Vulnerabilities
All
My stack
Known exploited
Public exploit
Critical (9+)
Likely exploited (EPSS ≥ 50%)
Changed, 24 h
New, 24 h
Sort:
risk
modified
published
CVSS
EPSS
| CVE | CVSS | EPSS | Vendor, product | Description | Modified |
|---|---|---|---|---|---|
| CVE-2026-120338 | 10.0 critical | — | prefecthq, prefab-ui | Prefab UI's Generative UI sandbox runs caller-supplied Python inside Pyodide, Python compiled to WebAssembly, in a Deno subprocess. The feature is documented as isolation for untrusted code, including LLM-written UI. From prefab-ui 0.14.0 through 0.20.2 the launcher starts Deno with unrestricted --allow-read, --allow-write, and --allow-net, and the runner leaves the Deno API on the JavaScript global object. Pyodide's js module proxies that global, so the Python can call Deno.readTextFileSync and Deno.writeTextFileSync. A caller who can supply the code can read and write files as the server OS user, and can persist code execution by writing a .pth file into the server virtualenv's site-packages. The usual remote path is FastMCP's GenerativeUI provider, which registers the MCP tool generate_prefab_ui and forwards that tool's code argument into this sandbox. On a FastMCP server with no authentication configured, any client who can reach the server can call the tool. With authentication, any client allowed to call tools can. Generative UI is opt-in, and the host must have Deno installed. Confirmed on prefab-ui 0.20.2. Fixed in 0.21.0: after the runtime and Prefab sources load, the runner revokes Deno read, write, and network permission before it accepts code, the process is started with --no-prompt so a permission request cannot grant them again, and bootstrap network access is limited to cdn.jsdelivr.net. The fix is commit 030f074 (pull request 538). | 20 h ago |
| CVE-2025-11969 | 9.4 critical | — | wso2, wso2 api manager, wso2 identity server as key manager, wso2 open banking am | The System REST APIs do not correctly enforce access control mechanisms, allowing users with lower privilege roles to execute administrative actions that should be restricted. This weakness permits unauthorized operations on the affected product. Successful exploitation could enable a malicious actor to perform unauthorized administrative operations, including executing administrative functions. APIs exposed through the WSO2 API Manager's API Gateway are not affected by this vulnerability. | 6 h ago |
| CVE-2026-84170 | 8.3 high | — | apache software foundation, apache wicket | Allocation of resources without limits or throttling in Apache Wicket. org.apache.wicket.markup.html.form.upload.resource.AbstractFileUploadResource is the server side of the form-less upload offered by FileUploadToResourceField, which posts the selected files to a separately mounted resource rather than through a Form. The resource is an application-wide singleton with no component to consult, so newResourceResponse() takes the three limits that bound the upload - the total request size, the maximum size of a single part and the maximum number of parts - from the request parameters maxSize, fileMaxSize and fileCountMax, and hands them to the multipart parser as they arrive. Nothing binds those numbers to the field that is supposed to be uploading. A request naming a very large maxSize is therefore parsed with no effective ceiling on the total size, and omitting fileMaxSize and fileCountMax leaves the per-part and part-count limits unset as well. The values the application configured through FileUploadToResourceField#setMaxSize(), #setFileMaxSize() and #setFileCountMax() are consulted only when the client omits the corresponding parameter, so against a client that supplies one they have no effect at all. Parts above the parser's in-memory threshold are spooled to the container's temporary directory, and the file parts are then copied into the folder the application's IUploadsFileManager owns, under a sub-folder the request names, where they remain. A part that is not a file is read into the heap in full while the request is parsed. One request, repeatable, can therefore consume as much disk or as much heap as the client chooses to send. The resource carries no authorization annotation and the default authorization strategy permits every resource request, so unless the application authorizes the resource itself the request needs no authentication, no session, no prior page view and no user interaction. An application is affected where it uses FileUploadToResourceField, or mounts FileUploadResourceReference itself. This is not the ordinary upload path: the resource is a singleton that has to be created and mounted deliberately. An application that configures no upload limit at all loses nothing, because Wicket applies no total-size limit of its own by default; what the defect removes is a limit the application set. Uploads through Form with FileUploadField, and through AjaxFileDropBehavior, are not affected, because both read their limits from the component rather than from the request. The upload resource does not exist in the 8.x line. This issue affects Apache Wicket: from 9.14.0 through 9.24.0, from 10.0.0 through 10.9.1. Users are recommended to upgrade to version 9.25.0 or 10.12.0, which fix the issue. | 13 h ago |
| CVE-2026-83594 | 8.3 high | — | apache software foundation, apache wicket | Memory allocation with an excessive size value in Apache Wicket. org.apache.wicket.markup.html.form.upload.resource.FileUploadToResourceField uploads files to a separately mounted resource rather than through a form, and is told the outcome afterwards over its own Ajax callback. That callback reads the request parameter filesInfo, a JSON array carrying a client file name, a size and a content type for each file, and turns it into the list of FileUploadToResourceField.UploadInfo that onUploadSuccess() receives. Nothing binds that JSON to an upload the resource performed and nothing re-derives it from the stored files, so the number of entries and every value in them are whatever the request supplied. The callback URL is rendered into the page, no prior upload is required, and both GET and POST reach it, so any client that can load the page can have the component report files that were never uploaded, without authenticating beyond whatever the page itself requires and without any user interaction. UploadInfo#get(), which the class offers as the way to obtain the bytes of an upload, allocates a byte array from UploadInfo#getSize() before it opens the file. The array is therefore sized by the number the request supplied, and the allocation happens even for a file that was never stored. One small request can cost the server close to two gigabytes of heap, and the request is trivially repeatable, so a few of them exhaust the heap for every request the Java virtual machine is serving. A size that the conversion to int turns negative instead throws NegativeArraySizeException, which the exception handler inside that method does not cover, so it escapes into the application. An application is affected where it uses FileUploadToResourceField, which is not the ordinary upload path through FileUploadField and Form and has to be extended and mounted deliberately. The allocation additionally requires the application to call UploadInfo#get() rather than reading UploadInfo#getFile() itself. Where the application denies the ENABLE action on the field through its authorization strategy, the callback is refused like any other listener. The 8.x line does not contain this component and is not affected. This issue affects Apache Wicket: from 9.14.0 through 9.24.0, from 10.0.0 through 10.11.0. Users are recommended to upgrade to version 9.25.0 or 10.12.0, which fix the issue. | 13 h ago |
| CVE-2026-120316 | 8.2 high | — | http4k | http4k before 6.62.0.0-ee contains a path traversal vulnerability in ThymeleafTemplates Caching and HotReload that allows unauthenticated attackers to read files outside baseTemplateDir by supplying ../ segments in ViewModel.template() values. Attackers can influence request-driven template selection so Thymeleaf's FileTemplateResolver resolves escaped paths without containment checks, disclosing any file readable by the server process. | 10 h ago |
| CVE-2026-82956 | 7.6 high | — | apache software foundation, apache wicket | Improper neutralization of input during web page generation in Apache Wicket. org.apache.wicket.ajax.AjaxNewWindowNotifyingBehavior notices that a page instance has been opened in a second browser window or tab. It asks the browser for the value of window.name, which reaches the server as the request parameter windowName, and remembers what the browser reported in a field of the behaviour. That field is part of the page's serialized state, so the remembered value stays in the page store for as long as the page instance does. On every later render of that page instance the behaviour builds the precondition of its Ajax call by substituting the remembered value into a single-quoted JavaScript string literal, and Wicket writes that precondition into the page as raw JavaScript, because the precondition of an Ajax call listener is script supplied by the caller and is emitted without quoting or escaping. Nothing on that path escapes the value. The window name is client-supplied and Wicket applies no validation to it, so a remembered name containing a single quote ends the literal and the rest of it runs as script in the application's origin, every time that page instance is rendered. An application is affected only where it has explicitly added AjaxNewWindowNotifyingBehavior to a component. No part of the framework installs the behaviour, so an application that does not use it is not affected. The name the behaviour generates for itself, when the browser reports no name at all, is a random UUID and is not client-supplied. This issue affects Apache Wicket: from 8.0.0 through 8.19.0, from 9.0.0 through 9.24.0, from 10.0.0 through 10.11.0. Users are recommended to upgrade to version 8.20.0, 9.25.0 or 10.12.0, which fix the issue. | 13 h ago |
| CVE-2026-84172 | 6.3 medium | — | apache software foundation, apache wicket | Files accessible to external parties in Apache Wicket. org.apache.wicket.markup.html.PackageResourceGuard decides whether a file on the classpath may be streamed to a client as a package resource. HTML is on the default allow list, so the only thing that separates a static HTML file, which may be served, from a component's markup template, which may not, is a check for a class file of the same name beside it. That check calls getResource() on the class loader that loaded wicket-core. The resource itself is reached through different class loaders: the scope class is resolved through the application's IClassResolver, which uses the thread context class loader, and org.apache.wicket.core.util.resource.ClassPathResourceFinder#find() then tries the scope class' own loader, the context class loader, and the wicket-core loader last. Where wicket-core is loaded by a class loader that cannot see the application's classes, the check finds no class file, concludes the file is not a component markup template, and serves it. A request for the package resource URL of a component class then returns that component's markup template with status 200. What is disclosed is developer-authored template content: component structure, developer comments, markup that is rendered only under some conditions, inline script, and the endpoints a template names. Access is read-only, and no user data, credential or session material is reachable this way. Such a deployment is rare. The ordinary packaging, a web archive with wicket-core in WEB-INF/lib, puts the framework and the application in one class loader, so the check finds the class file and the template is refused; those deployments are not affected. The exposure is confined to deployments that give wicket-core a class loader which cannot see the application's component classes: wicket-core placed in a servlet container's shared library directory, in an enterprise archive library directory with the application in the web archive, as an application server global module, or as an OSGi bundle, where wicket-core's bundle cannot resolve a class held by an application bundle. Markup shipped by Wicket's own modules, and by any other code sharing a class loader with wicket-core, remains protected. An attacker has to name a component class, and Wicket writes fully qualified class names into the URL of every package resource it renders. This issue affects Apache Wicket: from 8.0.0 through 8.19.0, from 9.0.0 through 9.24.0, from 10.0.0 through 10.11.0. Users of Apache Wicket 7.x or older, which are no longer supported, should upgrade to a supported version. Users are recommended to upgrade to version 8.20.0, 9.25.0 or 10.12.0, which fix the issue. | 13 h ago |
| CVE-2025-13647 | 5.8 medium | — | wso2, wso2 identity server, wso2 open banking iam, wso2 identity server as key manager | The affected component fails to enforce expected validation constraints on the `content-type` header when processing certain requests. This allows an attacker to submit a crafted request with a manipulated `content-type` header. Successful exploitation can lead to the unintended exposure of the file paths for certificate files. This information disclosure could potentially be leveraged in subsequent attacks or compromise the confidentiality of the system. | 2 h ago |
| CVE-2024-5617 | 4.8 medium | — | wso2, wso2 api manager, wso2 identity server, wso2 carbon identity management | The account lock feature fails to correctly enforce account lockouts on the primary user store when users with identical usernames exist in both primary and secondary user stores. This occurs because the lock condition is only applied to the secondary user store account in such scenarios, allowing continued authentication attempts against the primary user store account. This functional flaw enables brute force attacks. By systematically attempting different passwords against the primary user store account, an attacker can potentially gain unauthorized access to sensitive information or accounts, as the intended account lockout mechanism does not prevent repeated login attempts. | 6 h ago |
| CVE-2024-7075 | 4.6 medium | — | wso2, wso2 enterprise integrator, wso2 api manager, wso2 identity server | The user search function within the management console fails to properly encode user-supplied input before rendering it in the browser. This allows an attacker to inject malicious scripts that are then executed by the user's browser in the context of the management console. Successful exploitation enables an attacker to redirect the user's browser to a malicious website, modify the web page's user interface, or extract information from the browser. However, due to the use of the httpOnly flag on sensitive session cookies, session hijacking or similar attacks are prevented. | 3 h ago |
| CVE-2026-18800 | 4.4 medium | — | wso2, wso2 api manager, wso2 traffic manager, wso2 api control plane | When an Event Publisher output adapter is configured with properties that are irrelevant to its function, the affected products log these irrelevant properties. This can inadvertently include sensitive user credentials or other confidential information depending on the nature of the misconfigured property. A malicious actor with access to the 'wso2carbon' log files could retrieve sensitive information logged by the system. This may potentially lead to unauthorized access if the logged data contains credentials or other critical secrets. | 2 h ago |
| CVE-2026-107457 | 2.3 low | — | qt | A memory corruption vulnerability in the QMqttClient class of the Qt MQTT module. Under certain MQTT 5 connection settings between client and server, QMqttClient may corrupt memory, which can cause the application to crash. | 3 h ago |
| CVE-2026-120310 | 2.3 low | — | parse-community, parse-server | Parse Server before 8.6.109 and 9.0.0 before 9.10.6-alpha.1 returns stored field values in object update responses without checking read access, so clients whose class-level permissions or object ACLs allow update but not read can see object data. Attackers can update such objects to read atomic operation results (Increment, Add, AddUnique, Remove), Cloud Code trigger-set values, and, for dot notation keys under certain conditions, the whole stored parent field. | 10 h ago |
| CVE-2026-82558 | 2.3 low | — | apache software foundation, apache wicket | Missing authorization in Apache Wicket. org.apache.wicket.protocol.ws.api.WebSocketBehavior, through which Wicket's native WebSocket support hands a message sent by the client to the application, calls its onMessage() callbacks without checking whether the client may invoke the component the behavior is attached to. Wicket opens a single WebSocket connection per page, and a message received on it is delivered to every WebSocketBehavior of that page. For an Ajax behavior Wicket makes this check before invoking it, through Behavior#canCallListener(Component), which refuses a component that is not visible or not enabled, including one for which the IAuthorizationStrategy denies the RENDER or ENABLE action. A WebSocketBehavior attached to such a component, or to a child of one, therefore receives the client's text and binary messages, together with a WebSocketRequestHandler that writes back to that client, where an Ajax behavior on the same component would be refused. An application is affected where it attaches a WebSocketBehavior that overrides onMessage() to a component that it hides, disables, or restricts through its IAuthorizationStrategy, for example with MetaDataRoleAuthorizationStrategy or the AuthorizeAction annotation. The client needs a session of their own and a page of their own that contains such a component: messages are delivered only to pages in the page store of the client's own session, so this gives no access to another user's session, page or connection. Messages pushed by the server are not affected. This issue affects Apache Wicket: from 8.0.0 through 8.19.0, from 9.0.0 through 9.24.0, from 10.0.0 through 10.11.0. Older, unsupported releases from 6.0.0 onwards are also affected. Users are recommended to upgrade to version 8.20.0, 9.25.0 or 10.12.0, which fix the issue. | 14 h ago |
| CVE-2026-13575 | unscored | — | no CPE yet | HTTP::Session versions before 0.54 for Perl allow memcached command injection via unfiltered session ids in HTTP::Session::Store::CHI. The session id from the client's cookie reaches the CHI cache unchanged. The sibling HTTP::Session::Store::Memcached rejects ids that contain control characters, but the CHI store has no such check, and CHI does not escape keys for its memcached drivers. With a memcached driver, a CR and LF in the session id end the cache command early, and the server reads the bytes after them as further commands. | 16 h ago |
| CVE-2026-74759 | unscored | — | no CPE yet | Google::Auth versions 0.12 for Perl allow server side request forgery via unvalidated credential_source URLs in AWS external account credentials. An external_account configuration with the AWS subject token type selects Google::Auth::ExternalAccountCredentials::Aws. Its retrieve_subject_token sends requests to the imdsv2_session_token_url, url and region_url fields of credential_source without the host check that the other credential classes apply. Any caller that builds credentials from a configuration it does not fully control issues GET and PUT requests to hosts the configuration names, including internal services and link-local metadata endpoints. | 19 h ago |
16 of 16
Download these 16 as CSV · subscribe to this list.
EPSS scores: 2026-11-01. NVD cursor: 2026-10-29 00:22. The News page shows only notable CVEs (KEV, your stack, new criticals, EPSS over 50%, or active exploitation per SSVC); this table has all of them.