CVE Feed

BusyBox dpkg read_package_field() steps past a NUL terminator on malformed .deb packages, causing an out-of-bounds heap read.

BusyBox httpd IP deny rules with invalid CIDR prefix lengths fail open, leaving a parsed IP with a zeroed mask so the rule matches no clients.

A DLL hijacking vulnerability in the FileMaker Pro installer for Windows allowed a local user to execute arbitrary code with elevated administrator privileges by placing a malicious DLL file in the installer directory. This vulnerability is addressed in FileMaker Pro version 26.0.3.

An authorization bypass vulnerability in the FileMaker Server Web Publishing Engine allowed requests containing an extended privilege header to bypass the disabled Custom Web Publishing with XML setting and access the XML Web Publishing interface. This vulnerability is addressed in FileMaker Server version 26.0.3.

An out-of-bounds read vulnerability in FileMaker Server for Linux allowed an attacker uploading a specially crafted image file to a container field to disclose process memory during thumbnail generation in FileMaker WebDirect. This vulnerability is addressed in FileMaker Server version 26.0.3.

A heap buffer overflow vulnerability in the FileMaker Server database engine block parsing routine allowed a maliciously crafted .fmp12 database file to cause memory corruption, potentially leading to arbitrary code execution. This vulnerability is addressed in FileMaker Server version 26.0.3.

Cinnamon's Kotaemon (all versions up to and including v0.12.0) multi-user chat interface contains multiple vulnerabilities due to incorrect authorization and improper access controls. There are four handler methods in `libs/ktem/ktem/pages/chat/control.py` that load a Conversation record by its ID without comparing the requester's `user_id` to the conversation's owner `Conversation.user`. This allows any authenticated user to perform the following actions: 1. Read other user's chat transcripts, RAG retrieval history, AI-generated plots, and chat suggestions. 2. Permanently delete another user's conversation. 3. Rename another user's conversation. 4. Overwrite another user's conversation's chat suggestion list.

Rejected reason: ** REJECT ** DO NOT USE THIS CANDIDATE NUMBER. ConsultIDs: CVE-2026-18944. Reason: This candidate is a reservation duplicate of CVE-2026-18944. Notes: All CVE users should reference CVE-2026-18944 instead of this candidate. All references and descriptions in this candidate have been removed to prevent accidental usage.

In Flatpak before 1.18.1, the revokefs writer, used by the flatpak-system-helper to receive repository data from unprivileged callers, validated file paths by rejecting literal .. components but did not prevent symlink traversal. A malicious local user in an active local session could obtain two revokefs sessions via the system helper, create a symlink in one session pointing into the other session's directory, and retain a file descriptor through that symlink. This allowed the attacker to modify files belonging to a different revokefs session after they had been validated and imported by the system helper. In particular, an attacker could use this to tamper with ostree commit objects in the system repository after they passed signature verification, enabling root-controlled file writes to attacker-chosen paths and local root privilege escalation.

In Flatpak before 1.18.1, a malicious sandboxed app can replace ~/.var/app/$appid/.ld.so with a symlink, causing regenerate_ld_cache to write files at an arbitrary location. The filenames and content are not attacker controlled, making this hard to exploit.

MLflow's statsmodel flavor, versions 2.1.0 to 3.14.0, omits the MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False security control entirely in _load_model(), which allows a remote attacker to execute arbitrary code via a crafted MLmodel artifact.

MLflow's dspy flavor, versions >= 2.0, applies the MLFLOW_ALLOW_PICKLE_DESERIALIZATION=False security control only when the model_path ends in .pkl, which allows a remote attacker to execute arbitrary code via a crafted MLmodel artifact.

orval before 8.29.0 fails to escape the operationId parameter when emitting it into generated TanStack Query mutator options metadata objects. Attackers can inject arbitrary JavaScript code through a crafted operationId in an OpenAPI specification that executes when generated hooks are called.

orval @orval/core before 8.28.0 contains a code injection vulnerability in the form-data serializer that fails to escape multipart property names in generated template literals. Attackers can inject ${...} expressions into OpenAPI schema property names that execute as live interpolation when the generated client builds FormData bodies with consumer process privileges.

orval before 8.29.0 fails to escape OpenAPI media-type keys when emitting them into single-quoted Content-Type string literals in generated code. Attackers can inject JavaScript through crafted media-type keys in OpenAPI specifications that executes when generated fetch operations or mock resolvers are invoked.

orval versions before 8.30.0 contain a code injection vulnerability in the @orval/core factory generator that fails to escape date default values in new Date() calls. Attackers can inject arbitrary expressions through apostrophes in OpenAPI schema defaults to execute code with the privileges of the consumer process when factoryMethods and useDates options are enabled.

orval versions 8.14.0 through 8.28.1 contain a code injection vulnerability in the @orval/effect generator that converts OpenAPI schema defaults into template literals. Attackers can inject arbitrary JavaScript expressions via schema defaults containing ${...} syntax, which are executed at module scope when the generated code is built or imported.

orval versions before 8.29.0 contain a code injection vulnerability in the @orval/hono generator that fails to escape OpenAPI path values in single-quoted route literals. Attackers can craft an OpenAPI document with an apostrophe in a static path segment to inject arbitrary JavaScript code that executes when the generated TypeScript module is imported.

Plex Media Server before 1.43.3.10861 allows an admin user to write arbitrary files that may be executed on load. The preference TranscoderH264Options is appended verbatim to x264's option string on every transcode. At startup, all .so files are run without signature, execute bit, or symbol checks.

Plex Media Server before 1.43.3.10861 allows an authenticated user to request arbitrary internal or external addresses via the '/video/:/transcode/universal' path parameter.

Plex Media Server before 1.43.3.10861 does not correctly neutralize URL values included in 'searchOne,' allowing an attacker to call other plugins' functions and supply their own parameters.

Plex Media Server before 1.43.3.10861 allows SSRF via '/player/timeline'. An attacker using any X-Plex-Token value can include a full URL in the 'protocol' parameter and force the Plex server to POST to the attacker's chosen destination.

Plex Media Server before 1.43.3.10861 builds a file path from the url parameter without checking it for ../ sequences, allowing path traversal via '/system/agents/media/get'. A remote attacker with a valid session token could read any file that the target user can access. This access includes the PlexOnlineToken, which grants control of the Plex account and server. A LAN-adjacent attacker with a client-supplied X-Forwarded-For header could exploit the same issue.

A weakness has been identified in Neethuharii CafeManagement. Impacted is an unknown function of the file CafePortalLogin.php of the component Login Handler. This manipulation of the argument uname causes sql injection. It is possible to initiate the attack remotely. The exploit has been made available to the public and could be used for attacks. This product is using a rolling release to provide continious delivery. Therefore, no version details for affected nor updated releases are available. The vendor was contacted early about this disclosure but did not respond in any way.

A security flaw has been discovered in Neethuharii CafeManagement. This issue affects some unknown processing of the file AddProductCode.php. The manipulation of the argument image results in unrestricted upload. The attack may be performed from remote. The exploit has been released to the public and may be used for attacks. This product utilizes a rolling release system for continuous delivery, and as such, version information for affected or updated releases is not disclosed. The vendor was contacted early about this disclosure but did not respond in any way.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, when a configured authenticator or authorizator class cannot be loaded, Server.initializeAuthenticator and Server.initializeAuthorizatorPolicy treat the failure as though no custom class was configured and fall back to AcceptAllAuthenticator or PermitAllAuthorizatorPolicy. A misspelled class name, missing dependency, constructor failure, or classpath problem can therefore start the broker with authentication or authorization disabled even though the operator configured those controls. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, H2PersistentQueue derives a session's message-map name as queue_ plus the client ID and its metadata-map name as queue_ plus the client ID plus _meta. A durable session whose client ID ends in _meta can therefore make its message map collide with another client's metadata map. The colliding sessions read and write the same H2 MVStore map with incompatible value types, which can corrupt queue head and tail data and cause message loss, misdelivery, failed queue reloads, or exposure of queued content across sessions. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, PostOffice.publishWill publishes a client's Last-Will message without applying the canWrite authorization and reserved-topic checks used for a normal PUBLISH. A client can configure a Will for a topic that the client is not permitted to write and cause the broker to publish the unauthorized message when the client disconnects unexpectedly. This issue allows unauthorized message injection into restricted topics. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, the broker does not enforce a maximum length for pending per-session message queues. When a fast publisher sends messages to a slow subscriber whose in-flight window is full, queued messages can accumulate without bound in memory or persistent storage. Remote clients can use this condition to exhaust broker resources and cause a denial of service. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, Moquette does not limit the depth of topic names and topic filters before processing them through recursive CTrie insertion and matching operations. A remote client can publish or subscribe with a deeply nested topic, causing a StackOverflowError that disrupts session processing and can deny service to broker clients. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, PostOffice.subscribe parses a shared-subscription filter through SharedSubscriptionUtils.extractShareName before validating the complete $share/{shareName}/{topicFilter} structure. A remote client can send a filter such as $share/grp without a topic-filter portion, causing a StringIndexOutOfBoundsException while calculating the share name. The exception terminates command handling on the shared session event loop and can deny service to other client sessions assigned to that loop. This issue is fixed in version 0.18.1.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, SessionEventLoop.run catches only InterruptedException, and SessionEventLoopGroup does not restart a terminated loop. An MQTT command that raises an uncaught exception can terminate an event loop shared by multiple client sessions, preventing every co-located client from processing PUBLISH, SUBSCRIBE, PUBACK, and other commands. An attacker can select client IDs that map across the available loops to disrupt session processing for the entire broker. This issue is fixed in version 0.18.1.

Frictionless before 5.19.1 contains an OS command injection vulnerability in the explore console command that allows an attacker who supplies a crafted Data Package descriptor to execute arbitrary operating system commands as the user who explores it. Attackers can place shell metacharacters in resource path values within a datapackage.json descriptor, which are passed unsanitized to os.system through a shell, causing arbitrary command execution in the victim's security context when they run the explore command against the untrusted package. This vulnerability was also addressed in version 5.20.0rc2 of the pre-release branch.

BusyBox romfs volume ID parsing uses unbounded strlen on attacker-controlled metadata, causing a heap buffer overflow when processing crafted filesystem images.

A unit confusion in BusyBox TLS Montgomery reduction buffer allocation causes a pre-authentication heap buffer overflow when processing a crafted ClientKeyExchange message.

Moquette is a lightweight Java MQTT broker. Prior to 0.18.1, when pattern-based ACL rules are configured, AuthorizationsCollector.canDoOperation substitutes client ID and username values directly into rules containing %c or %u and then treats the result as an MQTT topic filter. A client that uses + or # in either identity can broaden the substituted filter and gain cross-tenant read and write access. A # identity can also produce an invalid filter that triggers a NullPointerException in Topic.match and disrupts session processing. This issue is fixed in version 0.18.1.

Integer overflow in the packet buffer growth logic in PgBouncer through 1.25.2 allows an unauthenticated remote attacker to cause a denial of service. Sufficiently large input makes the buffer size computation overflow, leaving the growth loop unable to terminate. Because PgBouncer serves all clients from a single process, this saturates a CPU core and stalls every pooled connection until the process is killed. Both unauthenticated and authenticated code paths can reach the overflow.

Missing validation of a mandatory attribute in the SCRAM client-final-message parser in PgBouncer through 1.25.2 allows an unauthenticated remote attacker to crash the process. A malformed message can make the parser report success while leaving a required value unset, which is then dereferenced as a NULL pointer. The crash occurs before any credential is verified, so no valid account is required. Because PgBouncer serves all clients from a single process, this terminates every pooled connection.

SQL injection vulnerability in Moodle Socialwall plugin v.3.0 through v.3.3 allows an attacker to execute arbitrary code via crafted HTTP requests