jane-php versions prior to 7.12.0 have a vulnerability in their $ref resolution during code generation. The software resolves external $ref targets using file_get_contents without restricting the URI scheme or path, allowing attacker-controlled OpenAPI documents to trigger server-side request forgery (SSRF) and local file inclusion (LFI) via path traversal. This can lead to arbitrary HTTP requests and local file reads at build or CI time, with the fetched content inlined into generated code.
Refitter versions prior to 2.1.2 contain a critical vulnerability (CVE-2026-63344) where unescaped OpenAPI content-type keys are directly embedded into Refit attribute strings. This allows an attacker to inject arbitrary code that executes at assembly load time via a crafted content-type key in the OpenAPI specification. The vulnerability has a CVSS 4.0 score of 9.3 and affects Refitter 2.0.0 and earlier. A fix is available in version 2.1.2.
Refitter versions prior to 2.1.2 contain a vulnerability where unescaped OpenAPI header parameter names are emitted directly into Refit attribute strings. This allows an attacker to inject code that executes at assembly load time via a crafted header parameter name in the OpenAPI specification. The flaw was verified on Refitter 2.0.0 with .NET 8 and can lead to code execution at application startup.
Refitter versions prior to 2.1.2 allow arbitrary code execution via unescaped OpenAPI path strings embedded directly into Refit attribute strings. An attacker can craft a single-line path that closes the attribute and interface context, injects a file-scoped class with a .NET ModuleInitializer method, and reopens namespaces/interfaces to maintain valid code. This injected code executes at assembly load, enabling code execution at application startup. The vulnerability affects Refitter 2.0.0 and earlier and is fixed in version 2.1.2.
NATS Server versions prior to 2.12.12 and between 2.14.0-RC.1 and before 2.14.3 have a vulnerability where the MQTT-over-WebSocket path can cause a crash on WebSocket-only JetStream servers if MQTT is not enabled. An unauthenticated client can trigger this by accessing the MQTT path, leading to a server process crash. This affects deployments that enable WebSocket but leave MQTT disabled. The issue is fixed in versions 2.12.12 and 2.14.3.
NATS Server versions prior to 2.12.12 and between 2.14.0-RC.1 and before 2.14.3 have a vulnerability in the MQTT CONNECT packet parser. An unauthenticated MQTT client can send incomplete CONNECT packets that cause the server to retain large amounts of memory until authentication times out, leading to memory exhaustion. This issue is fixed in versions 2.12.12 and 2.14.3. Deployments not using MQTT are not affected.
NATS Server versions up to 2.12.11 and between 2.14.0-RC.1 and before 2.14.3 contain a vulnerability in MQTT retained message and QoS1+ durable delivery. This flaw allows delivery of messages on topics that should be denied by subscription deny rules, bypassing normal live delivery checks. The issue affects MQTT subscribers with broad wildcard permissions but specific topic denials. Fixed in versions 2.12.12 and 2.14.3.
The Halo 2 aggregation circuit in the snark-verifier library mishandles proof elements encoded as the point at infinity (identity element). This occurs because the scalar-one shortcut path assumes all input points are non-identity, but assigned proof points can represent the identity (0,0). This flaw affects recursive aggregation circuits that rely on this assumption, potentially causing incorrect multi-scalar multiplication (MSM) or accumulator computations. The issue was fixed in version 0.2.6 by updating the sum helper to correctly handle identity inputs as no-ops. The vulnerability is rated medium severity with a CVSS score of 6.9.
NLTK versions from 3.9.4 up to but not including 3.10.4 contain a vulnerability in the nltk.pathsec.ZipFile archive extraction component. The extraction process fails to properly check for hardlinks inside the extraction root, allowing a local attacker with write access to the NLTK data root to pre-plant a hardlink that points outside the root. When NLTK extracts a package, it overwrites the target of that hardlink outside the data root, leading to arbitrary file truncation and overwriting. This can result in complete integrity loss of the targeted file. The vulnerability requires local access and suitable permissions to create hardlinks and is mitigated by a fix released in version 3.10.4.
NLTK versions 3.10.0 through 3.10.3 contain a vulnerability in the allowlisted pickle loader where numpy submodule file I/O functions were not properly denied. This allowed crafted pickle data to perform arbitrary file reads or file creations during unpickling when numpy is installed. The issue was fixed in version 3.10.4 by expanding the denylist to cover numpy submodules and restricting allowed globals in model loading.
NLTK versions from 3.9.4 up to but not including 3.10.4 contain a vulnerability in the CorpusReader view methods where intermediate-directory symlinks inside the corpus root can bypass scoped root sandbox protections. This allows reading files outside the intended corpus root but within the global allowed roots, defeating intended path validation. The issue arises because view methods use a different file access path that does not enforce the scoped root check, allowing symlink traversal. The vulnerability is fixed in version 3.10.4.
NETAPP StorageGRID versions 12.1.0 up to but not including 12.1.0.2 have an information exposure vulnerability. This flaw allows a remote unauthenticated attacker to obtain low-sensitivity credentials via a man-in-the-middle attack when certain non-default features are enabled. The vulnerability has a low severity score and does not involve high-impact conditions such as privilege escalation or data integrity compromise. A fix is available in version 12.1.0.2.
NLTK version 3.10.3 contains an incomplete fix for JVM argument injection and untrusted JAR execution vulnerabilities in the nltk.internals.java() function. While the patch blocks unsafe launcher tokens in the 'options' argument, it fails to validate the first item of the 'cmd' argument sequence, allowing an attacker to supply launcher tokens such as '-jar' or '@argfile' to execute arbitrary external JAR files. This bypass affects both direct calls to nltk.internals.java() and the higher-level StanfordSegmenter interface. The issue is fixed in version 3.10.4.
NLTK versions 3.6 through 3.10.3 contain a vulnerability in the XMLCorpusView streaming XML reader where processing XML files with large comment or CDATA sections can cause a quadratic CPU denial of service. This occurs because the XML parsing function repeatedly performs expensive regex validations on growing buffers while reading the XML in 1 KiB blocks. The issue is fixed in version 3.10.4.
A vulnerability in the NLTK Python package downloader affects versions 3.6.7 through 3.10.3. The downloader's index update function performs a breadth-first search to compute package closures without cycle detection or depth limits. Malicious or compromised index data containing self-referencing or mutually referencing collections causes an infinite loop, leading to 100% CPU usage and unbounded memory growth until process termination. This flaw can cause denial of service during downloader operations that rely on index parsing.
NLTK's ARFF_Formatter component improperly handles class labels when generating Weka ARFF files, allowing attacker-controlled labels to inject arbitrary ARFF schema directives. This occurs because class labels are embedded without escaping, enabling injection of ARFF metacharacters that can alter the ARFF file structure. The vulnerability affects any application using WekaClassifier.train() or ARFF_Formatter.from_train() with untrusted training data. A fix has been merged but not yet released as of the latest information.
The Natural Language Toolkit (NLTK) DependencyGraph component uses unbounded recursion in its tree() and triples() methods to walk parsed dependency structures, which can lead to a crash via RecursionError when processing crafted long dependency chains. This occurs because these methods do not use the class's cycle detection and lack recursion depth limits, unlike other parts of NLTK. An attacker supplying a maliciously long dependency-parse file can cause denial of service by crashing the process. This issue affects versions prior to 3.10.4 and has been fixed in 3.10.4.
NLTK versions prior to 3.10.4 contain a vulnerability in the Synset.tree() and Synset.acyclic_tree() methods where recursive traversal lacks a maximum call-depth safeguard. This can lead to a crash via unhandled RecursionError when processing a long, genuinely acyclic relation chain in a custom WordNet-format lexical resource. The default bundled WordNet corpus is not affected due to its shallow relation depths. The vulnerability does not affect Synset.closure(), which uses an iterative traversal method. A fix is available in version 3.10.4.
A malicious file identified by the SHA-256 hash a74f60e5deac9ecf7953ee0f855a5a493832b6b143141e75526bf954720f303f was observed twice on November 19, 2026, by OffSeq Mirage honeypot sensors. The file is associated with attacker activity involving the MITRE ATT&CK technique T1105 (Ingress Tool Transfer). The malicious payload could be a dropper, web shell, miner, or post-exploitation tool delivered or staged during the attack. The attacker infrastructure was located in Vietnam. This indicator can be used to detect the presence of this malicious file in environments by matching the hash against endpoint detection and response (EDR), antivirus (AV), and threat intelligence feeds.
NLTK's CCG lexicon parser suffers from uncontrolled recursion when parsing deeply nested parenthesized categories, leading to a denial of service via an unhandled RecursionError. This occurs because the parser functions do not limit recursion depth or input length, allowing crafted inputs to exceed Python's recursion limit and abort parsing. The issue affects versions prior to 3.10.4 and has been fixed in version 3.10.4.