CVE-2026-93320 is a medium severity vulnerability in moby BuildKit where the system may be tricked into performing file operations on special file inodes instead of regular files. This can cause operations to be blocked or, on rootful workers, allow unintended access to host devices. The issue arises from a confused deputy scenario where BuildKit mishandles file types during processing.
Capacitor is a cross-platform native runtime for web applications. From 6.0.0 until 6.2.2, 7.6.9, 8.3.5, 8.4.3, and 8.5.1, the Android and iOS WebView navigation guard validates a target URL's host and scheme but not its path, allowing a victim who activates an untrusted link to navigate a frame to /_capacitor_http_interceptor_. The native proxy can fetch an attacker-selected URL and return the response as a document at the application's own origin, allowing script in that response to access same-origin storage, cookies, and registered Capacitor plugin capabilities. Applications remain affected when CapacitorHttp is disabled because affected releases serve the proxy path regardless of that setting. This issue is fixed in versions 6.2.2, 7.6.9, 8.3.5, 8.4.3, and 8.5.1.
CVE-2026-72668 is a high-severity vulnerability in Elastic Kibana versions 9.4.0 through 9.4.6 involving an unintended proxy or intermediary ('Confused Deputy') issue in the Kibana Agent Builder. This flaw allows a non-administrative user who can edit a shared agent to cause privileged operations to be executed under the identity of a higher-privileged user. If the same user can also author workflows, this can escalate to full administrative control of Kibana and the Elasticsearch cluster.
MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, an HTTP transport deployment with READ_ONLY_MODE=false accepts a request without an Authorization identity and permits attacker-controlled Atlassian service headers, including X-Atlassian-Confluence-Url, to select a public attacker hostname or one allowed by MCP_ALLOWED_URL_DOMAINS. A caller can then invoke confluence_upload_attachment or the Jira attachment variant in src/mcp_atlassian/jira/attachments.py with a server-local file_path and cause the MCP process to send the file to the selected attachment endpoint. This issue is fixed in version 0.22.0.
MCP Atlassian is a Model Context Protocol (MCP) server for Atlassian products (Confluence and Jira). Prior to 0.22.0, the Jira update_issue attachments argument is converted into local paths and routed to the attachment upload implementation without workspace validation. A caller can make the MCP server read arbitrary local files and attach them to a Jira issue, using the server as a confused deputy to exfiltrate the contents. The advisory traces the vulnerable input and processing flow through jira update_issue, attachments, upload_attachment, and file_path, which identify the affected entry points, controls, and code paths. This issue is fixed in version 0.22.0.
Metacat is data repository software that helps researchers preserve, share, and discover data. Prior to 3.4.2, MetacatSolrIndex.query forwards the client-controlled qt parameter through Apache SolrJ from search endpoints such as /d1/mn/v2/query/solr/ to its privileged Solr backend. An unauthenticated client can select the /admin/file handler, and SolrJ reformats the parameter into a request accepted even when handleSelect=false is configured on Solr 7.0 or later. When Solr returns the selected core configuration file, Metacat embeds the raw content in an XML processing error response, disclosing internal files such as solrconfig.xml and enabling infrastructure profiling. This issue is fixed in version 3.4.2.
CVE-2026-45723 is an improper input validation vulnerability in siderolabs omni prior to versions 1.6.6 and 1.7.3. An authenticated Operator can manipulate the TalosVersion field to cause the application to issue HTTP GET requests to unintended paths on the configured image-factory host. This can lead to same-host endpoint probing and possible disclosure of internal diagnostic information. The vulnerability does not allow redirection to other hosts or write operations. The issue is fixed in versions 1.6.6 and 1.7.3.
Nuxt OG Image generates OG Images with Vue templates in Nuxt. From 6.0.2 until 6.7.0, nuxt-og-image exposes the unauthenticated /_og/d/** route when the documented defaults security.strict = false and security.secret = "" are used, and base64url-decodes the fonts parameter through decodeOgImageParams. Attacker-controlled fonts[].path values flow through loadDefinedFonts into the font-assets/node.js binding, which performs a server-side fetch without validating the URL scheme, origin, resolved address, or redirects. This permits blind requests to loopback, private, link-local, cloud metadata, and other internal HTTP services, while differences in the outer response status and timing can reveal service reachability. Slow targets can also occupy OG image render workers for the configured fetch and render timeouts. This issue is fixed in version 6.7.0.
Strimzi Kafka Operator versions prior to 1.0.1 contain an improper privilege management vulnerability. An attacker able to create a Kafka custom resource can manipulate the entityOperator's watchedNamespace field to cause the operator to create a Role with full Secret CRUD permissions in a target namespace. This Role is bound to a ServiceAccount in the attacker's namespace, allowing the attacker to mint tokens and access Secrets across namespaces where the operator has permissions. The vulnerability is fixed in versions 1.0.1 and 1.1.0 by disabling the watchedNamespace feature by default and requiring explicit enabling.
Anyquery is an SQL query engine built on top of SQLite. Prior to 0.4.5, anyquery server exposes URL-capable SQLite virtual table modules such as json_reader and log_reader through its unauthenticated MySQL-compatible server port without restricting outbound destinations. A remote attacker can provide a loopback, private-network, or link-local cloud metadata URL, causing go-getter in the Anyquery server process to fetch the selected resource and expose its response as queryable table data. This permits internal network probing, access to internal APIs, and disclosure of cloud credentials; low-integrity impact is possible when a reached internal API performs state-changing actions. This issue is fixed in version 0.4.5.
A Pre-authentication SSRF vulnerability exists in the SMA1000 Appliance Work Place interface due to an unintended alternate access path. A remote unauthenticated attacker could potentially exploit this vulnerability to gain unauthorized access to sensitive functionality and perform unauthorized operations.
Wallos is an open-source, self-hostable personal subscription tracker. Prior to version 5.0.0, the fix for CVE-2026-33407 (GHSA-hhjq-82f8-m6rc, "SSRF via HTTP Proxy Environment Variable") hardened endpoints/logos/search.php by disabling cURL proxying (CURLOPT_PROXY = '' + CURLOPT_NOPROXY = '*'). However, Wallos ships a second, near-identical, unauthenticated logo-image search endpoint — endpoints/payments/search.php — that was not given the same hardening. It still passes the HTTP_PROXY/HTTPS_PROXY environment variable straight into CURLOPT_PROXY. This issue has been patched in version 5.0.0.
MagicMirror² versions prior to 2.37.0 contain a server-side request forgery (SSRF) vulnerability in the ADD_CALENDAR handler of the calendar module. This handler accepts attacker-controlled URLs and authentication data via an unauthenticated Socket.IO namespace, allowing SSRF without proper validation. The vulnerability can lead to exfiltration of internal service response data if the response is a valid iCal calendar. The issue is fixed in version 2.37.0.
CVE-2026-73424 is a medium severity vulnerability in the Astro web framework's Vercel adapter versions from 10.0.3 up to but not including 11.0.3. The issue allows unauthenticated GET requests to render routes that are intended to be protected by Vercel edge path rules or split edge middleware. This occurs because the adapter accepts the x_astro_path parameter for the public /_isr function based solely on the x-vercel-isr header, leading to unintended proxy or intermediary behavior (CWE-441). The vulnerability is fixed in version 11.0.3.
The Elastic Cloud on Kubernetes (ECK) operator reads a list of secret references from an annotation on secrets it manages, and it accepts the namespace recorded in each reference without validating that the reference is authorized for the resource being reconciled. A user whose Kubernetes permissions are limited to their own namespace can write that annotation, trigger a reconcile, and cause the operator to use its cluster-wide secret permissions to copy the contents of a secret from any other namespace into a secret the user can read.
Sub2API is an AI API gateway platform designed to distribute and manage API quotas from AI product subscriptions. From 0.1.135, to 0.1.168, platform API keys issued to tenants are exchanged for upstream requests made with shared provider accounts (ChatGPT/Codex OAuth, OpenAI platform keys, or an operator-configured base URL) that belong to the operator, not to the caller. The `POST /responses/*subpath` wildcard routes spliced the client-supplied subpath into the upstream URL with no validation. This lets an authenticated tenant relay requests to arbitrary upstream endpoints using pooled account credentials via a path traversal. This vulnerability is fixed in 0.1.169.
A flaw was found in the `odh-model-controller`. An authenticated user with permissions to create custom resources can exploit a vulnerability in the `loadSecret` function. This function improperly reads the Secret namespace from user-controlled input without validation. This allows an attacker to read sensitive API keys and cloud credentials from other namespaces, leading to information disclosure.
In versions of the Datadog Android application prior to v545-5.9.2, OnCallNotificationActivity is declared exported with no permission guard. A co-installed application can launch it with attacker-controlled Intent extras, including a full-screen lock-screen message, an arbitrary on-call page ID, and an arbitrary Intent to run inside the Datadog process.
This requires:
A malicious application co-installed on the victim's device.
An active Datadog session in the Android app.
Impact: After a single tap on the Acknowledge button, the app sends a forged on-call acknowledgement to the backend under the victim's session, launches the attacker-supplied Intent from within the Datadog process (reaching otherwise non-exported components), and turns on the screen while dismissing the keyguard.