Unbounded recursion in the a2ui web_core v0_9 renderer (DataContext.resolveSignal / resolveDynamicValue) lets a remote attacker who can supply A2UI dynamic-value or function-call expression trees crash or hang the rendering context through call-stack exhaustion, an availability-only denial of service. All versions listed through 0.10.6 (EUVD enumerates 0.10.0 through 0.10.6) are affected; the CVSS 4.0 score is 6.9 with VA:L only, and the project's own assessed vector (CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L) confirms no confidentiality or integrity impact and no authentication or user interaction requirement. A public PoC reference exists in issue #2388; it is not verified as working exploit code, and there is no KEV listing, so this is a moderate, availability-only issue rather than a top-tier emergency, prioritized as 'this_month' by vuln.today. A vendor fix is available as PR #2432 / commit c76e17272e52c12c6dcf68905b1df978c39e9aa5, which introduces a MAX_DYNAMIC_VALUE_DEPTH guard of 1000.
Cross-application artifact access in Google Agent Development Kit (ADK) for Python lets a caller in one ADK application list, read, overwrite or delete file-backed artifacts belonging to a different application when both applications share the same FileArtifactService root directory and operate under the same user_id (plus the same session_id for session-scoped artifacts). This is an authenticated, low-privilege issue (CVSS 4.0 5.3, vector PR:L) and only manifests in that specific shared-root, colliding-identity deployment; single-application deployments, apps with distinct roots, and apps with distinct user_ids are not affected, so it is a bounded tenant-isolation flaw rather than an emergency. Affected ranges are 1.19.0 before 1.39.1 and 2.0.0a1 before 2.7.0, with 2.6.0 being only partially fixed; there is no CISA KEV listing and no confirmed working exploit, only a public issue/PR reference (google/adk-python#5618 and #5619), and vendor patches are released in 1.39.1 and 2.7.0.
Essential Addons for Elementor (WordPress plugin) versions 6.7.3 up to but not including 6.8.7 fail to apply the plugin's administrator-approval requirement to accounts created through its email one-time-password registration flow, so unauthenticated visitors can register and immediately log in without an administrator ever approving them. Exploitation is only possible on sites that have both the approval requirement and the email OTP registration flow enabled at the same time; sites using approval without the OTP flow, or not using the plugin's registration features, are not exposed, which substantially limits the number of affected installations despite the plugin's wide deployment. The assessed impact is integrity-only and moderate (CVSS 3.1 base 5.3, AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:N), and no confirmed working exploit code or active exploitation is known - only a public PoC/advisory reference exists and has not been verified as working exploit code.
Unauthenticated email relay in the Royal Elementor Addons WordPress plugin before 1.7.1070 allows remote attackers to replace the recipient list of form notification emails and relay attacker-composed messages from the site's own domain. The flaw is only reachable when a form's notification recipient field is explicitly configured to echo a submitted value - a non-default setup - so sites using static or hardcoded recipient addresses are not affected. Impact is bounded to integrity (spoofed or relayed email), with no confidentiality or availability consequence, giving a CVSS 3.1 score of 5.3; a public PoC reference exists but is not verified as working exploit code, and the issue is not listed in CISA KEV.
Unauthenticated attackers can forge a valid payment confirmation and mark store orders as paid without any payment in the Sadad QA Payment WordPress plugin through version 1.28, because the authenticity check on incoming payment notifications is keyed on a merchant secret that is empty by default while the remaining signed data is attacker-controlled. Scored CVSS 5.3 (AV:N/AC:L/PR:N/UI:N, integrity-only impact), the flaw is exploitable against default installations with a reachable confirmation callback, but a deployment where the operator has configured a non-empty merchant secret breaks the forgery precondition. A public PoC reference exists (a WPScan vulnerability entry); it is not verified as working exploit code, and the vuln.today priority label is 'this_month' - a moderate-severity integrity issue rather than a top-tier emergency.
Unauthenticated attackers can mark arbitrary high-value WooCommerce orders as paid on any store running the BitPay Gateway for WooCommerce plugin through version 4.1.3, because the plugin accepts a payment-provider transaction confirmation without checking that it belongs to the order being completed or that the confirmed amount matches the order total. The only prerequisite is that the site has the BitPay payment method active and that the attacker possesses one genuine - even very low-value - BitPay transaction to reuse as the confirmation. Impact is limited to payment-status falsification (integrity) rather than site takeover or data exposure, and a public proof-of-concept reference exists (the WPScan advisory entry); it is not verified as working exploit code, no CISA KEV entry is recorded, and vuln.today schedules this at this_month priority rather than act_now.
Authenticated attackers holding a WordPress administrator account can store an encoded template-injection payload in a Smart Popup by Supsystic text block, which the plugin's template engine later evaluates server-side, allowing PHP constants such as database credentials and authentication keys to be read. Only sites running the plugin before version 2.0.2 are affected. Exploitation requires high privileges (PR:H) with no victim interaction, which confines realistic exposure to trusted or insider administrators or a hijacked admin session; the vulnerability is not in CISA KEV, and the available WPScan reference is a public PoC reference that is not verified as working exploit code. The vendor has released a fix, and the priority label is this_month.
Heap out-of-bounds reads in Assimp's MD2 (Quake II model) importer (all versions through 6.0.5) let an attacker who supplies a crafted .md2 file crash the importing application and potentially expose adjacent heap memory. Attacker-controlled numFrames and numVertices header fields overflow the 32-bit frame-size product used in MD2Importer::ValidateHeader(), wrapping the computed offsets past the fileSize bounds check so later parsing reads outside the allocated buffer. The attack is local and file-based: it needs a host application that feeds an untrusted .md2 file to Assimp, requires no authentication and no network reachability, and yields only denial of service plus limited disclosure (no code execution or write). No public exploit code or active exploitation was identified at time of analysis, and an upstream fix has been merged.
Missing authorization in the update_fees AJAX handler of the PiWeb Disable payment method / Partial payment for WooCommerce plugin (all versions before 1.1.10.9) lets unauthenticated attackers read and modify other customers' WooCommerce orders. By taking a nonce from their own order-pay page and iterating sequential order_id values, an attacker can swap payment methods, trigger fee recalculation, and disclose order line items and totals for orders they do not own, with a worst-case impact of low confidentiality and low integrity loss (CVSS 4.0 6.9; independently assessed as CVSS 3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N). No public exploit code has been identified at time of analysis and the issue is not in CISA KEV; the vendor fix is version 1.1.10.9 and this is prioritised for remediation this month.
Unauthenticated information disclosure in Dolibarr ERP/CRM before 24.0.2 lets remote users read unpublished draft job positions through the public recruitment script htdocs/public/recruitment/view.php. Exposure is confined to deployments that have enabled the optional RECRUITMENT_ENABLE_PUBLIC_INTERFACE setting; there, walking sequential provisional 'ref' values reveals job titles, descriptions, salary budgets and recruiter names and emails, with confidentiality-only impact and no effect on integrity or availability. No public exploit has been identified at time of analysis, and the vendor patch is available in 24.0.2 (fixes in pull request 41108).
Information disclosure in the Vite development server 6.4.4 lets a network attacker retrieve files that server.fs.deny is supposed to block - including .env files and certificates - by appending loader query parameters such as ?raw, ?import&raw, or ?import&url&inline to the denied path, receiving the contents with HTTP 200. Exploitation requires that the dev server was explicitly exposed to the network via --host or server.host, that the sensitive file sits inside a server.fs.allow directory, and that it matches a server.fs.deny pattern; default local-only development is not affected, and the impact is limited to disclosure (no code execution or file modification). The vendor rates this CVSS 4.0 base 6.9, the flaw is fixed in Vite 6.4.5, and no public exploit code was identified at time of analysis.
ToolJet versions from 3.20.2-lts before 3.20.242-lts can be configured with instance-level signup disabled, yet an unauthenticated remote attacker can still create an account by POSTing to /api/signup without an organizationId, because SignupDisableGuard treats the string 'false' as truthy. Successful requests join the default workspace and return a valid session, so access is limited to resources available to an ordinary member of that workspace; a proof-of-concept procedure is documented in the vendor advisory. The vendor-released fix is 3.20.242-lts.
Unauthenticated Slack token exfiltration in n8n (before 1.123.83, 2.41.4, and 2.42.1) stems from the Slack Trigger webhook failing open on signature verification when no signing secret is configured on the Slack API credential. An attacker who knows the webhook URL can name a file for the node to download, causing the workspace access token to be attached to a request sent to an attacker-controlled host, with the response readable back - a full-response SSRF that leaks a Slack credential. Authentication is not required (PR:N per the CVSS vector), but exploitation depends on the unprotected configuration and knowledge of the webhook URL; no public exploit code or confirmed active exploitation was identified at time of analysis, and the priority is moderate rather than emergency.
OData $filter injection in n8n's Microsoft Outlook contact lookup and Microsoft SharePoint v1 item update nodes lets an attacker who can feed untrusted data into those nodes' expression-bindable parameters dump an entire bound mailbox contact list or redirect SharePoint writes onto unintended list items. n8n builds these $filter string literals without escaping, so a single injected quote closes the literal and rewrites the query; the resulting impact stays inside the scope of the stored Microsoft 365 credential. Affected are npm n8n releases before 1.123.85, 2.x before 2.43.5, and 2.44.x before 2.44.1; exploitation is conditional on a workflow that binds untrusted external input (webhook or form data) into these operations, and no public exploit code has been identified at time of analysis. Vendor guidance treats upgrading as the only complete fix, with workflow auditing and access restriction as short-term measures.
Prototype pollution in Highcharts Dashboards edit mode lets an attacker who controls a component's editableOptions[].propertyPath write attacker-chosen values onto Object.prototype when a user edits that option in the dashboard, after which the injected property is visible to every script on the page. The flaw affects the @highcharts/dashboards npm package from 1.0.0 through 4.2.1 and is fixed in 4.2.2; exploitation requires the board configuration passed to Dashboards.board() to come from an untrusted source (API, CMS, user-generated content) and requires a low-privileged user to open the component in edit mode and trigger the edit. Impact is application-dependent because the injected property stays inert until other page code reads it, and no public exploit has been identified at time of analysis; the vendor states that updating is the complete fix.
Prototype pollution in Highcharts Dashboards (1.0.0 through 4.2.1) and Grid Pro (2.0.0 through 3.1.0) lets an attacker who controls a table or component column ID cause DataTable.setCell and related writers to store attacker-supplied values on Object.prototype or the Object constructor under integer (row-index) keys. The unsafe write is reachable by default in Dashboards through the Highcharts component's connector write-back (allowConnectorUpdate), taking the column ID from columnAssignment when a point is updated, and through cell editing and the public DataTable API in Grid Pro; a point update, cell edit, or API write is required, along with user interaction. Successful pollution influences array and object reads and for-in enumeration across the page, which can divert application control flow and disrupt rendering; the assessed impact is low integrity and high availability with no confidentiality impact, rated medium severity (CVSS 4.0 base 6.9, CWE-1321). No public exploit code and no confirmed active exploitation were identified at time of analysis, and vendor patches are available (Dashboards 4.2.2, Grid Pro 3.2.0).
Remote client crash in Don't Starve Together (Steam AppID 322330) up to build 747465: a malicious public dedicated-server operator can publish a lobby whose world_gen_data carries an invalid Klei persistent blob header length, and when a victim views or selects that listing the client decodes the attacker-controlled data and corrupts native memory, terminating the game. No authentication is required, but the attack is gated by user interaction - the victim must actively browse and choose the crafted lobby, so merely running the game is not enough. Impact is availability only (client crash) with no confidentiality, integrity, or code-execution impact established; no public exploit code and no confirmed active exploitation were identified at time of analysis, and a vendor fix is available.
Command injection in the BSON encoding path of the MongoDB Java Driver lets an application that wraps externally supplied, untrusted bytes in a raw BSON document emit a structurally inconsistent command, so the database can interpret attacker-supplied trailing content as additional fields of the surrounding command and alter the data read or written under the application's credentials. All versions covered by the cpe:2.3:a:mongodb:java_driver wildcard are affected, and the vendor fixed the issue in the 5.14.0 release dated October 28, 2026. Severity is medium (CVSS 4.0 base 6.9; independently assessed as CVSS:3.1 AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:L/A:N): no authentication to the driver or database and no user interaction are required beyond the ability to influence those bytes, but exposure is confined to the non-default pattern of building commands from untrusted raw documents, applications using typed or validated documents are not exposed, and no public exploit has been identified at time of analysis.
Token issuance in pretix's SSO identity-provider role can return a valid token for a response type that the client was never configured to use, bypassing the client-secret requirement associated with the authorization-code flow. The flaw affects pretix deployments acting as an OAuth/OIDC identity provider for customer accounts where at least one SSO client exists, and exploitation further depends on running JavaScript on a redirect_uri page already present in the client allowlist, plus high privilege and some user interaction. The vendor addressed it in the 2026.8.1 release; no public exploit was identified at time of analysis.
Missing authorization in GROWI's collaborative editing (Yjs) connection allows unauthenticated guests on 'Readonly' guest-access pages, logged-in read-only users, and non-active users (pending, suspended, or invited) with view access to modify page drafts without edit permission. The tampered content persists only as the page's editing draft and is committed to page history under a legitimate user's name when that user later opens the editor and saves, while the page update API still rejects the attacker's own save. Affected versions v7.0.0 through v8.0.6 are fixed in v8.0.7; no public exploit was identified at time of analysis, and the flaw is integrity-only (CWE-862) with a moderate this-month priority.
SQL injection in Recipe Card Blocks Lite (WordPress plugin) affects versions 3.5.0 up to but not including 3.6.4 and allows authenticated users with Contributor-level access or above to inject SQL through an unsanitized block or shortcode attribute, with high confidentiality impact and no integrity/availability impact; this is a legitimate but moderate-priority vulnerability, not a drop-everything emergency. The official CVSS vector uses PR:H while the plugin description cites Contributor access, so the exact privilege threshold should be confirmed with the vendor. EPSS is 0.18% (8th percentile), the vuln.today priority label is this_month, no CISA KEV entry is present, no public exploit identified at time of analysis, and vendor patch 3.6.4 is available.
Blind SQL injection in the Thrive Automator WordPress plugin affects all versions before 11.2 and is reachable by any authenticated account holding editor-level privileges or higher, which supplies a request parameter whose array key is neither sanitized nor escaped before being placed into a SQL query. Successfully exploited, it gives that editor-level user a read-oriented primitive against the WordPress database (confidentiality high; integrity and availability impact none), so the practical attacker set is limited to insiders or already-compromised editor accounts rather than anonymous visitors. No public exploit code has been identified at time of analysis, EPSS is low at 0.18% (8th percentile), and the vendor has released version 11.2 as the fix.
SQL injection in the Welcart e-Commerce WordPress plugin before 2.12.6 lets authenticated users holding the Editor role or higher inject arbitrary SQL into plugin queries, exposing stored data such as order, customer and payment records to read-out. The flaw is a sanitise-then-unmask mistake: the plugin escapes a parameter, then strips that escaping before the value reaches the query, so the protection it appears to apply is not in force at execution time. Only sites running versions below 2.12.6 are affected, exploitation requires a privileged account rather than anonymous access, vendor patches are available in 2.12.6, and no public exploit has been identified at time of analysis, with EPSS at 0.18% (7th percentile).
Stored cross-site scripting in the GeoDirectory WordPress plugin before 2.8.196 lets authenticated users with the Contributor role or higher inject script through a post badge shortcode attribute that is written into an inline event handler without JavaScript-context escaping. The payload persists in the badge and fires when a victim - typically an administrator - views and clicks it, so an attacker must first obtain a Contributor-or-higher account and then rely on victim interaction. This is a moderate, not emergency, priority: EPSS is 0.17% (6th percentile), there is no CISA KEV listing, and no public exploit was identified at time of analysis; the vendor has released a fix in 2.8.196.
Stored cross-site scripting in the GeoDirectory WordPress plugin before 2.8.196 allows authenticated users holding the Contributor role or higher to save a malicious post badge click-through link value on a listing, which is later rendered without proper sanitisation or escaping in an anchor element. The payload executes in the browser of any higher-privileged user who views the affected listing - the advisory specifically calls out an administrator reviewing moderated content - but only after that victim opens the listing, so user interaction is required. Exposure is concentrated on community and multi-author directory sites that hand Contributor-level accounts to semi-trusted users and have admins manually review submissions; EPSS is 0.17% (6th percentile), there is no KEV listing, and no public exploit is identified at time of analysis, which supports a patch-worthy but non-emergency priority of fixing this month. The complete fix is updating to GeoDirectory 2.8.196.
Stored cross-site scripting in the MP3 Audio Player WordPress plugin (versions 5.9 through before 5.16) lets an authenticated user with Contributor-level access or higher inject JavaScript through one of the plugin's audio player shortcode attributes, which the plugin writes into an HTML attribute without escaping. The injected script does not fire on its own: it runs in the browser of another user - typically an administrator previewing a pending post - who views the crafted content, giving the attacker script execution in a higher-privileged session. Exploitation requires a valid Contributor account plus victim interaction, no public exploit code has been identified at time of analysis, and the low EPSS score (0.17%, 6th percentile) with a this_month priority label reflects a measured rather than emergency-level risk; the vendor has released a fix in 5.16.
Stored cross-site scripting in the 'Copy Anything to Clipboard' WordPress plugin before version 5.5.5 lets an authenticated user with the contributor role or higher plant an unsanitized shortcode attribute value that the plugin later writes back into the page via client-side script. The payload only fires when a higher-privileged user views and interacts with the poisoned content, so exploitation requires victim interaction on top of a valid contributor-level account. Our assessment rates this medium severity (CVSS 6.8, assessed vector AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N), with an EPSS of 0.17% (6th percentile) and no public exploit identified at time of analysis.
Stored cross-site scripting in the Page Builder by SiteOrigin WordPress plugin before 2.38.0 lets an authenticated user holding the contributor role or higher save a widget property that is neither sanitized on storage nor escaped when the page builder interface renders it. The payload executes only when a higher-privileged user (editor or administrator) opens the affected post in the page builder editor, so it is a privilege-escalation-by-victim scenario rather than an attack on anonymous visitors. The vendor has released a fix in 2.38.0; EPSS is low at 0.17% (6th percentile), there is no CISA KEV entry, and no public exploit code was identified at time of analysis, which places this in a routine monthly maintenance cycle rather than an emergency response.
Stored cross-site scripting in the Royal Addons for Elementor WordPress plugin (versions 1.7.1037 through before 1.7.1070) allows authenticated users with Contributor-level access or higher to inject JavaScript via the map widget's size setting, which executes in the browser of any user who views the affected page, including an administrator reviewing a pending submission. The flaw is CWE-79 and is assessed as medium severity; the assessed vector is CVSS:3.1/AV:N/AC:L/PR:L/UI:R/S:C/C:L/I:L/A:N, reflecting low privileges required and user interaction. A vendor patch is available in 1.7.1070; no public exploit code has been identified at time of analysis, and EPSS is low at 0.17% (6th percentile).
A heap-based buffer overflow in the Bluetooth OBEX packet handling of the Alpine iLX-F511 in-car head unit (version 7.0.000 per EUVD) lets an attacker within Bluetooth range send an over-long OBEX payload that is copied into a heap buffer without adequate length validation, resulting in arbitrary code execution with root privileges. No authentication or user interaction is required, but the CVSS vector is AV:P and the vendor description explicitly limits the threat to physically present attackers, so the practical constraint is Bluetooth-OBEX proximity to a discoverable/pairable unit rather than internet reachability - which is why the base score sits at 6.8 despite high confidentiality, integrity and availability impact. No CISA KEV entry and no public exploit code were identified at time of analysis; the finding was reported through the Zero Day Initiative (ZDI-CAN-29012 / ZDI-26-829) and vuln.today prioritizes it as this_month.
Unauthenticated code execution as root on Alpine iLX-F511 in-car head units (version 7.0.000) is possible because the Android Auto connection handler copies attacker-supplied data into a fixed-length stack buffer without validating its length (CWE-121). Exploitation requires physical presence at the unit (CVSS AV:P); an attacker who can establish a wired Android Auto session with the head unit and deliver over-length data can corrupt the stack and redirect execution with root privileges, with no authentication (PR:N) and no action by the vehicle owner (UI:N). CVSS 3.1 base score is 6.8; CISA KEV does not list this issue, no public exploit code has been identified at time of analysis, and no EPSS estimate was supplied with the available intelligence.
Command injection in the 'diagapp' diagnostic component of the Alpine iLX-F511 in-dash infotainment head unit lets a physically present attacker run arbitrary commands as root, because a user-supplied string is passed to a system call without validation. Exploitation requires hands-on physical access to the head unit and interaction with diagapp; no authentication or user interaction is needed once that access is obtained, and the flaw cannot be reached remotely or over a network. A public PoC reference exists in the form of the ZDI advisory (ZDI-26-825 / ZDI-CAN-28750), though it is not verified as working exploit code; no confirmed active exploitation (CISA KEV) is reported, and the vendor-reported CVSS of 6.8 reflects the physical attack vector that keeps realistic exposure limited to shared, rented, serviced or resold vehicles and workshop settings.
Stack-based buffer overflow in the framework service of the Alpine iLX-F511 in-vehicle head unit allows an attacker with hands-on access to the device to execute arbitrary code as root. Only firmware version 7.0.000 is identified as affected, and no authentication or user interaction is required once the attacker can reach the service, which keeps the CVSS complexity low. The physical attack vector (AV:P) is the decisive constraint: the flaw is not network-reachable, EPSS and CISA KEV data provided do not indicate active exploitation, and no public exploit code has been identified at time of analysis. Reported by ZDI as ZDI-CAN-29007 / ZDI-26-822 and tracked in ENISA EUVD as EUVD-2026-112197.
Juniper Networks Session Smart Conductor contains an incorrect permission assignment flaw (CWE-732) in its RoutingManager component: processes that write routing credential material into the /tmp directory create it with permissions that other local users can read. The exposed data is sensitive routing credential information, giving confidentiality-only impact; the independent assessment scores this as a local attack (AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N) requiring an existing account on the Conductor host, which conflicts with the vendor's submitted vector (AV:N/AC:H/PR:N/UI:N/S:C) and should be confirmed against advisory JSA112183. The vendor score is 6.8 and the case is triaged for this month rather than immediate action; no CISA KEV listing and no public exploit code were identified at time of analysis.
Missing SSL/TLS certificate verification in an Acronis Cyber Protect 17 component on Linux and Windows before build 42587 lets an unauthenticated attacker who can occupy an adjacent-network man-in-the-middle position present a forged certificate that the component accepts, causing confidentiality and integrity impact. The issue is rated CVSS 6.8 (AV:A/AC:H/PR:N/UI:N) and is fixed in build 42587; no public exploit was identified at time of analysis.
Path traversal in Dell PowerProtect Data Domain (Feature Release 7.7.1.0 through 8.8, LTS2025 8.3.1.0 through 8.3.1.40, and LTS2026 8.6.1.0 through 8.6.1.20) allows a high-privileged, authenticated remote attacker to escape an intended restricted directory and disclose information, yielding high confidentiality impact with no integrity or availability effect (CVSS 3.1: 6.8, scope changed). Exploitation requires administrative-level credentials on the appliance plus network reachability, so unauthenticated or low-privileged users cannot trigger it; the realistic exposure is an insider or an attacker who already controls an admin account. No public exploit code has been identified at time of analysis, and Dell published advisory DSA-2026-416 covering the issue; this is a moderate-severity, limited-impact issue that does not rise to emergency status.
Improper validation of integrity check values in Dell PowerProtect Data Domain lets a local attacker who already holds high privileges bypass a protection mechanism, with high impact on confidentiality, integrity and availability (CVSS 6.7, CWE-354). Affected builds span Feature Release 7.7.1.0 through 8.8 and the LTS2024 (7.13.1.0-7.13.1.80), LTS2025 (8.3.1.0-8.3.1.40) and LTS2026 (8.6.1.0-8.6.1.20) trains, per Dell advisory DSA-2026-416. The dominant limiting factors are the local access vector and the pre-existing high-privilege requirement, so realistic exposure is essentially confined to trusted administrators or insiders on the appliance rather than remote or low-privileged users; no public exploit has been identified at time of analysis and the issue is not in CISA KEV.
Path traversal in gopass's filesystem storage backend lets a write-capable collaborator on a shared, git-synced password store plant a crafted entry name containing ../ sequences; when a victim runs gopass rm -r on that entry, Prune calls os.RemoveAll on a path outside the store root and can delete arbitrary directory trees the victim can write to. The issue affects gopass before 1.17.0 and is fixed in 1.17.0. Exploitation requires the victim to run recursive delete against the crafted entry, so it is local, interactive, and bounded by filesystem permissions; no confirmed public exploit code is identified, and the advisory states a PoC is available upon request.
Unauthenticated order-key disclosure and cancellation in PayPlug WooCommerce (Official) 2.9.0 through 3.1.0 allows remote attackers to obtain another customer's WooCommerce order key and cancel that customer's unpaid guest order. The plugin does not verify that the requester owns the order named in a payment request; exploitation requires no privileges or user interaction and is limited to unpaid guest orders. A vendor patch is available in 3.1.1, EPSS is low at 0.16%, and no public exploit identified at time of analysis; this is a genuine but moderate-priority issue, not a critical emergency.
Unauthenticated broken access control in the PayPlug for WooCommerce (Official) WordPress plugin before 3.1.1 lets remote attackers read, modify, and permanently delete arbitrary WooCommerce orders by supplying an order identifier to AJAX actions that perform no capability or ownership check. Sites running versions 2.8.0 up to but not including 3.1.1 are affected, and because WooCommerce order IDs are typically sequential integers, the required identifier is readily enumerable - a weak limit rather than a real barrier. CVSS is 6.5 with an assessed vector of AV:N/AC:L/PR:N/UI:N, EPSS is low at 0.15% (percentile 4%), there is no CISA KEV entry, and no public exploit code is confirmed; the vendor has released a patch in 3.1.1, and priority is this_month.
Unauthenticated settings tampering in the WPOrLogin WordPress plugin before 3.1.0 lets remote attackers rewrite the plugin's hidden-login, registration and password-recovery slugs, which can lock administrators out of the WordPress login interface. The flaw is a missing-authorization issue (CWE-862): the settings-update routine performs neither a capability check nor nonce validation, so no credentials or user interaction are required (CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:L/A:L, 6.5). Only sites running an affected plugin version are exposed; a vendor fix is available in 3.1.0, EPSS is low at 0.15% (4th percentile), and no public exploit code or CISA KEV entry was identified at time of analysis, so the realized impact is operational lockout rather than data disclosure or code execution.
Cross-site request forgery in the WordPress plugin Media Library Assistant before 3.42 allows a remote attacker to permanently delete every trashed attachment on a site by luring a logged-in Editor-or-higher user to attacker-controlled content. The Empty Trash media action performs no CSRF nonce check, so a forged cross-origin request issued in the victim's active WordPress session is honored. No public exploit code or active exploitation has been identified at time of analysis (EPSS ~0.11%), and impact is scoped to media already in the trash rather than live library items; the vendor has released 3.42 to address the flaw.
Cross-site request forgery in the WordPress plugin Media Library Assistant (versions before 3.42) allows an attacker to craft a forged bulk-action request that omits the CSRF nonce field entirely, skipping the plugin's conditional validation and causing a logged-in user with the Author role or above to trash or delete media attachments. Per the assessed vector (AV:N/AC:L/PR:N/UI:R/S:U/C:N/I:H/A:N), the attacker is unauthenticated, but exploitation only succeeds when a privileged victim with an active WordPress session is induced into issuing the forged request. There is no public exploit identified at time of analysis, EPSS is 0.11% (1st percentile), and the issue carries a this_month priority - a moderate, not urgent, integrity-only risk.
Cross-ObjectType object disclosure in Open Object (Objects API) versions 2.4.3 through 4.1.0 allows an authenticated holder of a valid, non-superuser API token with a read grant on just one ObjectType to retrieve the complete versioned record of an object belonging to any other ObjectType through GET /api/v2/objects/{uuid}/{index}, including sensitive fields such as a citizen BSN in government deployments. The same token correctly receives 403 from the normal retrieve endpoint, so only the history-detail route leaks data (CWE-639, CVSS 6.5, PR:L). The GitHub security advisory (GHSA-55w6-rqp5-j5wx) includes a local reproduction against the official 4.1.0 image; there is no CISA KEV entry or other indication of in-the-wild exploitation at time of analysis, and remediation is available in 2.5.1, 3.6.2, 4.0.2 and 4.1.1.
Cookie scope forgery in Happy DOM 20.14.5 and earlier lets an attacker-controlled origin plant a cookie that a victim origin subsequently receives, enabling cookie injection and session fixation in server-side rendering or prerendering pipelines that use the library's CookieContainer for cookie handling. The flaw is only reachable when the same Window/cookie container first fetches an attacker-controlled origin and then a request to the victim origin, so test-only usage of the headless DOM carries no impact; the assessed vector is AV:N/AC:H with low confidentiality and integrity impact. It is fixed in 20.14.6, a public PoC reference exists (it is not verified as working exploit code), no CISA KEV listing or active exploitation is indicated, and the priority is a normal monthly patch cycle rather than emergency response.
Confidentiality exposure in PostgreSQL Anonymizer (Dalibo) before 3.3.0 lets an authenticated database role that holds the TEMPORARY privilege defeat the extension's dynamic-masking validation by defining an anonymization function with a SQL-standard function body, so masked data can be read rather than anonymized. Only deployments running the extension in versions prior to 3.3.0 are affected; the mismatch is resolved in 3.3.0 and later. Severity is rated 6.5 (CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N) with high confidentiality impact but no integrity or availability effect, and no public exploit code or confirmed active exploitation was identified at time of analysis.
Data masked by PostgreSQL Anonymizer can be disclosed in cleartext when a masking rule is implemented through a sub-query: the outer result set is anonymized, but rows returned by the nested query are not, so a database account mapped to a masked role can read values the rule was meant to hide. Only deployments that use dynamic/role-based masking with at least one sub-query-based masking rule are exposed; the attacker must already hold an authenticated PostgreSQL account mapped to a masked role (PR:L), and the impact is confidentiality-only, with no integrity or availability effect. The issue is fixed in PostgreSQL Anonymizer 3.3.0 and later; no public exploit code has been identified at time of analysis, and this is a moderate-severity issue suitable for the monthly patch cycle rather than an emergency response.
PostgreSQL Anonymizer fails to invalidate query plans that were prepared before a masking rule was added or before the session's role was changed, so an already-connected session can keep returning unmasked data until that session ends. Exposure requires an authenticated database session (PR:L per the assessed vector) holding a prepared plan that predates the masking configuration change; fresh sessions and newly prepared plans are masked correctly, which limits this to a defense-in-depth data-exposure issue rather than an external threat. The vendor-resolved issue is fixed in PostgreSQL Anonymizer 3.3.0 and later; no public exploit was identified at time of analysis and there is no KEV entry.
Out-of-bounds write in Zephyr's Ethernet L2 layer (subsys/net/l2/ethernet/ethernet.c) lets a remotely sent packet that the device forwards from a non-Ethernet interface to Ethernet corrupt memory preceding the packet buffer, causing crashes, packet corruption or heap inconsistency (CWE-787, CVSS 6.5, availability-weighted). Affected releases are Zephyr 4.1.0 through 4.4.2 built with CONFIG_NET_L2_ETHERNET_RESERVE_HEADER (default-y on nRF54L and NXP Wi-Fi zero-copy TX configurations) plus a forwarding or tunnelling path such as CONFIG_NET_IPV4_FORWARDING/CONFIG_NET_IPV6_FORWARDING from 802.15.4/OpenThread or PPP to Ethernet/Wi-Fi, IPv6 multicast forwarding, or the IPIP tunnel - plain Ethernet-to-Ethernet forwarding is not affected. The write only materialises in production builds compiled without CONFIG_ASSERT and without CONFIG_NET_BUF_HARDENING; the attacker controls neither the bytes written nor the offset, and no public exploit has been identified at time of analysis (not in CISA KEV).
Denial of service in Grafana OSS and Grafana Enterprise lets an authenticated user holding only the Viewer role crash the server by submitting a crafted server-side math expression that recurses without a depth limit (CWE-674). Per the assessed CVSS vector (AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:H, 6.5), the attacker must be logged in but needs no special UI interaction, and the impact is availability only - no confidentiality or integrity effect. A transient crash is reachable with Viewer privileges alone, while making the outage survive restarts additionally requires the elevated permission to provision alert rules; EUVD lists affected branches spanning the 7.4.0 line through 13.2.x for both OSS and Enterprise editions. No public exploit code or CISA KEV entry was identified at time of analysis, and the priority signal is this_month rather than act_now.