CVE Intelligence
Início / CVEs

Base de CVEs

Vulnerabilidades publicadas, com severidade, exploração conhecida e fontes oficiais.

Severidade
Limpar

2.140 resultado(s)

Exportar esta busca: RSS · CSV · STIX
CVE-2026-108947 Média · 5.3 CWE-20 12/10/2026
yoshitomo-matsubara torchdistill main_util.py load_ckpt deserialization
A flaw has been found in yoshitomo-matsubara torchdistill up to 1.1.5. This affects the function load_ckpt of the file torchdistill/common/main_util.py. Executing a manipulation of the argument src_ckpt can lead to deserialization. The attack needs to be launched locally. The exploit has been published and may be used. The project maintainer explains: "This is very similar to an internal report that GH auto-generated for the owner before, which I found tolerable for this project at that time. So I will close this as duplicate."
CVE-2026-84317 Alta · 8.8 CWE-20 12/10/2026
Nginx UI: MCP nginx_config_add skips config directive validation → authenticated nginx directive injection (RCE on the official image)
Nginx UI is a web user interface for the Nginx web server. Prior to 2.4.1, internal/config/validation.go (ValidateConfigFile) is a security control that blocks dangerous nginx directives — js_*, perl_*, load_module of the njs/perl modules, and user root. It is enforced on three of the four config-write paths: api/config/add.go, api/config/modify.go, and mcp/config/config_modify.go. The fourth path — the MCP tool nginx_config_add (mcp/config/config_add.go) — does NOT call it: it writes the caller-supplied content straight to disk and reloads nginx. An authenticated MCP caller (or an MCP agent induced via prompt injection) can therefore inject exactly the directives the control is meant to block, leading to nginx code execution on deployments where the njs/perl module is available (e.g. the official Docker image). This issue is fixed in version 2.4.1.
CVE-2026-84316 Alta · 8.8 CWE-20 12/10/2026
Nginx UI: nginx config directive denylist (ValidateConfigFile) is bypassable via same-line ";" separation → authenticated directive injection (RCE on the official image)
Nginx UI is a web user interface for the Nginx web server. Prior to 2.4.1, config.ValidateConfigDirectives splits configuration by newline and checks only the first directive token on each line even though Nginx permits multiple semicolon-separated directives on one line. An authenticated caller can place a benign directive first and a restricted load_module, js_import, perl, or user root directive later on the same line, or can use an unrestricted include directive to load restricted content. These mechanisms bypass validation on REST, nginx_config_modify, site, and stream write paths; when an executable module is available, the injected configuration can execute code in the Nginx worker context. This issue is fixed in version 2.4.1.
CVE-2026-78673 Alta · 7.5 CWE-20 12/10/2026
Maximo AI Service Improper Input Validation
IBM Maximo AI Service 9.1.0, and 9.2.0 could allow a remote attacker to cause a denial of service due to improper input validation.
CVE-2026-104364 Baixa · 3.3 CWE-20 12/10/2026
IBM® Db2® Denial of Service
IBM Db2 11.5.0 through 11.5.9, and 12.1.0 through 12.1.5 could allow a local attacker to cause a denial of service due to improper input validation.
CVE-2026-57143 Alta · 8.8 CWE-20 12/10/2026
PraisonAI: Server-Side Request Forgery (SSRF) in SearxNG / search_web tools via attacker-controlled searxng_url parameter
PraisonAI is a multi-agent teams system. Prior to praisonaiagents 1.6.62, searxng_search and the search_web SearxNG provider pass the LLM-controlled searxng_url parameter directly to requests.get without validating its scheme, host, port, or resolved address. Prompt-injected content can direct the default search tool to internal JSON services, distinguish open and closed internal ports, and attempt cloud-metadata requests, with returned JSON results entering agent context. This issue is fixed in praisonaiagents 1.6.62.
CVE-2026-94451 Média · 6.5 CWE-20 12/10/2026
Quarkus-hibernate-orm-rest-data-panache: quarkus-hibernate-orm-rest-data-panache: information disclosure and dos via unvalidated sort parameter in rest data panache
Quarkus REST Data Panache provides a way to automatically generate REST endpoints for data entities. A flaw was found where certain endpoints do not properly check the sort parameter provided by users. This could allow an attacker to learn information they should not have access to or slow down the system by forcing the database to perform complex tasks, potentially causing a denial of service.
CVE-2026-84286 Crítica · 9.8 CWE-20 12/10/2026
OOB Read/Write via Negative Array Index in exllamav3_ext Kernel Table
ExLlamaV3’s EXL3 loader uses an untrusted, model-supplied index to select a CUDA kernel without first verifying that the index is within bounds. A malicious model can force the code to read and invoke an out‑of‑range function pointer, resulting in a reproducible crash. Under specific conditions involving certain host memory layouts, it may be possible for an attacker to exploit this vulnerability to achieve controlled indirect kernel dispatch during model load.
CVE-2026-109311 Média · 6.5 CWE-20 12/10/2026
vLLM before 0.29.0 Heap Overflow DoS via IAMF Audio Transcription Upload
vLLM before 0.29.0 passes uploaded audio to the PyAV/FFmpeg av.open() fallback with only a file-size check, exposing FFmpeg's IAMF parser heap overflow. Authenticated remote attackers can repeatedly upload malformed IAMF containers to /v1/audio/transcriptions, corrupting the heap and crashing the API process.
CVE-2026-93693 Média · 6.5 CWE-20 12/10/2026
Unauthenticated queue manipulation through connector SSRF and response injection in Progress ShareFile Storage Zones Controller
In Progress ShareFile Storage Zones Controller v5.12.6 and below, improper validation of outbound request destinations and parameters, unsafe JSONP response formatting, and insufficient validation on queue service can be chained by a remote unauthenticated attacker to issue server-side requests to unintended local handlers and perform unauthorized queue operations.
CVE-2026-21130 Média · 5.1 CWE-20 12/10/2026
Improper input validation in CreditCardRetrieveActivity of Samsung Checkout prior to version 5.1...
Improper input validation in CreditCardRetrieveActivity of Samsung Checkout prior to version 5.1.03.1 allows local attackers to launch an arbitrary activity with Samsung Checkout privileges.
CVE-2026-21129 Média · 5.1 CWE-20 12/10/2026
Improper input validation in GuestCreditCardRegisterActivity of Samsung Checkout prior to version...
Improper input validation in GuestCreditCardRegisterActivity of Samsung Checkout prior to version 5.1.03.1 allows local attackers to launch an arbitrary activity with Samsung Checkout privileges.
CVE-2026-21128 Média · 5.1 CWE-20 12/10/2026
Improper input validation in CreditCardRegisterActivity of Samsung Checkout prior to version 5.1...
Improper input validation in CreditCardRegisterActivity of Samsung Checkout prior to version 5.1.03.1 allows local attackers to launch arbitrary activity with Samsung Checkout privileges.
CVE-2026-21125 Alta · 8.6 CWE-20 12/10/2026
Improper input validation in parsing sheifd in libsimba.heifdec.media.samsung.so prior to SMR Oct...
Improper input validation in parsing sheifd in libsimba.heifdec.media.samsung.so prior to SMR Oct-2026 Release 1 allows local attackers to potentially execute arbitrary code.
CVE-2026-21124 Alta · 8.6 CWE-20 12/10/2026
Improper input validation in getting cover image in libsimba.heifdec.media.samsung.so prior to...
Improper input validation in getting cover image in libsimba.heifdec.media.samsung.so prior to SMR Oct-2026 Release 1 allows local attackers to potentially execute arbitrary code.
CVE-2026-87765 Média · 5.5 CWE-20 11/10/2026
Improper Input Validation in Kernel APsys DVFS
In the kernel apsys DVFS driver, there is a possible out-of-bounds read or hardware register corruption due to improper input validation of the frequency index. This could lead to system instability or denial of service with local privileges.
CVE-2026-66085 Média · 5.5 CWE-20 11/10/2026
Information Disclosure in Modem
In modem, there is a possible information disclosure due to missing filtering of sensitive data in crash dumps. This could lead to local information disclosure with no additional execution privileges needed.
CVE-2026-107825 Média · 4.0 CWE-20 09/10/2026
OWASP Coraza WAF: ProcessURI silently drops QUERY_STRING and ARGS_GET on URI parse failure — defense-in-depth bypass for non-net/http integrations
OWASP Coraza WAF is a golang modsecurity compatible web application firewall library. From 3.0.0 until 3.8.0, ProcessURI in internal/corazawaf/transaction.go handles a url.ParseRequestURI failure by retaining the raw URI but leaving QUERY_STRING, ARGS_GET, ARGS_GET_NAMES, and the GET-derived portion of ARGS empty. An unauthenticated attacker can place control bytes in a URI passed directly by integrations such as coraza-spoa, coraza-proxy-wasm, custom FFI hosts, or WASM hosts, causing Coraza to omit query parameters that the downstream integration may still process and allowing rules targeting those variables to be bypassed. The bundled coraza/v3/http integration is not affected because Go net/http rejects such malformed request targets before calling Coraza. This issue is fixed in version 3.8.0.
CVE-2026-48484 Média · 6.5 CWE-20 09/10/2026
pyLoad: Lack of Input Size Validation Leads to Denial of Service (DoS) and Process Termination
pyLoad is a free and open-source download manager written in Python. Prior to 0.5.0b3.dev101, the API `rpc` function in `api_blueprint.py` handles `multipart/form-data` uploads by reading the whole content of the uploaded file into memory with `file.read()`. This occurs before the data is sent to the underlying function. Since there is no size limit set at this point, a large file upload can exhaust the server's available memory which led to process termination. Version 0.5.0b3.dev101 contains a patch.
CVE-2026-103413 Alta · 8.8 CWE-20 09/10/2026
Apache Camel Karavan: unvalidated Kubernetes resources applied from a project's kubernetes.yaml
Improper input validation vulnerability in Apache Camel Karavan. When a deployment was started, Karavan unmarshalled a project's `kubernetes.yaml` and applied every resource it contained to the cluster without restricting the resource kinds, without rejecting security-sensitive pod options, and without pinning the target namespace. An authenticated user of any role could therefore have Karavan apply arbitrary Kubernetes resources within the reach of its service account, including pods requesting hostNetwork, hostPID, hostIPC, hostPath volumes, host ports, privileged containers, privilege escalation or added capabilities. This issue affects Apache Camel Karavan: from 4.0.0 before 4.22.1. Users are recommended to upgrade to version 4.22.1, which fixes the issue.
CVE-2026-107734 Alta · 7.1 CWE-20 08/10/2026
SumatraPDF: SyncTeX Argument Injection in Inverse Search Enables Arbitrary Command Execution via External Editors
SumatraPDF is a multi-format reader for Windows. In 3.5.2 and earlier, an attacker-controlled SyncTeX source filename is substituted for the %f placeholder in an external editor command line without safe Windows argument quoting, and the resulting command line is passed to CreateProcessW(). A user with an external editor configured or auto-detected who opens a PDF with a crafted .synctex.gz file and invokes inverse search can inject command-line flags; the resulting impact depends on the target editor interpreting those flags and can include unintended editor actions or code execution through a malicious extension. No broader impact is claimed beyond the advisory-supported conditions. No fixed version is available as of this review.
CVE-2026-107726 Crítica · 9.3 CWE-20 08/10/2026
Hazelcast: Arbitrary member memory access by low-privileged client
Hazelcast is a unified real-time data platform combining stream processing with a fast data store. Prior to 5.4.5, 5.5.10, and 5.6.1, improper validation of data supplied by a malicious client able to connect to a cluster allows arbitrary reads from a cluster member's Java heap, off-heap data, and JVM process address space. The same flaw can crash cluster members and, in some Hazelcast Enterprise Edition configurations, corrupt memory with possible arbitrary code execution. Both slim and full distributions are affected. This issue is fixed in versions 5.4.5, 5.5.10, 5.6.1, and 5.7.0.
CVE-2026-107720 Alta · 7.4 CWE-20 08/10/2026
fast-jwt: createVerifier accepts unsigned JWTs when key is '' or null and algorithms is explicitly set
fast-jwt provides fast JSON Web Token (JWT) implementation. Prior to 6.3.1, fast-jwt createVerifier accepts an unsigned JWT when key is an empty string or null and algorithms is a non-empty allowlist. Falsy synchronous keys bypass prepareKeyOrSecret, allowedAlgorithms remains active, hasKey is false, and the empty signature avoids the verifySignature gate. An attacker can therefore submit a token containing arbitrary claims without possessing a signing key, resulting in authentication or authorization bypass. Claim validators still run, and non-empty keys, an empty key without algorithms, and the async key resolver path do not have this behavior. This issue is fixed in version 6.3.1.
CVE-2026-107717 Média · 6.5 CWE-20 08/10/2026
Banks: User-controlled prompt input can be parsed as privileged chat messages
Banks generates meaningful LLM prompts using a simple template language. Prior to 2.5.0, Banks Prompt.chat_messages() attempts to parse every line of rendered template output as ChatMessage JSON. When an application renders untrusted data and passes the returned ChatMessage objects to an LLM provider, attacker-controlled JSON can cross the prompt boundary and become a system, assistant, or tool message because ChatMessage.role accepts arbitrary strings. This can override application instructions, alter the intended prompt structure, or confuse downstream tool and message handling. This issue is fixed in version 2.5.0.
CVE-2026-88257 Média · 5.3 CWE-20 08/10/2026
beam_mcp: nested tool argument constraints advertised but not enforced
Improper Input Validation vulnerability in BeamMCP.Schema in ScriptKittyOS beam_mcp allows an MCP client to reach a tool's dispatch function with arguments that violate the input schema the server advertised. BeamMCP.Schema.validate/2 checked type, required, additionalProperties, enum and numeric bounds on the top-level arguments object only. Constraints inside nested objects and on array items (items, minItems, maxItems, minLength, maxLength, pattern, nested required, enum and additionalProperties: false) were advertised by tools/list and never checked at tools/call or prompts/get, and keywords outside the enforced subset (oneOf, anyOf, $ref) were advertised and ignored. A host whose dispatch code relies on the schema it declared receives values the schema forbids, such as an out-of-range number or an undeclared key inside a nested object. What the host does with such a value decides the impact. This issue affects beam_mcp: from 0.1.0 before 0.10.1.