof" (The Register, 2026-10-05). Citrix's own guidance for this flaw names no such script and tells customers to follow their standard incident response processes if they identify signs of compromise (Citrix, 2026-10-03).
The headline says Citrix has published no workaround. The Triage line follows watchTowr's root-cause analysis for CVE-2026-88771 and GTIG's log artifacts for CVE-2026-88772, which GTIG advises alerting on individually, and the zero-day timing is cited to watchTowr. Only Mandiant's DTLS and UDP/443 controls are limited to CVE-2026-88772 and its upstream IP allow-listing is not tied to one flaw, the KEV sentence drops the US directive, a reference to an exploitation table that does not exist is removed, and the takeaway adds GTIG's sector statement.
For CVE-2026-88771, the precursor in the appliance's own logs is header or username values carrying shell metacharacters together with crash-message text such as "PPE unexpectedly died" (watchTowr Labs, 2026-09-28). For CVE-2026-88772, GTIG names a DTLS handshake-failure line, packet-engine terminations and watchdog "NOT restarting" lines, and advises alerting on each on its own rather than requiring both (GTIG/Mandiant, 2026-09-29). Citrix has published no workaround (watchTowr, 2026-09-27), and Mandiant's DTLS and UDP/443 controls apply to CVE-2026-88772 only, while its upstream IP allow-listing is not scoped to one CVE (GTIG/Mandiant, 2026-09-29).
Bishop Fox published on 2026-10-01 an end-to-end unauthenticated root exploit chain over TCP 19009 against unpatched R81.10 and R82.10 servers and a patch-state scanner. It says Check Point's second indicator does not fire on this route, that the first indicator is the crash signature of a separate CVE, and that the hotfix closes three defects; its hunting advice is file-integrity monitoring.
Bishop Fox published on 2026-10-01 a lab reproduction of an unauthenticated root code-execution chain on unpatched R81.10 and R82.10 management servers, run end to end over TCP 19009 (Bishop Fox, 2026-10-01). The chain mints a session by presenting the server's own internal certificate name, which the server accepts from a caller-supplied string without checking that a certificate was presented, then uses the file service to write a file anywhere as root; overwriting a stock scheduled-task file makes the system's cron daemon run the attacker's command as root with no further request, and the output is read back over the same session (Bishop Fox, 2026-10-01). Bishop Fox confirmed two further routes from the same write to code execution and says closing any one leaves the others standing (Bishop Fox, 2026-10-01). The post describes the chain but links only a patch-state scanner, not the exploit script.
Bishop Fox reads the hotfix as closing three defects while the advisory names two: the forged login, the unbounded file write and the upgrade-service parameter validation, with either of the first two alone breaking the chain; this is its comparison of the shipped code, not a Check Point statement (Bishop Fox, 2026-10-01). Its scanner reads patch state from outside with one benign request to the management port and, per its README, does not detect prior compromise (Bishop Fox, 2026-10-01).
Bishop Fox says Check Point's second indicator at the time (an upgrade-tools load-error grep, as sk1000171 then listed it) fires only when an upgrade-tools load fails and is never reached by this route, so it leaves neither the error line nor the parent-directory sequence, and that a genuine upgrade can trip it; it calls the first indicator, an over-long username next to a core dump, the crash signature of the separate CVE-2026-91843 and not evidence of this traversal (Bishop Fox, 2026-10-01). An exposed, unpatched server with neither log sign is therefore not shown to be clean. Its hunting advice is file-integrity monitoring of scheduled-task directories, startup and SSH key paths, the normally empty upgrade-tools directory, the CPM configuration files and anything new in the output-staging directory, paired with execution telemetry, cron reload events and an alert on the management Java process spawning a shell, because a scheduled hash check can miss stock files restored byte for byte (Bishop Fox, 2026-10-01). It also notes that the Trusted Clients restriction is a permissions setting that emits an implied rule, so a rule-base audit alone can miss it (Bishop Fox, 2026-10-01).
Flink's own customer notification of 2026-09-25, seen by RETAIL-NEWS, says an unauthorised person reached an internal system with compromised credentials, which corrects the earlier statement that no initial-access vector had been disclosed. The 10,000 figure now says customers, the two outlets' differing readings of the 100 ETH demand are stated, and Flink's more cautious notice on the order details is carried.
Flink's own customer notification of 2026-09-25, which RETAIL-NEWS saw, says an unauthorized person reached one of the delivery service's internal systems with compromised credentials (RETAIL-NEWS, 2026-09-25). Which credentials were used, how they were obtained and whether multi-factor authentication applied are not stated. heise reports the 10,000 ransom notes for customers (heise online, 2026-09-26), and Flink says employees are also being contacted directly (NL Times, 2026-09-25). The outlets read the 100 ETH differently: heise as the sum customers are to raise (heise online, 2026-09-26), NL Times as what the group says the company itself would have to pay for all user data to be deleted (NL Times, 2026-09-25). Flink's customer notice was more cautious about data scope: it said there was no concrete indication that the additional order details had been accessed (RETAIL-NEWS, 2026-09-25).
CISA added both CVEs to its KEV catalog on 2026-10-02, and Zammad's own statement of 2026-10-01 says the code-execution flaw is exploitable only on 6.5 and older (which receive no security fixes), that 7.0 and later are not affected and 7.2.0 hardens the code, and that it cannot confirm the root-escalation flaw because DIVD gave it no technical details. Zammad also says it first received a report about the code-execution flaw in August 2026, before the exploitation DIVD dates to 2026-09-21.
CISA added both CVE-2026-102489 and CVE-2026-102490 to its Known Exploited Vulnerabilities catalog on 2026-10-02 and describes them as chainable (CISA KEV catalog, 2026-10-02). Zammad published its own statement on 2026-10-01: exploitation of CVE-2026-102489 is "only possible on Zammad 6.5 and older, because of the runtime environment those versions use", those versions no longer receive security fixes, Zammad 7.0 and later are not affected, and the code is hardened in 7.2.0, the current stable release (Zammad, 2026-10-01). For CVE-2026-102490 Zammad says "DIVD has not given us any technical details about this vulnerability", so it cannot confirm the flaw, its scope or the affected versions, names no fix, and disputes DIVD's handling of the disclosure, citing a report to Zammad on 2026-09-24 and public scanning and disclosure on 2026-09-26 (Zammad, 2026-10-01). Zammad also says it first received a report about CVE-2026-102489 in August 2026 and analysed it then, which is before the exploitation DIVD dates to 2026-09-21, so the vendor may have known of the flaw before it was exploited (Zammad, 2026-10-01).
Citrix's bulletin CTX697174 (2026-10-03) fixes CVE-2026-88779, a SAML-triggered memory overflow that the September fixed builds do not cover; Citrix confirms targeted attacks and tells SAML-configured appliances that already took the September builds to upgrade again, to 14.1-73.41, 13.1-64.28 or the FIPS builds.
Citrix published a second NetScaler bulletin on 2026-10-03, CTX697174, for CVE-2026-88779, a memory overflow reachable when the appliance is configured as a SAML service provider or identity provider; Citrix says it has observed targeted attacks and scopes the impact to denial of service (Citrix, 2026-10-03; Citrix, 2026-10-03). The flaw is independent of the vulnerabilities disclosed in the earlier bulletin CTX697096 (Citrix, 2026-10-02), and Citrix tells operators who upgraded to the September builds and meet the SAML precondition to upgrade again (Citrix, 2026-10-03). The fixed builds are 14.1-73.41, 13.1-64.28, 14.1-73.41 FIPS and 13.1-37.282 (Citrix, 2026-10-03). A SAML-configured appliance on 14.1-73.37 or 13.1-64.23 is therefore fixed against the exploited command-injection and DTLS flaws described above but not against this one.
Upgrade every internet-facing NetScaler ADC/Gateway to 14.1-73.37+ (14.1-FIPS 14.1-73.37 FIPS+) or 13.1-64.23+ (13.1-FIPS/NDcPP 13.1-37.279+) now (a SAML-configured appliance needs the later builds named for CVE-2026-88779, and an identity provider 14.1-73.46 or 13.1-64.29); on a 13.1-branch appliance run show ns variable first; if it returns any variables, install 13.1-64.24 instead to avoid a known reboot loop. Capture logs, an instance snapshot where the appliance is virtual, a support bundle and a core dump from each exposed appliance BEFORE patching, since the upgrade can remove forensic evidence of prior exploitation.
Run Mandiant's compromise checks, extended by the Unit 42 artifacts described in the update below, Zscaler's (appliance-initiated downloads to unfamiliar addresses, an unexpected executable under /var, a Perl CGI handler added to the web server configuration) and Rapid7's (the setuid bit on /bin/sh, nsaaad crashes followed by Pitboss reboots, an outbound agent polling an external endpoint every 15 seconds), on every NetScaler that was internet-facing and unpatched at any time since early September, patched or not; isolate any hit, check both nodes of an HA pair, halt HA configuration sync until both nodes are validated, and rotate the LDAP and RADIUS bind credentials, SSL private keys and key-encryption material held in the configuration, since Rapid7 saw the configuration and backups staged and exfiltrated.
Find any NetScaler ADC or Gateway still on the end-of-life 12.1 or 13.0 branches and decide isolation or replacement now: they receive no security updates and Citrix has not said whether they are affected.
Copy the application and network logs of every self-hosted Zammad instance, then upgrade it to Zammad 7.2.2 (7.2.0 is affected by the 27 advisories Zammad published on 2026-10-06); a 6.5 instance that cannot be upgraded yet gets the 6.5.5 emergency fix from Zammad's new 6.5 package repository, and an older instance that cannot be upgraded goes offline.
On every instance that was internet-reachable on 6.3.0 to 6.5.4 since 2026-09-21, run DIVD's log-check script, and treat a hit, or any DEB or RPM host where code ran as the zammad account, as compromised.
Apply the R82.20 Security Hotfix or the appropriate Jumbo Hotfix Accumulator to every Check Point Security Management / Multi-Domain Security Management / Log / Multi-Domain Log / SmartEvent server now; no LivePatch exists for this issue.
On every affected server that was reachable on TCP/19009 while unpatched, check /etc/cron.d, authorized_keys, the startup paths and anything new under $FWDIR/conf/SMC_Files for unexpected changes, alongside Check Point's revised log indicators in sk1000171.
Upgrade every SAML-configured NetScaler (add authentication samlAction or add authentication samlIdPProfile) that is still below 14.1-73.41, 13.1-64.28, 14.1-73.41 FIPS or 13.1-37.282, including appliances on the 14.1-73.37 or 13.1-64.23 builds, to at least those builds; identity providers then need the builds named for CVE-2026-107406. Where the upgrade has to wait and NetScaler Console virtual patching is available, confirm with show appfw signatures that the Global Deny List signature version is at least v24, which Citrix offers as a bridge for CVE-2026-88779 only.
Before restarting a SAML-configured NetScaler that has crashed or rebooted repeatedly, preserve its logs and crash artifacts, open a Citrix support case, which Citrix asks for when an appliance is experiencing the impact, and search its authentication logs for shell metacharacters next to Pitboss crash-message text, because Rapid7 saw a crash-and-reboot cycle used to make the call-home crash handler run an attacker's payload as root.
2026-10-04T0405Z-intel· Sonnet 5.5 · window 26 h · 3 entries published
Verification & coverage notes
Coverage window: 26 h (gap 24.0 h to 2026-10-03T0404Z-intel), a standard window. Sunday; the vendor calendar was quiet and the only fresh high-value item was the NetScaler SAML zero-day that S1 and S2 reached independently (via the ENISA EUVD sweep and a heise lead). Published 3 new entries and appended changelog records to 4 existing ones. Mechanical KEV sweep (kev-window.txt): 0 additions dated in the window (catalogue 2026.10.02); S1's store-wide KEV cross-check found two gaps, handled below. No intel/ drops (README only), so no S5.
Verification: five iterations. Iterations 1 to 4 each returned NEEDS_FIXES with 18, 11, 2 and 6 truth plus editorial findings, all remediated and re-gated; the 2 of iteration 3 included an F4 on a legacy status field, so a further pass followed. Iteration 5 (all 184 claims answered, all ok) returned NEEDS_FIXES with one low-confidence truth finding (an unsourced 'only way' in the 88771 Detection paragraph) and four advisories; the run ended on decision rule 5 (truth plus editorial of at most 2, no F1 or F4): the five remediations were applied and the gate re-run (57 pass, 0 warn, 0 fail), with no further verifier pass, so the final text of those three edits (88771 Detection paragraph, 88779 Detection line, Flink correction record) was checked by the gate and not by a verifier. Residual: 1. The next audit should re-read them. One advisory (the recreation entry's event_date) was declined with its reason.
Updated (changelog): 4 entries. The fourth, 2026-09-23/cve-2026-93616-check-point-security-mgmt-path-traversal (type update), came from verifier iteration 1's missed-angle finding: Bishop Fox's 2026-10-01 end-to-end exploit method, patch-state scanner and its reading that Check Point's indicators miss the route; S1 had logged no delta after reading only sk1000171, and a scoped follow-up (FU2) found the delta. The other three: 2026-09-28/cve-2026-88771-citrix-netscaler-preauth-rce-zero-day-kev (type update: the fixed-build floor for SAML-configured appliances moves to 14.1-73.41 / 13.1-64.28 because CVE-2026-88779 is not fixed by the September builds); 2026-10-02/zammad-cve-2026-102489-102490-exploited-divd-breach (type update: Zammad's own statement names 7.2.0 as the target, says 6.5 and older receive no fixes and that it cannot confirm the root flaw; both CVEs now carry the KEV listing of 2026-10-02); 2026-09-27/flink-lpg-group-crowdfund-extortion-order-hub-breach (type improvement: Flink's customer notice names compromised credentials, so the earlier "vector not disclosed" statement was corrected and T1078 added).
Contradiction: Zammad. Zammad's own statement says it first received a report about CVE-2026-102489 in August 2026, which contests DIVD's framing of two previously unknown flaws exploited from 2026-09-21 (DIVD's timeline shows its report to Zammad on 2026-09-24); the entry carries both and attributes each (sourcing_note, update section).
Contradiction: Flink data scope. Flink's own customer notice of 2026-09-25 (RETAIL-NEWS) says the additional order details (floor, doorbell name, delivery notes) may have been affected but gives no concrete indication of actual access, while heise quotes Flink saying delivery notes leaked in individual cases; the entry carries both and attributes each.
Contradiction: CVE-2026-88779. Citrix scopes the impact to denial of service with no integrity impact; heise relays a researcher's report of a downloaded binary on a patched honeypot and a watchTowr reproduction, and ACSC says "potential exploitation". The entry carries both and states that code execution is unconfirmed (sourcing_note), priority high, not critical.
Single-source: 2026-10-04/chatgpt-custom-gpt-clickfix-sideloaded-in-memory-rat and 2026-10-04/recreation-platform-upload-webshell-card-data-foothold (Huntress first-hand SOC telemetry and sample analysis; no independent confirmation found; the second names no vendor, CVE or victim, confidence medium). Lookback inclusion (PD-7 d): both Huntress posts (2026-09-28, 2026-09-30) were first coverage from the huntress record's rotational lookback.
Reduced reading: the community.citrix.com blog pages (403 to extract and WebFetch) and the Cyber Press page were read through the reader proxy; the Citrix bulletin, heise, ACSC and every other cited page were read directly.
borderline-drop: Exchange Server September SU V2 / CVE-2026-96940, authenticated cross-mailbox read, no exploitation or PoC, regular patch cycle; published about 7 h before the window.
borderline-drop: Dolibarr CVE-2026-89013 (CSIRT Italia reports exploitation), niche open-source ERP/CRM with no constituency deployment evidenced and no scale stated; same footing as the held IBM Guardium row.
borderline-drop: Microsoft Digital Defense Report 2026 (annual report, found late via the msft-secblog pass, published 2026-10-01), vendor statistics and a policy piece with no defender decision; the Swiss ransomware table's caption and unit are unreadable in the PDF text (29 to 59, +103% could not be assigned a meaning), leaving only a 2% Russia-nexus notification share for Switzerland. Flagged to the audit as an annual-report calendar miss, not a recovery request.
borderline-drop: SConnect (Thales) CVE-2026-18397, fixed since August, end of life, no exploitation, no Swiss deployment evidenced.
borderline-drop: DTU (Denmark) breach, out of nexus, no actor, no technique beyond compromised profiles (PD-11 breach gate fails).
borderline-drop: Oxygen Forensics DOJ complaint, provenance and ownership allegation about police forensic tooling; no adversary behavior (no evidence-bound techniques[] for an incident entry), Swiss use unverified by any source, first reported 2026-09-24. Re-open if a Swiss licensee or authority guidance surfaces.
borderline-drop: OpenAI notices to over 100 organizations and the California AG subpoena, awareness; documents dated 2026-09-30 to 10-02 and the three covered OpenAI entries already hold the substance.
borderline-drop: ShinyHunters "Rey" detention report, unconfirmed by any authority (Reuters via two relays), no defender consequence; revisit on an authority confirmation.
borderline-drop: DragonForce second-stage backdoor (Lab52, 2026-10-01), single-source loader and MQTT-fallback analysis on a legacy-shape entry, no victim, sector or initial-access context.
borderline-drop: Kiteworks "over 100 advisories" scale note (SecurityWeek roundup); MFT Server could not be confirmed from the vendor's advisory listing read this fire; the covered entry already carries the 9.5.1 floor.
borderline-drop: WordPress CVE-2026-87902 KEV listing of 2026-09-25, the covered entry already says exploited (PD-13 bookkeeping, nothing shipped).
borderline-drop (S1, S4 exclusions, all with reasons in the findings files): Fortra BoKS, GitLab AI Gateway CVE-2026-90970, Dell CSM, Chrome 154.0.8037.97, Mozilla MFSA 2026-97 to 103, Zimbra 10.1.21 (watch for a PoC), WatchGuard AP, OpenAM CVE-2026-105115 (watch for a PoC), Digi / cPanel / UTMStack EUVD items, Foreman, PAN GlobalProtect App, 0patch WalletService, FUJIFILM/Sharp MFP; Bundesagentur fuer Arbeit outage (agency rules out an attack), IFAPME re-post, STMicroelectronics / Neff Drexel and the Gentlemen 2026-10-03 batch (leak-site claims only), Epic Systems pause, Frontline Education, Vicksburg, LDLC, Zenfirst, ICO reprimands, Garante IQVIA fine, Polish healthcare claims.
Backlog (state/coverage_backlog.md): all six open rows re-gated on today's facts by S1, S2 and S4; none clears the gate, none was struck, none published. IBM MQ / Langflow: dated note appended (IBM's bulletin covers IBM MQ Appliance only); held to 2026-10-11. MikroTik CVE-2026-84411, IBM Guardium CVE-2026-85542, ARA-Region Lyss-Limpachtal, Netech: no change, held to 2026-10-14. Qilin / Touring Club Suisse: no change; strikes on 2026-10-05 as scheduled. Securitas / Everest: no change; held to 2026-10-10. No row added.
Rotation: the digest's sources.rotation cursor ranked each domain; 49 records attempted by the previous two fires were excluded as the belt-and-braces rule says, except that S4's pool (12 breach sources, all attempted by those fires) was ranked on the cursor alone. ssd-disclosure (3 consecutive failures) was promoted to the S1 slice as a fetch-gap record and failed again.
Leads for the audit, not recovered here: ENISA Threat Landscape 2026 (published 2026-09-22, no store entry; public administration the most targeted sector), Zscaler ThreatLabz 2026 Ransomware Report (2026-09-30, listing only), Sophos "TerminalFix and Lorem Ipsum Loader" (2026-09-30, possible changelog record on the 2026-08-31 TerminalFix entry).
Candidate sources added: citrix-netscaler-security-bulletins (first-party NetScaler PSIRT, feed verified; S1 recommends essential tier on promotion), vaud-soc-cyber-threat-review (cantonal government SOC monthly review, no feed route yet). Recipe fixes: acronis-tru, dcod-ch and intrinsec feeds pinned (verified by S3 and S2), plus recipe notes on ten records. Surfaced but not added: IAPP news, Cybernews (feed 403; extract falls to the metered reader), Bay Area Labs.
Reader-pool use: a handful of calls to read community.citrix.com pages (S1, S2) and one Cyber Press page (this fire); no other metered fetch.
Coverage gaps: ssd-disclosure (captcha on every transport), inside-it-ch (article bodies behind the Vercel checkpoint; teasers only), msrc-blog (landing shell, no dated posts; msrc-update-guide carried the CVE data), fbi-cyber-alerts (extract returns the terms modal; WebFetch of /PSA works, no in-window PSA), depthfirst, pwn-ai (dormant), bsi-de (weekend quiet), cert-at (quiet); Reuters, justice.gov and Telegraph pages unreachable on every transport (read through relays for dropped items only).