<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0"><channel><title>SecureRoot: critical and exploited vulnerabilities</title><link>https://www.secureroot.com/vulnerabilities/</link><description>Vulnerabilities in Linux and Unix server software that need patching now.</description><lastBuildDate>Tue, 03 Nov 2026 17:06:56 +0000</lastBuildDate><item><title>CVE-2026-97229: GitLab CI/CD regular expression integer overflow allows unauthenticated RCE</title><link>https://www.secureroot.com/cve/CVE-2026-97229</link><guid>https://www.secureroot.com/cve/CVE-2026-97229</guid><pubDate>Thu, 29 Oct 2026 00:00:00 +0000</pubDate><description>An integer overflow in the Onigmo regular-expression compiler in the Ruby that GitLab CE/EE ships lets an unauthenticated attacker run code on the GitLab server by submitting a specially crafted regular expression in a CI/CD configuration; it bypasses the incomplete fix for CVE-2026-93577 in GitLab 19.4.1. GitLab fixed it in 19.4.2, 19.3.5 and 19.2.8 on October 12, 2026, and the underlying Ruby bug is fixed in Ruby 4.0.8, 3.4.12 and 3.3.13, so other Ruby applications that compile attacker-contro</description></item><item><title>CVE-2026-46300: Linux kernel ESP-in-TCP page-cache write to root (Fragnesia)</title><link>https://www.secureroot.com/cve/CVE-2026-46300</link><guid>https://www.secureroot.com/cve/CVE-2026-46300</guid><pubDate>Wed, 28 Oct 2026 00:00:00 +0000</pubDate><description>skb_try_coalesce() in the kernel networking core (net/core/skbuff.c) drops the shared-fragment marker when it merges socket buffers, so the IPsec ESP receive path (ESP-in-TCP) can decrypt in place over page-cache pages of files the attacker can only read. A local unprivileged user can overwrite the cached copy of a SUID binary such as /usr/bin/su and get a root shell; Ubuntu notes it may also allow container escape. A public exploit exists, and CISA added it to KEV on 28 October 2026, noting it </description></item><item><title>CVE-2026-111497: Brotli decoder out-of-bounds write in Node.js enables RCE (Node2Shell)</title><link>https://www.secureroot.com/cve/CVE-2026-111497</link><guid>https://www.secureroot.com/cve/CVE-2026-111497</guid><pubDate>Tue, 27 Oct 2026 00:00:00 +0000</pubDate><description>An unchecked return value in the Google Brotli decoder (versions 1.0.3 to 1.2.0) turns a malformed block switch into an out-of-bounds write, so attacker-supplied Brotli data reaching Node.js through request bodies with Content-Encoding: br (Express 5 / body-parser 2), Brotli responses decoded by fetch(), or node:zlib Brotli calls can crash the process, leak memory or run code. Public exploits give a shell on official Node.js 24.22.0 builds and CISA added it to the KEV catalog on 2026-10-27 (due </description></item><item><title>CVE-2026-48356: Adobe Commerce / Magento unrestricted file upload (APSB26-73)</title><link>https://www.secureroot.com/cve/CVE-2026-48356</link><guid>https://www.secureroot.com/cve/CVE-2026-48356</guid><pubDate>Tue, 27 Oct 2026 00:00:00 +0000</pubDate><description>Adobe Commerce and Magento Open Source accept upload of files with a dangerous type, which Adobe rates critical (CVSS 9.6) and says can lead to arbitrary code execution; no authentication is required but a victim must open a crafted URL or page. It was fixed in the July 14, 2026 security update APSB26-73. CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-27, so it is being exploited in the wild.</description></item><item><title>CVE-2026-109618: PHP out-of-memory handling corrupts request memory, remote code execution</title><link>https://www.secureroot.com/cve/CVE-2026-109618</link><guid>https://www.secureroot.com/cve/CVE-2026-109618</guid><pubDate>Mon, 26 Oct 2026 00:00:00 +0000</pubDate><description>When PHP hits an out-of-memory error it aborts the request with longjmp(), which can leave request memory partly initialized; request cleanup then treats it as valid, which attackers can turn into remote code execution. An unauthenticated attacker can trigger the out-of-memory condition with crafted multipart POST requests, and with Apache mod_php this happens before any application code runs; PHP-FPM is also considered susceptible. Fixed in PHP 8.5.12, 8.4.27, 8.3.36 and 8.2.35; CISA added it t</description></item><item><title>CVE-2026-109619: PHP stale request state reused by the next request, remote code execution</title><link>https://www.secureroot.com/cve/CVE-2026-109619</link><guid>https://www.secureroot.com/cve/CVE-2026-109619</guid><pubDate>Mon, 26 Oct 2026 00:00:00 +0000</pubDate><description>PHP worker processes keep some global state (error, session, object and extension state) that points into a request&#x27;s memory; after a fatal error or a late warning it is not reset, and the next request on the same worker can use the freed memory (use after free). An unauthenticated attacker can chain two requests to the same worker to get remote code execution: PHP confirmed this against Apache mod_php and, through session upload progress plus an application fatal error, against PHP-FPM. Fixed i</description></item><item><title>CVE-2026-105482: Node.js dns.resolve*() silently truncates hostnames at a NUL byte</title><link>https://www.secureroot.com/cve/CVE-2026-105482</link><guid>https://www.secureroot.com/cve/CVE-2026-105482</guid><pubDate>Fri, 23 Oct 2026 00:00:00 +0000</pubDate><description>dns.resolve*() and dns.reverse() pass names to c-ares as C strings, so attacker.example\0.trusted.example is resolved as attacker.example without an error. Applications that validate a hostname (for example against an allowlist) and then resolve it can act on records for a different name. This is an incomplete fix of CVE-2026-48930; fixed in the October 21, 2026 releases (High, CVSS 7.4; Red Hat Important), with no known exploitation.</description></item><item><title>CVE-2026-105483: Node.js TLS hostname check bypass with fullwidth-digit IP addresses</title><link>https://www.secureroot.com/cve/CVE-2026-105483</link><guid>https://www.secureroot.com/cve/CVE-2026-105483</guid><pubDate>Fri, 23 Oct 2026 00:00:00 +0000</pubDate><description>tls.checkServerIdentity() does not treat hosts written with non-ASCII digits (such as fullwidth １２７.0.0.1) as IP addresses, although they resolve to a normal IPv4 address, and matches them against certificate DNS names instead. A certificate containing DNS:127.0.0.1 or DNS:*.0.0.1 can then authenticate an IP endpoint it was not issued for. This regression from the CVE-2026-48618 fix affects only 24.x and 26.x; fixed October 21, 2026 (High, CVSS 7.4), no known exploitation.</description></item><item><title>CVE-2026-106069: Node.js TLS hostname check bypass with leading-zero IPv4 hosts</title><link>https://www.secureroot.com/cve/CVE-2026-106069</link><guid>https://www.secureroot.com/cve/CVE-2026-106069</guid><pubDate>Fri, 23 Oct 2026 00:00:00 +0000</pubDate><description>tls.checkServerIdentity() normalizes hosts with WHATWG URL rules, reading a leading-zero octet as octal (0127.0.0.1 becomes 87.0.0.1), while the operating system resolver on some platforms, such as macOS, connects to 127.0.0.1. A server at the connected address can therefore be authenticated with a certificate issued for a different IP address. Fixed in the October 21, 2026 releases (High, CVSS 7.4; Red Hat Important); no known exploitation.</description></item><item><title>CVE-2026-110447: Nextcloud Collabora Built-in CODE Server (richdocumentscode) command injection</title><link>https://www.secureroot.com/cve/CVE-2026-110447</link><guid>https://www.secureroot.com/cve/CVE-2026-110447</guid><pubDate>Fri, 23 Oct 2026 00:00:00 +0000</pubDate><description>The Nextcloud app &#x27;Collabora Online - Built-in CODE Server&#x27; (richdocumentscode) before 26.4.303 has an unauthenticated OS command injection in proxy.php that runs commands as the web server user while the built-in CODE server is not running (after a restart, upgrade or crash). It was exploited in the wild in September and October 2026 to drop PHP web shells, and CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-23. Standalone Collabora Online servers (Docker or distro packa</description></item><item><title>CVE-2026-110264: GitLab Org-mode regex integer overflow allows unauthenticated RCE</title><link>https://www.secureroot.com/cve/CVE-2026-110264</link><guid>https://www.secureroot.com/cve/CVE-2026-110264</guid><pubDate>Thu, 22 Oct 2026 00:00:00 +0000</pubDate><description>GitLab CE/EE renders Org-mode files with the org-ruby gem, which builds a regular expression from an attacker-supplied #+TODO: keyword; a keyword with more than 32,767 repeats overflows a 16-bit counter in Ruby&#x27;s Onigmo regex engine and causes an out-of-bounds write at match time. An unauthenticated attacker can turn this into code execution on the GitLab server as the git user, exposing gitlab-secrets.json, CI/CD variables, runner and access tokens, and every repository. GitLab rates it critica</description></item><item><title>CVE-2026-106931: SiteGuard Server Edition statistics feature OS command injection (root)</title><link>https://www.secureroot.com/cve/CVE-2026-106931</link><guid>https://www.secureroot.com/cve/CVE-2026-106931</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>SiteGuard Server Edition, a host-based WAF module for Apache httpd, Nginx and IIS from EG Secure Solutions, has an OS command injection in its statistics feature that can run arbitrary OS commands as root (or SYSTEM). The statistics feature is enabled by default; JPCERT/CC scores it CVSS 4.0 8.5 (local, low-privileged attacker). EG Secure Solutions and JVN confirm exploitation in the wild, and CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-20.</description></item><item><title>CVE-2026-113672: Argo CD Jsonnet imports read any file in repo-server, including credentials</title><link>https://www.secureroot.com/cve/CVE-2026-113672</link><guid>https://www.secureroot.com/cve/CVE-2026-113672</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>Argo CD evaluates .jsonnet files in argocd-repo-server with an importer that does not confine import, importstr or importbin to the repository, so absolute paths, ../ traversal or symlinks can read any file the repo-server can read, such as /proc/self/environ, the service-account token, mounted private keys and repository credentials (returned by placing them in a ConfigMap). importstr of /dev/zero can also exhaust repo-server memory. An attacker needs to commit a .jsonnet file to a repository a</description></item><item><title>CVE-2026-113708: Argo CD Kustomize remote ref option injection runs commands in repo-server</title><link>https://www.secureroot.com/cve/CVE-2026-113708</link><guid>https://www.secureroot.com/cve/CVE-2026-113708</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>When Argo CD builds a Kustomization, Kustomize passes the ref or version value of a remote resource, base or component URL to git fetch without a &#x27;--&#x27; separator, so a value such as --upload-pack=&lt;command&gt; with a file:// remote runs that command inside argocd-repo-server. The command runs as the repo-server user, can read repository credentials, and its output comes back in the manifest-generation error. An attacker only needs to commit a Kustomization to a Git repository an Application builds, o</description></item><item><title>CVE-2026-96652: Plex Media Server unauthenticated SSRF via /player/timeline (Plex2Shell)</title><link>https://www.secureroot.com/cve/CVE-2026-96652</link><guid>https://www.secureroot.com/cve/CVE-2026-96652</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>Plex Media Server before 1.43.3.10861 has a server-side request forgery in /player/timeline: with any X-Plex-Token value an attacker can put a full URL in the &#x27;protocol&#x27; parameter and make the server POST to a destination of their choice, including itself on 127.0.0.1. Chained together, CVE-2026-96652, -96654, -96655 and -96656 let an unauthenticated remote attacker run code as the Plex Media Server user; Plex says they are being actively and widely exploited since a public proof of concept appe</description></item><item><title>CVE-2026-96654: Plex Media Server searchOne injection reaches loopback agent API (Plex2Shell)</title><link>https://www.secureroot.com/cve/CVE-2026-96654</link><guid>https://www.secureroot.com/cve/CVE-2026-96654</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>Plex Media Server before 1.43.3.10861 does not neutralize URL values passed to the loopback-only agent function &#x27;searchOne&#x27;, so a request that reaches it over loopback can call other plugins&#x27; functions with attacker-supplied parameters; Plex describes it as access to internal agent functions over the loopback interface. Chained together, CVE-2026-96652, -96654, -96655 and -96656 let an unauthenticated remote attacker run code as the Plex Media Server user; Plex says they are being actively and w</description></item><item><title>CVE-2026-96655: Plex Media Server owner admin token disclosure / transcode SSRF (Plex2Shell)</title><link>https://www.secureroot.com/cve/CVE-2026-96655</link><guid>https://www.secureroot.com/cve/CVE-2026-96655</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>Plex Media Server before 1.43.3.10861 lets an authenticated user make the server request arbitrary internal or external addresses via the path parameter of /video/:/transcode/universal; Plex lists the issue as disclosure of the server owner&#x27;s admin token. Chained together, CVE-2026-96652, -96654, -96655 and -96656 let an unauthenticated remote attacker run code as the Plex Media Server user; Plex says they are being actively and widely exploited since a public proof of concept appeared on Octobe</description></item><item><title>CVE-2026-96656: Plex Media Server file write to Codecs dir via transcoder prefs (Plex2Shell)</title><link>https://www.secureroot.com/cve/CVE-2026-96656</link><guid>https://www.secureroot.com/cve/CVE-2026-96656</guid><pubDate>Tue, 20 Oct 2026 00:00:00 +0000</pubDate><description>Plex Media Server before 1.43.3.10861 lets an admin (owner-token) user write arbitrary files that get executed: the TranscoderH264Options preference is appended verbatim to x264&#x27;s options, so x264&#x27;s dump-yuv can write a file into the Codecs directory, and every .so there is loaded at transcoder start without signature, execute-bit or symbol checks. Chained together, CVE-2026-96652, -96654, -96655 and -96656 let an unauthenticated remote attacker run code as the Plex Media Server user; Plex says </description></item><item><title>CVE-2026-112900: Pillow: Argument Injection in Ghostscript call, potentially leading into RCE</title><link>https://www.secureroot.com/cve/CVE-2026-112900</link><guid>https://www.secureroot.com/cve/CVE-2026-112900</guid><pubDate>Mon, 19 Oct 2026 00:00:00 +0000</pubDate><description>Pillow is a Python imaging library. From 2.6.0 until 13.0.0, Pillow is vulnerable to argument injection when invoking Ghostscript for EPS input. EpsImagePlugin.Ghostscript passes a caller-controlled input filename after -f without converting it to a safe absolute path, so a name beginning with a hyphen can be interpreted as a Ghostscript option. An attacker who controls the filename can inject Ghostscript arguments. Arbitrary command execution specifically requires a leading @ argument-file file</description></item><item><title>CVE-2026-101304: Multiple jail filesystem root escapes</title><link>https://www.secureroot.com/cve/CVE-2026-101304</link><guid>https://www.secureroot.com/cve/CVE-2026-101304</guid><pubDate>Thu, 15 Oct 2026 00:00:00 +0000</pubDate><description>When fdescfs(4) was mounted in a jail with the &quot;nodup&quot; option, opening /dev/fd/N returned a descriptor that did not inherit FD_RESOLVE_BENEATH or the Capsicum capability rights of descriptor N.

A process in a jail that has received a directory file descriptor from another jail can use these techniques to escape the jail&#x27;s filesystem root restriction.</description></item><item><title>CVE-2026-101307: Remotely triggerable buffer overflow in tcp_rack</title><link>https://www.secureroot.com/cve/CVE-2026-101307</link><guid>https://www.secureroot.com/cve/CVE-2026-101307</guid><pubDate>Thu, 15 Oct 2026 00:00:00 +0000</pubDate><description>When the PCM sample array needs to grow, the RACK stack doubles a 16-bit allocation counter.  Repeated doublings caused the counter to wrap to zero, resulting in a near-zero-length allocation followed by a copy of the existing samples into the undersized buffer.  A remote peer can trigger the expansion by sending a sufficient number of ACKs on an established connection.

A remote attacker who can establish a TCP connection to a system using the RACK stack can trigger an out-of-bounds write of ap</description></item><item><title>CVE-2026-111094: TeamCity On-Premises RCE via whitelist bypass in the build messages protocol</title><link>https://www.secureroot.com/cve/CVE-2026-111094</link><guid>https://www.secureroot.com/cve/CVE-2026-111094</guid><pubDate>Thu, 15 Oct 2026 00:00:00 +0000</pubDate><description>JetBrains TeamCity On-Premises allowed remote code execution through a bypass of the whitelist in the build messages protocol (TW-103081). JetBrains scores it CVSS 8.1 with no privileges required (AV:N/AC:H/PR:N), while NVD scores it 9.8 critical. It was fixed in 2026.2, 2026.1.4 and 2025.11.8 together with other high-severity issues, including a Kotlin DSL sandbox escape and an administrator account takeover via password reset.</description></item><item><title>CVE-2026-21589: Atlassian Data Center products: unauthenticated arbitrary file access</title><link>https://www.secureroot.com/cve/CVE-2026-21589</link><guid>https://www.secureroot.com/cve/CVE-2026-21589</guid><pubDate>Thu, 15 Oct 2026 00:00:00 +0000</pubDate><description>A path traversal flaw in Bitbucket, Confluence, Jira Software, Jira Service Management, Bamboo and Crowd Data Center, as well as Crucible and Fisheye, lets an unauthenticated attacker read specific files inside the web application root directory if they know the exact file name and path (directory listing is not possible). Atlassian says all versions before the fixed releases announced on October 5, 2026 are affected and rates it critical (CVSS 4.0 score 9.3); Atlassian Cloud is already patched.</description></item><item><title>CVE-2026-105133: AhsayCBS API authentication bypass (checkSysPwd)</title><link>https://www.secureroot.com/cve/CVE-2026-105133</link><guid>https://www.secureroot.com/cve/CVE-2026-105133</guid><pubDate>Wed, 14 Oct 2026 00:00:00 +0000</pubDate><description>AhsayCBS up to 9.15.2.69 and 10.3.4.44 has an improper authentication flaw in checkSysPwd (com/ahsay/obs/api/ApiStructsAction.java): manipulating the &#x27;random&#x27; argument lets a remote attacker bypass authentication. Huntress observed the two flaws chained in the wild from 2026-10-07 to drop JSP web shells and XMRig miners; Ahsay&#x27;s first patches (9.15.2.59 and 10.3.4.0) were incomplete and the full fix is the 9.15.2.70 and 10.3.4.45 hotfixes. CISA added both CVEs to the Known Exploited Vulnerabilit</description></item><item><title>CVE-2026-105134: AhsayCBS Replication Receiver OS command injection (UpdateReceivers.do)</title><link>https://www.secureroot.com/cve/CVE-2026-105134</link><guid>https://www.secureroot.com/cve/CVE-2026-105134</guid><pubDate>Wed, 14 Oct 2026 00:00:00 +0000</pubDate><description>AhsayCBS up to 9.15.2.69 and 10.3.4.44 has an OS command injection in the Replication Receiver endpoint /rps/api/json/UpdateReceivers.do via the &#x27;random&#x27; argument, giving remote code execution on the backup server. Huntress observed the two flaws chained in the wild from 2026-10-07 to drop JSP web shells and XMRig miners; Ahsay&#x27;s first patches (9.15.2.59 and 10.3.4.0) were incomplete and the full fix is the 9.15.2.70 and 10.3.4.45 hotfixes. CISA added both CVEs to the Known Exploited Vulnerabili</description></item><item><title>CVE-2026-59774: Gitea Org-mode #+INCLUDE lets anonymous users read server files, then RCE</title><link>https://www.secureroot.com/cve/CVE-2026-59774</link><guid>https://www.secureroot.com/cve/CVE-2026-59774</guid><pubDate>Wed, 14 Oct 2026 00:00:00 +0000</pubDate><description>Gitea&#x27;s repository markup endpoint (POST /{owner}/{repo}/markup) renders Org-mode text with a #+INCLUDE directive that can read any absolute path, and anonymous users may call it for any public repository. An unauthenticated attacker can read files readable by the Gitea service account, including app.ini; with the INTERNAL_TOKEN from it they can inject a Git hook through the internal logger and run commands as the Gitea OS user. Found by XBOW and fixed in Gitea 1.27.1 together with CVE-2026-6000</description></item><item><title>CVE-2026-103903: MikroTik RouterOS WinBox: low-privilege user can get admin session</title><link>https://www.secureroot.com/cve/CVE-2026-103903</link><guid>https://www.secureroot.com/cve/CVE-2026-103903</guid><pubDate>Mon, 12 Oct 2026 00:00:00 +0000</pubDate><description>In the RouterOS 7 WinBox login flow, the first stage is checked against the attacker&#x27;s own account while a separately supplied second-stage username decides which account&#x27;s rights are installed. A user with any WinBox-capable login can therefore get a full administrative session as admin without its password and raise their own account&#x27;s rights. It affects RouterOS 7.0 before 7.24.5; the fix is in 7.24.5, 7.23.8 and 7.25.</description></item><item><title>CVE-2015-3306: ProFTPD mod_copy unauthenticated SITE CPFR/CPTO file copy</title><link>https://www.secureroot.com/cve/CVE-2015-3306</link><guid>https://www.secureroot.com/cve/CVE-2015-3306</guid><pubDate>Thu, 08 Oct 2026 00:00:00 +0000</pubDate><description>ProFTPD&#x27;s mod_copy module in 1.3.5 lets unauthenticated clients use the SITE CPFR and SITE CPTO commands to copy arbitrary files on the server, which public exploits (including a Metasploit module) turn into remote code execution by copying content into a web-served directory. It was fixed in ProFTPD 1.3.5a and 1.3.6rc1 in 2015; CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-08.</description></item><item><title>CVE-2015-5477: ISC BIND TKEY query REQUIRE assertion failure (remote named crash)</title><link>https://www.secureroot.com/cve/CVE-2015-5477</link><guid>https://www.secureroot.com/cve/CVE-2015-5477</guid><pubDate>Thu, 08 Oct 2026 00:00:00 +0000</pubDate><description>A single crafted DNS query for a TKEY record makes BIND&#x27;s named hit a REQUIRE assertion and exit, a remote unauthenticated denial of service against both recursive and authoritative servers. ACLs and configuration options do not prevent it because the faulty code runs before those checks, and ISC lists no workarounds. It was fixed in July 2015 (BIND 9.9.7-P2 and 9.10.2-P3); CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-08, so any still-unpatched legacy BIND servers shou</description></item><item><title>CVE-2016-3081: Apache Struts 2 S2-032 RCE via method: prefix with Dynamic Method Invocation</title><link>https://www.secureroot.com/cve/CVE-2016-3081</link><guid>https://www.secureroot.com/cve/CVE-2016-3081</guid><pubDate>Thu, 08 Oct 2026 00:00:00 +0000</pubDate><description>Apache Struts 2.3.19 to 2.3.28 (except 2.3.20.3 and 2.3.24.3) let remote attackers pass a malicious OGNL expression through the method: prefix and execute arbitrary code on the server when Dynamic Method Invocation (DMI) is enabled. Apache fixed it in April 2016 (S2-032) in 2.3.20.3, 2.3.24.3 and 2.3.28.1. CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-08, so legacy Struts 2.3 web applications should be checked and rebuilt.</description></item><item><title>CVE-2021-3199: ONLYOFFICE Document Server /upload path traversal to RCE</title><link>https://www.secureroot.com/cve/CVE-2021-3199</link><guid>https://www.secureroot.com/cve/CVE-2021-3199</guid><pubDate>Thu, 08 Oct 2026 00:00:00 +0000</pubDate><description>ONLYOFFICE Document Server before 5.6.3, when JWT is used, has a directory traversal in /upload through a /.. sequence in an image-upload parameter, which can lead to remote code execution; public proof-of-concept code exists. It was fixed in Document Server 5.6.3 (&#x27;Fix Path Traversal vulnerability via image upload params&#x27;, bug #46113). CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-08.</description></item><item><title>CVE-2023-22894: Strapi private-field filtering leaks password hashes and reset tokens</title><link>https://www.secureroot.com/cve/CVE-2023-22894</link><guid>https://www.secureroot.com/cve/CVE-2023-22894</guid><pubDate>Thu, 08 Oct 2026 00:00:00 +0000</pubDate><description>Strapi 3.2.1 up to (not including) 4.8.0 lets API query filters reference private user columns, so an attacker can infer emails, password hashes and password reset tokens from API responses; Strapi also disclosed a method to exploit it without authentication to hijack admin accounts. Strapi 3.x is end of life and gets no fix; upgrade to 4.8.0 or later. CISA added it to the Known Exploited Vulnerabilities catalog on 2026-10-08.</description></item><item><title>CVE-2026-106553: OpenSSH: In sshd in OpenSSH before 10.6, credentials can incorrectly persist after failure of a GSS…</title><link>https://www.secureroot.com/cve/CVE-2026-106553</link><guid>https://www.secureroot.com/cve/CVE-2026-106553</guid><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><description>In sshd in OpenSSH before 10.6, credentials can incorrectly persist after failure of a GSSAPIAuthentication authentication attempt.</description></item><item><title>CVE-2026-106555: OpenSSH: In sshd in OpenSSH before 10.6, GSSAPIAuthentication authentication state can incorrectly…</title><link>https://www.secureroot.com/cve/CVE-2026-106555</link><guid>https://www.secureroot.com/cve/CVE-2026-106555</guid><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><description>In sshd in OpenSSH before 10.6, GSSAPIAuthentication authentication state can incorrectly be persisted across authentication attempts.</description></item><item><title>CVE-2026-106583: OpenSSH: In ssh in OpenSSH before 10.6, a $ or \ character can occur in a command-line username, le…</title><link>https://www.secureroot.com/cve/CVE-2026-106583</link><guid>https://www.secureroot.com/cve/CVE-2026-106583</guid><pubDate>Tue, 06 Oct 2026 00:00:00 +0000</pubDate><description>In ssh in OpenSSH before 10.6, a $ or \ character can occur in a command-line username, leading to injection.</description></item><item><title>CVE-2026-102489: Zammad 6.3-6.5 session hijack leading to RCE as the zammad user</title><link>https://www.secureroot.com/cve/CVE-2026-102489</link><guid>https://www.secureroot.com/cve/CVE-2026-102489</guid><pubDate>Fri, 02 Oct 2026 00:00:00 +0000</pubDate><description>Zammad 6.3.0 to 6.5.4 has a session hijack (session fixation) flaw that leads to remote code execution as the zammad user; the code is also present in 7.0.0 to 7.1.x but not exploitable because of the newer runtime environment. DIVD found it after attackers used it, chained with the CVE-2026-102490 local root flaw, to breach DIVD in September 2026. Zammad hardened the code in 7.2.0 and shipped an emergency 6.5.5 release for the unsupported 6.5 series; CISA added it to the Known Exploited Vulnera</description></item><item><title>CVE-2026-102490: Zammad DEB/RPM packages (packager.io) local privilege escalation to root</title><link>https://www.secureroot.com/cve/CVE-2026-102490</link><guid>https://www.secureroot.com/cve/CVE-2026-102490</guid><pubDate>Fri, 02 Oct 2026 00:00:00 +0000</pubDate><description>In Zammad installed from the packager.io DEB/RPM packages, service processes started as root and executed files owned and writable by the zammad account before dropping privileges, so anyone with code execution as the zammad user could become root within seconds (services auto-restart). All packaged releases from 1.5.0 through 7.2.1 are affected; source installs and the official container images are not. DIVD found it while investigating the breach of its own network, chained with CVE-2026-10248</description></item><item><title>CVE-2026-84411: MikroTik RouterOS web interface: one request can run code as root</title><link>https://www.secureroot.com/cve/CVE-2026-84411</link><guid>https://www.secureroot.com/cve/CVE-2026-84411</guid><pubDate>Fri, 02 Oct 2026 00:00:00 +0000</pubDate><description>An integer underflow in HTTP request body handling in the RouterOS web management service (the WebFig service on the www and www-ssl ports) is reachable before login. A single crafted request from an unauthenticated attacker can crash the service or run code on the router with full privileges (CVSS 9.8). It affects RouterOS before 7.24, including RouterOS 6; MikroTik fixed it in 7.24, 7.23.8 and 6.49.23. No exploitation was known when CISA published its advisory on 29 September 2026.</description></item><item><title>CVE-2026-84440: IBM Guardium Data Protection is affected by multiple vulnerabilities.</title><link>https://www.secureroot.com/cve/CVE-2026-84440</link><guid>https://www.secureroot.com/cve/CVE-2026-84440</guid><pubDate>Tue, 29 Sep 2026 00:00:00 +0000</pubDate><description>IBM Guardium Data Protection 12.2 is vulnerable to command injection in the SNMP alert notification functionality. An authenticated attacker who can influence policy alert text can cause attacker-controlled data to be executed as operating system commands by the SNMP alerter service, which runs with root privileges.</description></item><item><title>CVE-2026-84782: OpenSSL DTLS retransmission heap over-read (High)</title><link>https://www.secureroot.com/cve/CVE-2026-84782</link><guid>https://www.secureroot.com/cve/CVE-2026-84782</guid><pubDate>Tue, 29 Sep 2026 00:00:00 +0000</pubDate><description>OpenSSL&#x27;s DTLS retransmission logic mishandles a handshake write that was suspended part-way (WANT_WRITE): a retransmission can read past the message buffer and send leftover heap memory to the peer as plaintext handshake data, or crash the process. Only applications that use DTLS (TLS over UDP) are affected; plain TLS over TCP is not. OpenSSL rates it High; all branches from 1.0.2 to 4.0 are affected, and fixes for 3.0, 1.1.1 and 1.0.2 are available upstream only to premium support customers, s</description></item></channel></rss>
