| CVE-2018-8897 | A statement in the System Programming Guide of the Intel 64 and IA-32 Architectures Software Developer's Manual (SDM) was mishandled in the development of some or all operating-system kernels, resulting in unexpected behavior for #DB exceptions that are deferred by MOV SS or POP SS, as demonstrated by (for example) privilege escalation in Windows, macOS, some Xen configurations, or FreeBSD, or a Linux kernel crash. The MOV to SS and POP SS instructions inhibit interrupts (including NMIs), data breakpoints, and single step trap exceptions until the instruction boundary following the next instruction (SDM Vol. 3A; section 6.8.3). (The inhibited data breakpoints are those on memory accessed by the MOV to SS or POP to SS instruction itself.) Note that debug exceptions are not inhibited by the interrupt enable (EFLAGS.IF) system flag (SDM Vol. 3A; section 2.3). If the instruction following the MOV to SS or POP to SS instruction is an instruction like SYSCALL, SYSENTER, INT 3, etc. that transfers control to the operating system at CPL < 3, the debug exception is delivered after the transfer to CPL < 3 is complete. OS kernels may not expect this order of events and may therefore experience unexpected behavior when it occurs. | high | |
| CVE-2024-45599 | Cursor is an artificial intelligence code editor. Prior to version 0.41.0, if a user on macOS has granted Cursor access to the camera or microphone, any program that is run on the machine is able to access the camera or the microphone without explicitly being granted access, through a DyLib Injection using DYLD_INSERT_LIBRARIES environment variable. The usage of `com.apple.security.cs.allow-dyld-environment-variables` and `com.apple.security.cs.disable-library-validation` allows an external dynamic library to be injected into the application using DYLD_INSERT_LIBRARIES environment variable. Moreover, the entitlement `com.apple.security.device.camera` allows the application to use the host camera and `com.apple.security.device.audio-input` allows the application to use the microphone. This means that untrusted code that is executed on the user's machine can access the camera or the microphone, if the user has already given permission for Cursor to do so. In version 0.41.0, the entitlements have been split by process: the main process gets the camera and microphone entitlements, but not the DyLib entitlements, whereas the extension host process gets the DyLib entitlements but not the camera or microphone entitlements. As a workaround, do not explicitly give Cursor the permission to access the camera or microphone if untrusted users can run arbitrary commands on the affected machine. | low | |
| CVE-2026-26327 | OpenClaw is a personal AI assistant. Discovery beacons (Bonjour/mDNS and DNS-SD) include TXT records such as `lanHost`, `tailnetDns`, `gatewayPort`, and `gatewayTlsSha256`. TXT records are unauthenticated. Prior to version 2026.2.14, some clients treated TXT values as authoritative routing/pinning inputs. iOS and macOS used TXT-provided host hints (`lanHost`/`tailnetDns`) and ports (`gatewayPort`) to build the connection URL. iOS and Android allowed the discovery-provided TLS fingerprint (`gatewayTlsSha256`) to override a previously stored TLS pin. On a shared/untrusted LAN, an attacker could advertise a rogue `_openclaw-gw._tcp` service. This could cause a client to connect to an attacker-controlled endpoint and/or accept an attacker certificate, potentially exfiltrating Gateway credentials (`auth.token` / `auth.password`) during connection. As of time of publication, the iOS and Android apps are alpha/not broadly shipped (no public App Store / Play Store release). Practical impact is primarily limited to developers/testers running those builds, plus any other shipped clients relying on discovery on a shared/untrusted LAN. Version 2026.2.14 fixes the issue. Clients now prefer the resolved service endpoint (SRV + A/AAAA) over TXT-provided routing hints. Discovery-provided fingerprints no longer override stored TLS pins. In iOS/Android, first-time TLS pins require explicit user confirmation (fingerprint shown; no silent TOFU) and discovery-based direct connects are TLS-only. In Android, hostname verification is no longer globally disabled (only bypassed when pinning). | high | |
| CVE-2025-68146 | filelock is a platform-independent file lock for Python. In versions prior to 3.20.1, a Time-of-Check-Time-of-Use (TOCTOU) race condition allows local attackers to corrupt or truncate arbitrary user files through symlink attacks. The vulnerability exists in both Unix and Windows lock file creation where filelock checks if a file exists before opening it with O_TRUNC. An attacker can create a symlink pointing to a victim file in the time gap between the check and open, causing os.open() to follow the symlink and truncate the target file. All users of filelock on Unix, Linux, macOS, and Windows systems are impacted. The vulnerability cascades to dependent libraries. The attack requires local filesystem access and ability to create symlinks (standard user permissions on Unix; Developer Mode on Windows 10+). Exploitation succeeds within 1-3 attempts when lock file paths are predictable. The issue is fixed in version 3.20.1. If immediate upgrade is not possible, use SoftFileLock instead of UnixFileLock/WindowsFileLock (note: different locking semantics, may not be suitable for all use cases); ensure lock file directories have restrictive permissions (chmod 0700) to prevent untrusted users from creating symlinks; and/or monitor lock file directories for suspicious symlinks before running trusted applications. These workarounds provide only partial mitigation. The race condition remains exploitable. Upgrading to version 3.20.1 is strongly recommended. | medium | |
| CVE-2023-4759 | Arbitrary File Overwrite in Eclipse JGit <= 6.6.0 In Eclipse JGit, all versions <= 6.6.0.202305301015-r, a symbolic link present in a specially crafted git repository can be used to write a file to locations outside the working tree when this repository is cloned with JGit to a case-insensitive filesystem, or when a checkout from a clone of such a repository is performed on a case-insensitive filesystem. This can happen on checkout (DirCacheCheckout), merge (ResolveMerger via its WorkingTreeUpdater), pull (PullCommand using merge), and when applying a patch (PatchApplier). This can be exploited for remote code execution (RCE), for instance if the file written outside the working tree is a git filter that gets executed on a subsequent git command. The issue occurs only on case-insensitive filesystems, like the default filesystems on Windows and macOS. The user performing the clone or checkout must have the rights to create symbolic links for the problem to occur, and symbolic links must be enabled in the git configuration. Setting git configuration option core.symlinks = false before checking out avoids the problem. The issue was fixed in Eclipse JGit version 6.6.1.202309021850-r and 6.7.0.202309050840-r, available via Maven Central https://repo1.maven.org/maven2/org/eclipse/jgit/ and repo.eclipse.org https://repo.eclipse.org/content/repositories/jgit-releases/ . A backport is available in 5.13.3 starting from 5.13.3.202401111512-r. The JGit maintainers would like to thank RyotaK for finding and reporting this issue. | high | |
| CVE-2021-39134 | `@npmcli/arborist`, the library that calculates dependency trees and manages the `node_modules` folder hierarchy for the npm command line interface, aims to guarantee that package dependency contracts will be met, and the extraction of package contents will always be performed into the expected folder. This is, in part, accomplished by resolving dependency specifiers defined in `package.json` manifests for dependencies with a specific name, and nesting folders to resolve conflicting dependencies. When multiple dependencies differ only in the case of their name, Arborist's internal data structure saw them as separate items that could coexist within the same level in the `node_modules` hierarchy. However, on case-insensitive file systems (such as macOS and Windows), this is not the case. Combined with a symlink dependency such as `file:/some/path`, this allowed an attacker to create a situation in which arbitrary contents could be written to any location on the filesystem. For example, a package `pwn-a` could define a dependency in their `package.json` file such as `"foo": "file:/some/path"`. Another package, `pwn-b` could define a dependency such as `FOO: "file:foo.tgz"`. On case-insensitive file systems, if `pwn-a` was installed, and then `pwn-b` was installed afterwards, the contents of `foo.tgz` would be written to `/some/path`, and any existing contents of `/some/path` would be removed. Anyone using npm v7.20.6 or earlier on a case-insensitive filesystem is potentially affected. This is patched in @npmcli/arborist 2.8.2 which is included in npm v7.20.7 and above. | high | |
| CVE-2025-48064 | GitHub Desktop is an open-source, Electron-based GitHub app designed for git development. Prior to version 3.4.20-beta3, an attacker convincing a user to view a file in a commit of their making in the history view can cause information disclosure by means of Git attempting to access a network share. This affects GitHub Desktop users on Windows that view malicious commits in the history view. macOS users are not affected. When viewing a file diff in the history view GitHub Desktop will call `git log` or `git diff` with the object id (SHA) of the commit, the name of the file, and the old name of the file if the file has been renamed. As a security precaution Git will attempt to fully resolve the old and new path via `realpath`, traversing symlinks, to ensure that the resolved paths reside within the repository working directory. This can lead to Git attempting to access a path that resides on a network share (UNC path) and in doing so Windows will attempt to perform NTLM authentication which passes information such as the computer name, the currently signed in (Windows) user name, and an NTLM hash. GitHub Desktop 3.4.20 and later fix this vulnerability. The beta channel includes the fix in 3.4.20-beta3. As a workaround to use until upgrading is possible, only browse commits in the history view that comes from trusted sources. | low | |
| CVE-2024-50338 | Git Credential Manager (GCM) is a secure Git credential helper built on .NET that runs on Windows, macOS, and Linux. The Git credential protocol is text-based over standard input/output, and consists of a series of lines of key-value pairs in the format `key=value`. Git's documentation restricts the use of the NUL (`\0`) character and newlines to form part of the keys or values. When Git reads from standard input, it considers both LF and CRLF as newline characters for the credential protocol by virtue of calling `strbuf_getline` that calls to `strbuf_getdelim_strip_crlf`. Git also validates that a newline is not present in the value by checking for the presence of the line-feed character (LF, `\n`), and errors if this is the case. This captures both LF and CRLF-type newlines. Git Credential Manager uses the .NET standard library `StreamReader` class to read the standard input stream line-by-line and parse the `key=value` credential protocol format. The implementation of the `ReadLineAsync` method considers LF, CRLF, and CR as valid line endings. This is means that .NET considers a single CR as a valid newline character, whereas Git does not. This mismatch of newline treatment between Git and GCM means that an attacker can craft a malicious remote URL. When a user clones or otherwise interacts with a malicious repository that requires authentication, the attacker can capture credentials for another Git remote. The attack is also heightened when cloning from repositories with submodules when using the `--recursive` clone option as the user is not able to inspect the submodule remote URLs beforehand. This issue has been patched in version 2.6.1 and all users are advised to upgrade. Users unable to upgrade should only interact with trusted remote repositories, and not clone with `--recursive` to allow inspection of any submodule URLs before cloning those submodules. | high | |
| CVE-2025-64711 | PrivateBin is an online pastebin where the server has zero knowledge of pasted data. Starting in version 1.7.7 and prior to version 2.0.3, dragging a file whose filename contains HTML is reflected verbatim into the page via the drag-and-drop helper, so any user who drops a crafted file on PrivateBin will execute arbitrary JavaScript within their own session (self-XSS). This allows an attacker who can entice a victim to drag or otherwise attach such a file to exfiltrate plaintext, encryption keys, or stored pastes before they are encrypted or sent. Certain conditions must exist for the vulnerability to be exploitable. Only macOS or Linux users are affected, due to the way the `>` character is treated in a file name on Windows. The PrivateBin instance needs to have file upload enabled. An attacker needs to have access to the local file system or somehow convince the user to create (or download) a malicious file (name). An attacker needs to convince the user to attach that malicious file to PrivateBin. Any Mac / Linux user who can be tricked into dragging a maliciously named file into the editor is impacted; code runs in the origin of the PrivateBin instance they are using. Attackers can steal plaintext, passphrases, or manipulate the UI before data is encrypted, defeating the zero-knowledge guarantees for that victim session, assuming counter-measures like Content-Security-Policy (CSP) have been disabled. If CSP is not disabled, HTML injection attacks may be possible - like redirecting to a foreign website, phishing etc. As the whole exploit needs to be included in the file name of the attached file and only affects the local session of the user (aka it is neither persistent nor remotely executable) and that user needs to interact and actively attach that file to the paste, the impact is considered to be practically low. Version 2.0.3 patches the issue. | medium | |
| CVE-2025-55177 | Incomplete authorization of linked device synchronization messages in WhatsApp for iOS prior to v2.25.21.73, WhatsApp Business for iOS v2.25.21.78, and WhatsApp for Mac v2.25.21.78 could have allowed an unrelated user to trigger processing of content from an arbitrary URL on a target’s device. We assess that this vulnerability, in combination with an OS-level vulnerability on Apple platforms (CVE-2025-43300), may have been exploited in a sophisticated attack against specific targeted users. | medium | |
| CVE-2025-27148 | Gradle is a build automation tool, and its native-platform tool provides Java bindings for native APIs. On Unix-like systems, the system temporary directory can be created with open permissions that allow multiple users to create and delete files within it. This library initialization could be vulnerable to a local privilege escalation from an attacker quickly deleting and recreating files in the system temporary directory. Gradle builds that rely on versions of net.rubygrapefruit:native-platform prior to 0.22-milestone-28 could be vulnerable to a local privilege escalation from an attacker quickly deleting and recreating files in the system temporary directory. In net.rubygrapefruit:native-platform prior to version 0.22-milestone-28, if the `Native.get(Class<>)` method was called, without calling `Native.init(File)` first, with a non-`null` argument used as working file path, then the library would initialize itself using the system temporary directory and NativeLibraryLocator.java lines 68 through 78. Version 0.22-milestone-28 has been released with changes that fix the problem. Initialization is now mandatory and no longer uses the system temporary directory, unless such a path is passed for initialization. The only workaround for affected versions is to make sure to do a proper initialization, using a location that is safe. Gradle 8.12, only that exact version, had codepaths where the initialization of the underlying native integration library took a default path, relying on copying the binaries to the system temporary directory. Any execution of Gradle exposed this exploit. Users of Windows or modern versions of macOS are not vulnerable, nor are users of a Unix-like operating system with the "sticky" bit set or `noexec` on their system temporary directory vulnerable. This problem was fixed in Gradle 8.12.1. Gradle 8.13 release also upgrades to a version of the native library that no longer has that bug. Some workarounds are available. On Unix-like operating systems, ensure that the "sticky" bit is set. This only allows the original user (or root) to delete a file. Mounting `/tmp` as `noexec` will prevent Gradle 8.12 from starting. Those who are are unable to change the permissions of the system temporary directory can move the Java temporary directory by setting the System Property java.io.tmpdir. The new path needs to limit permissions to the build user only. | high | |
| CVE-2025-14174 | Out of bounds memory access in ANGLE in Google Chrome on Mac prior to 143.0.7499.110 allowed a remote attacker to perform out of bounds memory access via a crafted HTML page. (Chromium security severity: High) | high | |
