CVE Vulnerability Catalog

Multiple TP-Link Kasa smart home devices contain insufficient cryptographic protections in the local device communication protocol. An adjacent network attacker may intercept, replay or forge locally exchanged control messages, potentially resulting in unauthorized device control. Successful exploitation could allow an attacker to manipulate the operational state of an affected device, resulting in unauthorized state changes, disruption of normal device functionality or a denial-of-service condition.

whichllm before 0.5.16 contains a code injection vulnerability in the run and snippet commands that allows a remote attacker who controls a HuggingFace repository to achieve arbitrary code execution by crafting a malicious GGUF filename containing double quotes or other special characters. The script generation function in cli.py interpolates HuggingFace-derived values, including GGUF variant filenames from the Hub API siblings rfilename field, directly into Python source code without escaping, allowing the crafted filename to break out of the generated string literal and execute injected code on the user's machine before any model download occurs.

An application using Spring Security's WebAuthn support may be vulnerable to user verification bypass when using a distributed HTTP session store. Affected versions: Spring Security 7.1.0, 7.0.0-7.0.6, 6.5.0-6.5.11, 6.4.0-6.4.18.

The base directory (spring.cloud.config.server.svn.basedir) used by the Spring Cloud Config Server to clone SVN repositories to is susceptible to time-of-check-time-of-use (TOCTOU) attacks. Affected versions: Spring Cloud Config 5.0.0-5.0.4, 4.3.0-4.3.4, 4.0.0-4.2.8, 3.1.14 and earlier.

Cross-Site Request Forgery (CSRF) vulnerability in Lime Technology, Inc.'s Unraid OS version 6.12.14 and earlier allows remote attackers to escalate privileges via the Unraid authentication cookie's lax same-site policy.

CTFd v3.7.6 was discovered to be vulnerable to a man-in-the-middle attack.

Winter CMS is a content management system built on the Laravel PHP framework. In versions prior to 1.2.13, the backend did not validate the handler name submitted through the form postback _handler POST field, allowing an authenticated backend user to invoke arbitrary controller methods, including protected, private, and action-prefixed ones. While AJAX requests validate that handler names match the on[A-Z][\w+]* pattern, the postback path passed the submitted _handler value straight to the handler dispatcher with no such check, so any controller that exposes a public action or conditionally relaxes its $requiredPermissions check could be reached, bypassing the roles and permissions system. The built-in Users controller was affected because it set $requiredPermissions to null for the myaccount action, letting any authenticated backend user invoke user-management methods such as update_onDelete and update_onManualPasswordReset without holding the backend.manage_users permission. This issue is fixed in version 1.2.13.

Winter is a free, open-source content management system (CMS) based on the Laravel PHP framework. From 1.2.10 through 1.2.12, authenticated backend users with the backend.manage_editor permission can store custom Markup Styles that are compiled by the LESS parser and rendered without sanitization on every backend page, allowing stored cross-site scripting. This issue is fixed in version 1.2.13.

Winter is a free, open-source content management system (CMS) based on the Laravel PHP framework. Prior to 1.2.13, custom CSS supplied through the Brand Settings Styles field by a backend user with the backend.manage_branding permission is compiled by the LESS parser and rendered without sanitization on every backend page, allowing stored cross-site scripting against backend users. This issue is fixed in version 1.2.13.

An issue was discovered in LibreNMS 1.65. A remote authenticated attacker with normal privileges can extract all the information from the LibreNMS database via a SQL injection in the address parameter in the /ajax_table.php API endpoint.

An issue was discovered in LibreNMS 1.65. A remote authenticated attacker with normal privileges can extract all the information from the LibreNMS database via a SQL injection in the sort parameter in the /ajax_table.php API endpoint. This affects many files in includes/html/table.

An issue was discovered in LibreNMS 1.65. A remote authenticated attacker with normal privileges can execute arbitrary shell commands through a command injection in the /graph.php API endpoint.

Stalwart Mail Server does not compare an OAuth redirect target against any registered destination in its default configuration. The validation routine returns success immediately when client-authentication is disabled, which is the case in shipped settings. The redirect value is stored with the authorization code, and the login page sends the browser to it with the code, allowing an attacker to capture the code and exchange it for tokens to read the victim's mail.

Midday allows any team member to delete the team. The delete procedure authorizes the caller with a helper that returns true for every membership row regardless of role. Neighboring procedures check the role and refuse non-owners, but deletion lacks this check. Any invitee can remove the team and all scoped records, and the cleanup job uses stored bank-connection tokens against connected providers.

IDURAR ERP CRM changes the password of whichever account a request names rather than the account making the request. The update handler resolves the authenticated user from the token but issues the update against a filter built from the URL identifier, without comparing them. The route is behind an administrator token check only, so any admin session suffices. The sole guard rejects a single hardcoded demo address. An attacker can set an arbitrary password on any other admin account and sign in as it.

OpenMetadata accepts a caller-supplied post-authentication redirect target and appends the issued token to it. SamlLoginServlet reads the callback parameter and stores it in the session without validation, and the assertion consumer later formats it into a URL with the JWT and account details. The OIDC and OAuth2 handlers follow the same pattern. An attacker can craft a link that, after victim login, sends the token to an attacker-controlled destination, allowing account takeover.

one-api gates one of its two channel-pinning paths and not the other. The API key suffix path requires admin check, but the URL path parameter branch has no role check. The route is behind token authentication only, so any valid token holder can reach it. The value flows to the distributor, which loads the channel by ID without scoping, and sets the Authorization header to that channel's key. A low-privilege account can pin any channel, sending requests with an operator-configured provider key, bypassing group restrictions and model allowlists.

bestzip builds the argument list for the system zip utility without separating options from operands. The destination archive path and source paths are passed without a -- delimiter, so any source entry starting with a hyphen is interpreted as an option. zip accepts -T to test the archive and -TT to name the test command, so a source list containing those and a command string causes zip to run that command via shell. An application passing untrusted file names to bestzip can execute arbitrary commands.

FiftyOne renders a dataset field's description as markup. The sidebar field-information component passes the description to dangerouslySetInnerHTML without sanitization, allowing stored markup to execute on hover. Since the description is stored in the schema and travels with exported datasets, opening a dataset from another party can run malicious scripts in the app's origin.

SunEditor before 3.1.4 has a vulnerability in the Embed plugin that allows JavaScript execution in the editor. The plugin parses raw embed HTML and recreates external script elements, leading to XSS. If an application stores or reflects content without additional sanitization, an attacker can trigger stored or reflected XSS.