Just published.
The newest CVEs, straight from CVE.org and NVD.
-
CVE-2026-115473
Mongoose is an embedded web server and network library. Prior to 7.22, mg_path_is_sane in src/util.c read path.buf[1] without first requiring path.len to be at least two and read s[2] when only two bytes remained. Unauthenticated remote requests reaching mg_http_upload or mg_http_serve_dir with the affected short path shapes could therefore trigger one-byte heap out-of-bounds reads. On bare-metal or RTOS targets without memory protection, the adjacent byte could be exposed or influence whether a path was accepted, and instrumented builds could crash. This issue is fixed in version 7.22.
-
CVE-2026-102895
GPUStack is an open-source GPU cluster manager for AI model serving and GPU instance provisioning. Prior to 2.2.0, get_current_user in gpustack/api/auth.py treated a loopback request source as the built-in administrator when force_auth_localhost was false, which was the default. A remote request could appear to originate from 127.0.0.1 because client_ip_getter accepted a same-host reverse proxy peer or a client-supplied X-GPUStack-Real-IP loopback value. The bypass allowed unauthenticated administrative access, including user and model changes, persistent administrator API-key creation, disclosure of worker and cloud-provider secrets, and code execution through POST /v2/inference-backends/from-yaml. Exploitation was observed in the wild against a 2.1.1 deployment. This issue is fixed in version 2.2.0.
-
CVE-2026-102280
Sandboxie-Plus is an open source sandbox-based isolation software for Windows. Prior to 1.18.6, a process inside a sandbox could send MSGID_FILE_SET_ATTRIBUTES or MSGID_FILE_SET_SHORT_NAME to SbieSvc and cause the SYSTEM service to open and modify an existing file outside the sandbox. The direct primitive changes file attributes, timestamps, writeability, or 8.3 short names; a short-name redirection can additionally contribute to SYSTEM code execution when a privileged process resolves an attacker-controlled short-name path in a location SbieSvc can open. This vulnerability is fixed in 1.18.6.
-
CVE-2026-82418
IBM Watson Speech Services Cartridge 4.0.0 through 5.4.0 could allow a remote attacker to execute arbitrary code due to a use-after-free vulnerability.
-
CVE-2026-17493
IBM Watson Speech Services Cartridge 4.0.0 through 5.4.0 could allow a remote attacker to cause a denial of service due to an out-of-bounds read.
-
CVE-2026-19319
IBM Watson Speech Services Cartridge 4.0.0 through 5.4.0 could allow a remote attacker to obtain sensitive information due to an out-of-bounds read.
-
CVE-2026-16725
IBM Watson Speech Services Cartridge 4.0.0 through 5.4.0 could allow a remote authenticated attacker to read arbitrary files from the local filesystem due to path traversal.
-
CVE-2026-122235
Jan from 0.8.5 before 0.8.8 contains a workspace guard bypass vulnerability in the agent-tools gate.rs that allows attackers to write files outside the workspace via dangling symlinks. Attackers supplying a malicious repository can use prompt injection to make the write or edit tools follow symlinks, creating files like ~/.ssh/authorized_keys on the host.
-
CVE-2026-122140
Trawl 1.5.0 before 1.10.0 contains a server-side request forgery vulnerability that allows unauthenticated attackers to bypass the MCP outbound URL policy using IPv4-mapped IPv6 addresses. Attackers can supply hex-normalized addresses like ::ffff:7f00:1 directly or via redirects to fetch loopback, private, and cloud metadata services and read responses.
-
CVE-2026-122133
Mistral Vibe before 2.25.5 on Windows contains an untrusted search path vulnerability that allows attackers to execute code by placing git.exe in a repository root. Attackers can lure users into opening a crafted repository so SessionLogger and ProjectContextProvider run the malicious git.exe at session start before any trust prompt.
-
CVE-2026-122134
Mistral Vibe before 2.25.7 contains a remote code execution vulnerability that allows attackers to run commands by configuring malicious Git clean or process filter drivers in a repository. When a user opens a crafted repository, ProjectContextProvider._fetch_git_status runs git status at startup, invoking the filters and executing commands with user privileges before any trust prompt.
-
CVE-2026-122135
Mistral Vibe before 3.0.3 contains a remote code execution vulnerability that allows attackers to run commands by abusing repository-configured gpg.program during project context collection. A crafted repository setting gpg.program and log.showSignature with a signed commit makes git log execute the attacker's program at session start, before any trust prompt.
-
CVE-2026-122137
Powerlevel10k before 1.20.19 contains an insecure temporary directory vulnerability in the bundled gitstatus/install script that reuses attacker-creatable directories when mktemp fails. Local attackers can pre-create the predictable gitstatus-install.tmp directory to replace the extracted gitstatusd binary, which then executes with the victim's privileges.
-
CVE-2026-122127
CESNET ipfixcol2 2.7.1 before 2.9.0 contains a heap-based buffer overflow in Lz4Decoder.cpp that allows unauthenticated network peers to write past heap buffers. Attackers can declare a small ring buffer_size followed by a large decompressed_size to write up to 65535 controlled bytes and crash the collector.
-
CVE-2026-122129
CESNET ipfixcol2 2.7.1 before 2.9.0 contains a memory allocation vulnerability in Lz4Decoder::read_start_header() that allows unauthenticated attackers to exhaust memory by declaring unbounded ring buffer sizes. Attackers can open TCP connections sending LZ4 start headers with very large buffer sizes, forcing up to 4 GiB allocations per connection and denying collector service.
-
CVE-2026-122130
CESNET ipfixcol2 2.3.0 before 2.9.0 contains an integer overflow vulnerability in the Forwarder output plugin that allows unauthenticated IPFIX exporters to overflow a heap buffer by wrapping uint16_t message length counters. Attackers can send near-65535-byte data sets and large templates to a collector forwarding over TCP to crash it and perform 2-byte writes to attacker-influenced addresses.
-
CVE-2026-122013
A vulnerability was detected in Comfast CF-WR632AX 2.6.0.1. Affected by this vulnerability is an unknown functionality of the file /usr/bin/webmgnt of the component Multipart Parser. Performing a manipulation of the argument filename results in path traversal. It is possible to initiate the attack remotely. The exploit is now public and may be used. The vendor was contacted early about this disclosure but did not respond in any way.
-
CVE-2026-121968
Netrek-client-cow through 3.3.3 contains an out-of-bounds read vulnerability in short.c handleSMessage() that allows malicious servers to crash clients via an unchecked m_recpt index. Attackers controlling a Netrek server or intercepting the plaintext TCP connection can send SP_S_MESSAGE packets with m_recpt of 36 or more, reading past players[] and shipnos to deny service.
-
CVE-2026-121969
netrek-client-cow through 3.3.3 contains a stack-based buffer overflow vulnerability in handleSMessage() in short.c that allows malicious servers to overflow a 100-byte stack buffer by sending oversized SP_S_MESSAGE packets. A malicious Netrek server or man-in-the-middle on the plaintext TCP connection can send overlong or unterminated message bodies to write server-controlled bytes onto the stack and crash the client.
-
CVE-2026-115758
SumatraPDF is a multi-format reader for Windows. In version 3.6.1 and earlier, HttpGet() in src/base/Http_win.cpp unconditionally passes INTERNET_FLAG_IGNORE_CERT_CN_INVALID to WinINet, disabling hostname matching for HTTPS update metadata and crash-symbol requests while retaining certificate-chain validation. A network-positioned attacker with a Windows-trusted certificate for an attacker-controlled domain can impersonate the project's update host, inject promo text and a clickable URL before update metadata validation, and provide an attacker-controlled archive to ExtractSymbols(). Installer replacement is not directly possible because HttpGetToFile() validates TLS and MaybeStartUpdateDownload() restricts the download host. No fixed version is available as of this review.
-
CVE-2026-10859
IBM MQ 9.2.0.0 through 9.2.0.44 LTS, 9.3.0.0 through 9.3.0.42 LTS, 9.4.0.0 through 9.4.0.26 LTS, 9.4.0.0 through 9.4.5.1 CD, and 10.0.0.0 through 10.0.0.5 LTS and CD could allow an authenticated attacker to execute arbitrary code on consumer applications due to insecure deserialization of ObjectMessage payloads.
-
CVE-2026-110281
IBM Db2 Genius Hub 1.1.2, 1.1.3, 1.1.4, 1.1.5 improperly validates TLS certificate validation, which could allow an attacker to obtain sensitive information using man in the middle technique.
-
CVE-2026-110283
IBM Db2 Genius Hub 1.1.2, 1.1.3, 1.1.4, 1.1.5 could allow a remote authenticated attacker to cause a denial of service due to improper validation of user-influenced resource identifiers.
-
CVE-2026-108838
IBM Db2 Genius Hub 1.1.2, 1.1.3, 1.1.4, 1.1.5 could allow a local attacker to cause a denial of service due to uncontrolled resource consumption.
-
CVE-2026-108910
IBM Db2 Genius Hub 1.1.2, 1.1.3, 1.1.4, 1.1.5 could allow a remote authenticated attacker to obtain sensitive information due to missing required cryptographic step.
-
CVE-2026-104203
Traccar is an open source GPS tracking system. From 5.1 until 6.15.0, WebModule.java's MediaFilter Guice mapping and WebServer.java's Jetty ResourceServlet assign different routing meanings to path parameters at /api/media/*. An unauthenticated attacker who knows or guesses a device unique ID and media filename can insert a semicolon at the Servlet boundary, bypass MediaFilter's session and device authorization checks, and retrieve the protected media because Jetty still dispatches the noncanonical path to ResourceServlet. The demonstrated impact is read-only disclosure of a protected device media file; depending on the deployment and stored media, this could expose photographs, audio, video, or other uploaded files. The advisory does not establish directory listing, traversal, modification, account takeover, denial of service, or code execution. Reverse proxies that reject or rewrite semicolon path parameters may block the request. This issue is fixed in version 6.15.0.
-
CVE-2026-113374
Sonatype Nexus Repository 3 does not require realm administration permission when a user creates a Rut Auth capability. A user who holds only the capability creation permission (nexus:capabilities:create) can create this capability, which turns on header-based authentication for the instance. A remote client can then authenticate as an existing user, including an administrator, without credentials.
-
CVE-2026-113371
Sonatype Nexus Repository 3 returns the signing passphrase of an APT or YUM repository, through an internal REST endpoint, to a user account that holds only the read privilege for administering that repository. The account does not need the privilege to edit the repository.
-
CVE-2026-113372
Sonatype Nexus Repository 3 returns the private signing key of a YUM proxy or group repository to an account that holds only the read privilege for administering that repository. The account does not need the privilege to edit the repository. A party that holds the key can sign repository metadata as that repository.
-
CVE-2026-113373
Sonatype Nexus Repository 3 Pro does not check the local account status when a user signs in through a configured SAML or OAuth2/OpenID Connect identity provider. A user whose account was disabled or locked by an administrator in Nexus Repository can sign in again with valid identity provider credentials for the same account, and then receives a new session with the roles of that account. The attacker needs no Nexus Repository session and no Nexus Repository privilege. The attacker needs a named account that holds valid credentials at the identity provider. The anonymous user is not part of this attack path.
-
CVE-2026-82033
DeepSeek Harness prior to 0.1.2-alpha.2, and specifically commit 3e24087, contains a missing authentication vulnerability in its web profile HTTP and WebSocket RPC control plane, allowing any local process, including those running as a different OS user, to reach unauthenticated session methods such as session.create and session.prompt. Attackers can invoke session.prompt to trigger the agent's bash and filesystem tools, achieve arbitrary command execution as the victim user using their configured model credentials, intercept live session traffic and logs over the event mux stream, and bypass human tool-approval checks via POST /api/respond by matching only rpcId, approvalId, and sessionId values broadcast over the ungated stream.
-
CVE-2026-122120
In Yealink Plug-in Softwares before 2.35.219.0, there is an unauthorized file write vulnerability under specific scenarios, which may be exploited by attackers to achieve remote code execution.
-
CVE-2026-122124
On Yealink UVC40 E2 devices before 319.410.0.71, an attacker with short-term physical access can steal Wi‑Fi credentials via the USB interface. Sensitive data within the firmware configuration can be abused to gain root privileges on the device. Within Wi‑Fi coverage, the attacker may perform audio and video eavesdropping with no LED alert from the device.
-
CVE-2026-122148
In Yealink MeetingBoard Pro before 331.320.0.64, the Web management module has a front-end validation bypass vulnerability. An attacker with administrator privileges can modify the page DOM to bypass front-end UI control logic and invoke screenshot-related interfaces, triggering the screenshot function by circumventing front-end interface restrictions. Exploitation can occur only when the attacker has a valid administrator account within the local network.
-
CVE-2026-113780
ESF-IDF is the Espressif Internet of Things (IOT) Development Framework. From 5.5.5 until 5.5.6, 6.0.4, and 6.1.1, the esp_tee trap entry in esp_tee_vectors_plic.S and esp_tee_vectors_clic.S saves a caller-controlled register frame through the REE-supplied stack pointer before switching to the trusted TEE stack or validating the pointer. A local unprivileged REE application can direct this machine-mode write into TEE-exclusive memory on ESP32-C5, ESP32-C6, ESP32-C61, and ESP32-H2 devices, breaking the REE/TEE isolation boundary and potentially corrupting TEE control-flow data for code execution, denial of service, or system crashes. Both secure-service ecall and interrupt entry paths contain the vulnerable ordering. This issue is fixed in version 5.5.6 and is scheduled to be fixed in versions 6.0.4 and 6.1.1.
-
CVE-2026-113782
ESP-IDF is the Espressif Internet of Things (IOT) Development Framework. From 5.2.0 until 5.2.8, 5.3.6, 5.4.5, 5.5.5, and 6.0.3, the OpenThread Border Router Neighbor Discovery packet handler trusts an attacker-controlled ICMPv6 option length while copying into a fixed stack buffer. A host on the same infrastructure link as the Border Router can send a crafted ICMPv6 Router Advertisement to its infrastructure interface whose oversized length causes an out-of-bounds write into adjacent stack memory. Firmware is affected only when CONFIG_OPENTHREAD_BORDER_ROUTER=y; Thread end devices, routers, and RCP firmware without the Border Router feature are unaffected. Exploitation requires no authentication or user interaction and can crash or reboot the device or corrupt adjacent stack state. This issue is fixed in versions 5.2.8, 5.3.6, 5.5.5, and 6.0.3 and is scheduled to be fixed in version 5.4.5.
-
CVE-2026-113783
ESP-IDF is the Espressif Internet of Things (IOT) Development Framework. From 5.2.3 until 5.3.7, 5.4.5, 5.5.6, and 6.0.3, the lwIP DNS resolver's cached-lookup path in dns_lookup() in dns.c copies every cached address into a caller-provided single ip_addr_t output buffer without a length parameter or bounds check. When CONFIG_LWIP_DNS_MAX_HOST_IP is greater than 1 and a cached hostname has multiple A or AAAA records, a subsequent lookup through dns_gethostbyname() or dns_gethostbyname_addrtype() writes beyond the single-element buffer into adjacent stack memory. A network-positioned attacker who influences DNS responses can cause a reboot through stack-canary corruption or, without stack protection, overwrite stack variables with attacker-controlled address bytes and potentially redirect control flow. This issue is fixed in versions 5.5.6 and 6.0.3 and is scheduled to be fixed in versions 5.3.7 and 5.4.5. No fixed release is available for the end-of-life 5.2 line.
-
CVE-2026-122116
Certain Yealink terminal devices before 2026-04-21 contain an improper access control vulnerability in the Wi-Fi 802.1X PEAP network configuration function. This flaw exposes deployed 802.1X network access authentication credentials to unauthorized users. The latest Yealink firmware has addressed this vulnerability by restricting 802.1X configuration modification privileges to administrator accounts only by default.
-
CVE-2026-122118
In certain Yealink products before 2026-04-15, when the Wi-Fi AP function is enabled, a default 8-digit numeric PSK with insufficient strength is used. Attackers can exploit this weakness via brute-force attacks to gain unauthorized network access and intercept Wi-Fi traffic.
-
CVE-2026-76506
A vulnerability in the web-based management interface of Cisco UCS Central Software could allow an authenticated, remote attacker to achieve remote code execution or conduct information disclosure attacks on an affected device. This vulnerability is due to insufficient validation of user-supplied input in an HTTP request. An attacker could exploit this vulnerability by sending a crafted HTTP request to an affected endpoint on an affected device. A successful exploit could allow the attacker to execute arbitrary commands, write arbitrary files to the filesystem, and obtain sensitive information. To exploit this vulnerability, the attacker must have valid user credentials on the affected device. Cisco has released software updates that address this vulnerability There are no workarounds that address this vulnerability.
-
CVE-2026-76419
A vulnerability in the web-based management interface of Cisco Unified Communications Manager (Unified CM) and Cisco Unified Communications Manager Session Management Edition (Unified CM SME) could allow an unauthenticated, remote attacker to conduct a cross-site scripting (XSS) attack against a user of the interface. This vulnerability exists because the web-based management interface does not properly validate user-supplied input. An attacker could exploit this vulnerability by persuading a user of the interface to click a crafted link. A successful exploit could allow the attacker to execute arbitrary script code in the context of the affected interface or access sensitive, browser-based information.
-
CVE-2026-121926
A vulnerability was detected in yezhihao jt808-server up to a07f68c466b560b97ef41edb8a4966469abdf83e. Affected is the function WebMvcConfig.addInterceptors of the component Web Management API. The manipulation results in missing authentication. The attack may be launched remotely. The exploit is now public and may be used. This product operates on a rolling release basis, ensuring continuous delivery. Consequently, there are no version details for either affected or updated releases. The vendor was contacted early about this disclosure but did not respond in any way.
-
CVE-2026-20241
A vulnerability in the authorization mechanism of Cisco Identity Services Engine (ISE) and Cisco ISE Passive Identity Connector (ISE-PIC) could allow an unauthenticated, remote attacker to obtain an endpoint identity certificate. This vulnerability is due to improper authorization checks. An attacker could exploit this vulnerability by sending a crafted request to an affected device. A successful exploit could allow the attacker to bypass authorization to obtain an endpoint certificate.
-
CVE-2026-92704
SharpEmu is an experimental PlayStation 5 emulator for Windows and Linux. In version 0.0.3-release.3 and earlier, the repository-provided debugger frontend tooling is vulnerable while running. This tooling is not included in the release archives. The frontend exposes an HTTP API on 127.0.0.1:8765 without authentication or CSRF token. Its do_POST handler in tools/SharpEmu.DebuggerFrontend/debugger_frontend.py does not validate the Origin, Referer, Sec-Fetch-Site, or Host headers. The _read_json_body() function also accepts JSON sent with browser-simple content types instead of requiring application/json . A malicious page that can reach the loopback listener can send cross-origin requests to /api/launch , /api/command , and /api/connect . Chromium may require Local Network Access permission for pages served from public origins. Local-origin pages and other delivery paths are not consistently subject to this restriction. The /api/launch endpoint accepts an attacker-selected emulatorPath if it points to an existing file on the victim’s host. On POSIX systems, that file must also be executable. The endpoint additionally requires an existing ebootPath . The /api/command endpoint exposes debugger operations, including write-memory , set-register , and continue . The /api/connect endpoint can initiate outbound TCP connections. These operations can let an attacker execute an existing binary with the victim’s privileges, control the attached emulated process, or reach network services accessible from the host. DNS rebinding can additionally let the attacker read API responses interactively. This issue is fixed in version 0.0.3-release.3-patch.1
-
CVE-2026-91163
Warpgate is an open source SSH, HTTPS, and MySQL bastion host for Linux. Prior to 0.29.0, the Kubernetes protocol listener uses AcceptAnyClientCert in warpgate-protocol-kubernetes/src/server/client_certs.rs to accept certificates at the TLS layer and defer validation to warpgate-protocol-kubernetes/src/server/auth.rs, where authentication and lookup failures are rendered verbatim in an HTTP 500 response body. The Kubernetes application in warpgate-protocol-kubernetes/src/server/mod.rs lacks the outer request-logging wrapper used by the HTTP protocol, and failures in warpgate-protocol-kubernetes/src/server/handlers.rs can short-circuit before its inner log point. An unauthenticated network client can therefore induce failures and receive ASN.1/DER parser diagnostics or sea_orm database errors without the attempt being recorded in the gateway log. The disclosed information is limited to internal diagnostics, and no user-data, integrity, or availability impact is established. This issue is fixed in version 0.29.0.
-
CVE-2026-119746
ProSoft Technology ICX35 devices (X35-HWC-A and ICX35-HWC-E) contain a login lockout bypass in the integrated module web page. The account lockout mechanism that restricts repeated failed authentication attempts is enforced in client-side JavaScript, so a remote attacker can reset the lockout and continue submitting password guesses by deleting the relevant browser cookies, defeating the brute-force protection. This vulnerability has been addressed in version 1.14.158.
-
CVE-2026-39430
On HP Poly EagleEye Cube USB 1.3.1-001267 devices, the administrator authentication flow is broken: session management and access controls are not enforced device-side. Thus, an unauthenticated user can change administrator-level settings (without logging in) via a modified client that bypasses the login step or via a custom client to make the underlying administrator API calls. This allows unauthenticated users to perform administrative actions too such as exporting configuration files that include the internal device credentials for the administrator user and other network systems.
-
CVE-2026-39431
On HP Poly EagleEye Cube USB 1.3.1-001267 devices, one of the API functions available on a management service accessible via USB allows command injection. This allows attackers (without administrator-level access) to execute arbitrary commands on the device as root. This is due to insufficient input validation that allows an attacker to control part of the CSR certificate generation process. This provides full root access that permits full control over device operations, unauthorized access to data, and the ability to place a persistent backdoor on the device (which could operate via the Ethernet port).
-
CVE-2026-100407
Git for Windows is the Windows port of Git. From 2.44.0.windows.1 until 2.55.0.windows.3, the git-credential-wincred store operation in contrib/credential/wincred/git-credential-wincred.c uses _scwprintf() to size a formatted wide-character credential buffer without space for the terminating NUL when both password and oauth_refresh_token are present. The operation also passes a byte count to _snwprintf_s() where a wide-character count is required, allowing the terminator to be written one WCHAR past the heap allocation before CredWriteW() persists the credential. The resulting fixed two-byte NUL overwrite can terminate the helper with Windows heap-corruption status and can cause git credential approve to report success even though the OAuth credential was not stored. This issue is fixed in version 2.55.0.windows.3.
-
CVE-2026-100408
Git for Windows is the Windows port of Git. From 2.43.0.windows.1 until 2.55.0.windows.3, when a matching git:* credential exists in Windows Credential Manager and a git-credential-wincred erase input includes a password field, the operation in contrib/credential/wincred/git-credential-wincred.c allocates a buffer of exactly CredentialBlobSize bytes in match_cred_password() and then uses wcsncpy_s() to copy the corresponding number of wide characters plus a terminating NUL. The destination capacity is also supplied in bytes instead of WCHAR elements, allowing a fixed two-byte NUL write past the allocation before the password comparison determines whether CredDeleteW() should remove the credential; erase without a password field does not enter this comparison path. The helper can terminate with Windows heap-corruption status, and git credential reject can appear successful while the rejected credential remains stored. This issue is fixed in version 2.55.0.windows.3.
The newest published CVE records from CVE.org and NVD, refreshed every few minutes, with severity, exploit odds and fix status.