Bitwarden Server before 2026.5.0 contains a privilege escalation vulnerability that allows authenticated Custom users with ManageUsers permission to remove Admin accounts from an organization by exploiting a missing role hierarchy check in the bulk user-remove endpoint. Attackers can supply Admin organization-user IDs in a bulk DELETE request to bypass the guard enforced on the single-user removal path, effectively removing one or more Admin accounts from an organization.

justhtml versions 1.13.0 and earlier contain a parser-differential / mutation cross-site scripting (mXSS) vulnerability when using a custom SanitizationPolicy that preserves foreign namespaces (e.g., drop_foreign_namespaces=False with allowlisted SVG/MathML elements or raw-text containers such as <style>). Specially crafted input can sanitize into markup that appears safe but becomes unsafe when re-parsed by a browser or another HTML parser, allowing markup injection. The default safe configuration (sanitize=True) is not affected. Fixed in 1.14.0.

Pagekit CMS 1.0.18 contains a privilege escalation vulnerability that allows authenticated users with the 'user: manage users' permission to escalate privileges by assigning arbitrary custom roles to themselves due to missing authorization checks in UserApiController::saveAction(). Attackers can assign themselves a custom role with the 'system: manage packages' permission and then upload and install a malicious PHP package through the admin package installer to achieve remote code execution.

Control Web Panel before 0.9.8.1225 contains a blind SQL injection vulnerability that allows unauthenticated remote attackers to execute arbitrary SQL queries by submitting unsanitized input through the userRes POST parameter at the user endpoint. Attackers can exploit MySQL root privileges obtained via the injection to write arbitrary files using INTO DUMPFILE, enabling deployment of a PHP webshell to the web-accessible roundcube logs directory and achieving remote code execution as the cwpsvc account.

Ray prior to 2.56.0 contains an unsafe deserialization vulnerability in the WebDataset reader that allows attackers to achieve remote code execution by supplying a malicious tar archive to the read_webdataset() function. The _default_decoder() function in webdataset_datasource.py unconditionally calls pickle.loads() on tar entries with .pkl/.pickle extensions and torch.load() with weights_only=False on .pt/.pth entries, executing arbitrary code inside Ray remote workers on every worker that processes the malicious archive.

SuperPlane before 0.30.0 contains an SMTP header injection vulnerability that allows unauthenticated attackers to inject arbitrary SMTP headers by including CRLF sequences in the event payload title field delivered via webhook. Attackers can manipulate the unsanitized title field passed to the SMTP DATA command to add Bcc recipients for content exfiltration, forge the From address to bypass SPF and DKIM checks, or inject Content-Type and MIME boundary headers to corrupt message bodies for phishing.

SuperPlane before 0.27.0 contains a broken object-level authorization vulnerability in the CanvasService gRPC handlers that allows authenticated users with viewer-level access to one organization to access resources belonging to other organizations by supplying arbitrary canvas or queue UUIDs without organization scoping. Attackers can read cross-tenant execution history and event payloads containing sensitive secrets, write queue items and canvas events into victim organizations, delete arbitrary canvases, and disrupt automation workflows across tenant boundaries.

An insecure direct object reference (IDOR) vulnerability in the Fullstep V5 registration process allows authenticated users to access data belonging to other registered users through various vulnerable authenticated resources in the application. The vulnerable endpoints result from: '/api/suppliers/v1/suppliers//false' to list user information; and '/#/supplier-registration/supplier-registration//2' to update your user information (personal details, documents, etc.).

Zen is a firefox-based browser. Prior to 1.21.5b, Zen's glance and split-view context-menu actions, Open link in glance and Split link in new tab, load a page-controlled link URL with the System principal instead of the originating page's principal, allowing a malicious web page to place a link to a file URL that can load with System privileges when opened through either context-menu item and bypass the content-to-file security check that blocks an ordinary click. This issue is fixed in version 1.21.5b.

Inadequate access control in the registration process in Fullstep V5, which could allow unauthenticated users to obtain a valid JWT token with which to interact with authenticated API resources. Successful exploitation of this vulnerability could allow an unauthenticated attacker to compromise the confidentiality of the affected resource, provided they have a valid token with which to interact with the API.

Liman is open source server management software. Prior to 2.2.2 - 1103, an OS command injection vulnerability in the log rotation configuration endpoint allows an authenticated administrator to execute arbitrary operating system commands on the Liman server. The `ip_address` parameter is embedded directly into a shell command without sanitization, enabling shell escape via single-quote injection. This is fixed in 2.2.2 - 1103.

Coolify is an open-source and self-hostable tool for managing servers, applications, and databases. Prior to 4.0.0-beta.474, Coolify's API controllers consistently validate server ownership with Server::whereTeamId($teamId) before any operation. However, multiple Livewire web UI components accept server_id and destination_uuid from URL query parameters without any team ownership validation, allowing cross-team resource deployment. This vulnerability is fixed in 4.0.0-beta.474.

AgenticMail gives AI agents real email addresses and phone numbers. In @agenticmail/claudecode prior to version 0.2.39, @agenticmail/codex prior to version 0.1.33, @agenticmail/core prior to version 0.9.43, and @agenticmail/openclaw prior to version 0.5.71, two inbound-mail handlers act on a privileged effect without verifying that the sender is the operator, while a sibling handler in the same repo does. The higher-impact one: any external email routed to the bridge inbox causes the dispatcher to resume the operator's Claude Code session with `permissionMode: 'bypassPermissions'`, embedding the attacker-controlled `from`/`subject`/`preview` verbatim into the prompt the resumed agent reads — an indirect prompt injection into a fully-privileged agent (Bash/Write/Edit/WebFetch + the agenticmail MCP toolbelt) running as the operator's OAuth identity. The sibling operator-query email-reply hook gates the same untrusted-From provenance with `isOperatorReplySender(replyFrom, config.operato

AgenticMail gives AI agents real email addresses and phone numbers. In @agenticmail/api prior to version 0.9.64, a low-privileged authenticated AgenticMail agent can enumerate another agent's pending/claimed tasks by supplying the target agent name to `GET /api/agenticmail/tasks/pending?assignee=<name>`. The returned task objects include the task IDs and payloads. The same task IDs can then be used with the capability-style task mutation endpoints (`/tasks/:id/claim`, `/tasks/:id/result`, `/tasks/:id/complete`, `/tasks/:id/fail`) to claim, complete, or fail tasks assigned to a different agent. Because ordinary authenticated agents can discover agent names through `GET /api/agenticmail/accounts/directory`, the task ID effectively stops being a secret capability. This turns the intended capability model into a cross-agent authorization bypass. Version 0.9.64 contains a fix.

The Text Snippets plugin for WordPress is vulnerable to Stored Cross-Site Scripting via the plugin's `ts` shortcode in all versions up to, and including, 0.0.1 due to insufficient input sanitization and output escaping on user supplied attributes. This makes it possible for authenticated attackers, with contributor-level access and above, to inject arbitrary web scripts in pages that will execute whenever a user accesses an injected page.

Stirling-PDF is a locally hosted web application that facilitates various operations on PDF files. Prior to 2.9.0, the /api/v1/pipeline/handleData endpoint in app/core/src/main/java/stirling/software/SPDF/controller/api/pipeline/PipelineProcessor.java injects the STIRLING-PDF-BACKEND-API-USER API key into pipeline subrequests, allowing an authenticated ROLE_USER to retrieve the key through /api/v1/user/get-api-key, impersonate the internal service account, bypass normal rate limits, and access internal endpoints including /api/v1/info/requests/all and /api/v1/info/load/all. This issue is fixed in version 2.9.0.

Parse Server is an open source backend that can be deployed to any infrastructure that can run Node.js. Prior to 9.9.1-alpha.13 and 8.6.83, a LiveQuery subscriber could receive object field values they were not authorized to read when a single save changed both an object field and the subscriber's ACL read access, because leave and enter events included the wrong object state. This issue is fixed in versions 9.9.1-alpha.13 and 8.6.83.

Parse Server is an open source backend that can be deployed to any infrastructure that can run Node.js. Prior to 9.9.1-alpha.12 and 8.6.82, deeply nested $or, $and, and $nor query condition operators in the REST API or LiveQuery query handling could trigger exponential-time processing in the internal query-traversal helper and block the Node.js event loop. This issue is fixed in versions 9.9.1-alpha.12 and 8.6.82.

An out-of-bounds write issue in the virtio PCI transport in Firecracker 1.13.0 through 1.14.3 and 1.15.0 on x86_64 and aarch64 might allow a local guest user with root privileges to crash the Firecracker VMM process or potentially execute arbitrary code on the host via modification of virtio queue configuration registers after device activation. Achieving code execution on the host requires additional preconditions, such as the use of a custom guest kernel or specific snapshot configurations. To remediate this, users should upgrade to Firecracker 1.14.4 or 1.15.1 and later.

Deloitte AI Assist for Customer exposed unauthenticated API endpoints that allowed an attacker with knowledge of additional parameters to read from or inject content into the retrieval-augmented generation (RAG) corpus. On 2026-03-25, AI Assist for Customer restricted network access and enforced authentication for the previously exposed endpoints.