CVE Database

Dell PowerStore contains a Missing Authorization vulnerability. An attacker with access to a mapped host could exploit this vulnerability to read from or write to LUNs that the host is not authorized to access, bypassing per-initiator LUN access controls and leading to protection mechanism bypass.

sh provides Python process launching. Prior to 2.2.4, the _uid option in sh.py performs an incomplete privilege drop on Linux and Unix-like systems. When sh runs from an elevated process and launches a command with _uid set to an unprivileged user, the child changes its UID but can retain the parent process's supplementary groups because the privilege-drop sequence does not fully establish the target user's UID, primary GID, and supplementary groups. The child can therefore retain access to files or resources granted to privileged groups such as root, docker, disk, shadow, or sudo, violating the expected _uid privilege boundary. This issue is fixed in version 2.2.4.

The Apify MCP server enables AI agents to extract data from websites using ready-made scrapers, crawlers, and automation tools available on the Apify Store. Prior to 0.10.11, getActorMCPServerURL in src/mcp/actors.ts concatenates the trusted Actor standby URL with the attacker-controlled webServerMcpPath from an Actor definition without verifying the resulting origin, allowing a malicious Actor publisher to use a userinfo-style authority value to redirect connectMCPClient to a third-party host. The call-actor, fetch-actor-details, and actor-mcp tool-loading paths pass this URL to transports in src/mcp/client.ts that attach the victim Authorization bearer token, exposing the Apify API token and enabling access to Actors, stored data, and billable compute. A victim must invoke or inspect the attacker-controlled Actor. This issue is fixed in version 0.10.11.

Lemur manages TLS certificate creation. Prior to 1.9.1, StrictRolePermission and AuthorityCreatorPermission in lemur/auth/permissions.py call flask_principal.Permission.__init__() with zero Need objects when ADMIN_ONLY_AUTHORITY_CREATION and LEMUR_STRICT_ROLE_ENFORCEMENT are unset because both flags default to False. Flask-Principal Permission.allows() returns True when self.needs is empty, so the .can() authorization gate permits every authenticated identity, including the read-only role. A read-only user can access POST /api/1/authorities, POST /api/1/certificates/upload, POST /api/1/pending_certificates//upload, POST /api/1/notifications, PUT or DELETE /api/1/notifications/, and POST /api/1/domains to create root Certificate Authorities, upload arbitrary certificates, create or edit notifications that reach an SSRF sink, and create domain entries. Explicitly setting either flag to False continues to opt into the permissive behavior. This issue is fixed in version 1.9.1.

Saleor is an e-commerce platform. From 2.10.0rc1 until 3.21.67, 3.22.63, and 3.23.22, the account activation flow treats email verification as sufficient proof of account ownership and automatically associates anonymous commerce data with the newly activated account. An attacker can use accountRegister to create an account with a victim's email address before the victim registers. If the victim follows the activation link sent to that mailbox, Saleor activates the attacker-created account and saleor/graphql/account/mutations/account/confirm_account.py can merge anonymous orders and gift-card data for the same email address without requiring the account password or another authentication factor. The attacker can then access the merged order history and personal data, including names, addresses, and phone numbers. The patched supported lines disable automatic merging by default, while the redesigned 3.24.0 flow requires password confirmation before anonymous objects are linked. This issue is fixed in versions 3.21.67, 3.22.63, and 3.23.22.

Dell AppSync Version 4.6.0.0, Dell Metro Node Version 8.0.0, Dell UCC Edge Version 3.0.1, Dell VxRail Version 8.0.322, Dell PowerMax Version 10.3.0, Dell Unity Version 5.4, Dell PowerFlex Manager Version 4.5.4, Dell PowerFlex Intelligent Catalog Versions 46.377.00 and 46.382.00 and Dell PowerFlex Rack version 4.5.4 and prior versions, contain(s) an UNIX Symbolic Link (Symlink) Following vulnerability. A low privileged attacker with local access could potentially exploit this vulnerability, leading to Elevation of privileges.

A flaw was found in managed-serviceaccount. A compromised addon-manager pod, due to its ClusterRole granting excessive permissions, can read any secret across all namespaces. Additionally, it can approve arbitrary Certificate Signing Requests (CSRs), which could lead to information disclosure and privilege escalation within the cluster.

Improper input validation in the capabilities route handler in OpenSearch Dashboards - the size of the request payload is not bounded - might allow remote attackers to cause a denial of service via a crafted HTTP request.

Joomla! Core - [20260802] - Improper CORS origin validation in Joomla 4.0.0-5.4.7, 6.0.0-6.1.2 - An improper implementation prevented configured CORS origins from being properly validated in CORS requests.

Dell PowerStore SDNAS contains a Buffer Copy without Checking Size of Input vulnerability in NFS/RPC. An unauthenticated attacker with remote access could potentially exploit this vulnerability, leading to command execution and denial of service.

The RabbitMQ Java client library allows Java and JVM-based applications to connect to and interact with RabbitMQ nodes. Prior to 5.33.1, src/main/java/com/rabbitmq/client/impl/ValueReader.java permits ValueReader.readTable and ValueReader.readArray to call ValueReader.readFieldValue recursively for AMQP table type F and AMQP array type A values without a nesting-depth limit. A malicious AMQP server or network intermediary can send approximately 580 nested table levels in the pre-authentication connection.start frame, fitting within the default 131072-byte frame maximum, to trigger StackOverflowError. The error terminates the client input processing thread and causes denial of service. This issue is fixed in version 5.33.1.

The RabbitMQ Java client library allows Java and JVM-based applications to connect to and interact with RabbitMQ nodes. Prior to 5.33.1, src/main/java/com/rabbitmq/client/impl/ValueReader.java uses ValueReader.readBytes to accept a wire-declared contentLength below Integer.MAX_VALUE and allocate a byte array before checking the bytes available in the frame. A malicious AMQP peer can send a LongString or byte-array field with type tag S and a declared length such as 0x7FFFFFFE during the pre-authentication connection.start server-properties table, causing an approximately 2 GB allocation and OutOfMemoryError before readFully consumes data. The resulting memory exhaustion can terminate the JVM and cause denial of service. This issue is fixed in version 5.33.1.

The RabbitMQ Java client library allows Java and JVM-based applications to connect to and interact with RabbitMQ nodes. Prior to 5.33.0, com.rabbitmq.tools.jsonrpc.ProcedureDescription receives a javaReturnType value in an untrusted system.describe response and passes it through JSONUtil.tryFill, setJavaReturnType, and computeReturnTypeAsJavaClass to Class.forName(javaReturnType) with initialization enabled. An attacker able to answer the JsonRpcClient request through a shared broker or network interception can select a class already present in the victim JVM and trigger its static initializer, while JsonRpcClient.java later passes getReturnType output to mapper.parse and may also create type confusion. Successful exploitation can affect confidentiality, integrity, and availability in the client process. This issue is fixed in version 5.33.0.

authentik is an open-source identity provider. Prior to 2026.2.6 and 2026.5.5, the Remote Access Control endpoint list returns every configured endpoint to any authenticated user regardless of which applications the user may access, and the response includes connection settings that can contain stored credentials. The endpoint listing does not apply the access controls governing the endpoints, and the connection flow does not confirm that an endpoint belongs to the Remote Access Control application through which it was launched. Any authenticated user can therefore read every endpoint together with its host and stored credentials and can open a connection to an endpoint belonging to another application. This exposes stored credentials for managed RDP, SSH, and VNC targets and grants interactive access to systems the user was never authorized to reach. Deployments that do not use the enterprise Remote Access Control provider are not affected. This issue is fixed in versions 2026.2.6 and 2026.5.5.

authentik is an open-source identity provider. Prior to 2026.2.6 and 2026.5.5, the enterprise Google Chrome device-trust stages advance the flow without confirming that the out-of-band device attestation actually ran. Affected enterprise deployments place either a Google Chrome Endpoint stage with mode set to REQUIRED or the deprecated Google Chrome Device Trust Connector stage in an authentication flow. The device attestation occurs in a verification iframe that calls the Google Verified Access API and records the verified device on success, but the vulnerable stages treat the flow as passed as soon as the stage is submitted. An attacker who can reach such a stage, including after primary username and password authentication, can skip the verification iframe and authenticate from a device that was never verified. Where device trust is the only additional factor, that protection is fully bypassed, while other configured factors remain in force. This issue is fixed in versions 2026.2.6 and 2026.5.5.

ePA 3.x Integration implements the authorization workflow and writes Medical Information Objects to Germany's electronic patient record. Prior to 1.3.0, ePA 3.x Integration disables TLS certificate verification for both ePA connections in app/vau/VAUProtokoll.py and Konnektor connections in app/konnektor/Konnektor.py. A network-positioned attacker can present an arbitrary certificate, terminate the TLS connection, and intercept ePA traffic. The VAU protocol does not provide an effective fallback because its application-layer certificate validation is also broken in affected versions. The Konnektor session uses self.session.verify set to False while the client authenticates with self.session.cert, so an attacker impersonating the Konnektor can receive the client's mutual TLS certificate exchange and observe smartcard operations. This issue is fixed in version 1.3.0.

ePA 3.x Integration implements the authorization workflow and writes Medical Information Objects to Germany's electronic patient record. Prior to 1.3.0, ePA 3.x Integration leaves request_counter unchanged in app/vau/VAUProtokoll.py while constructing VAU messages. The frozen client request counter causes the server side to reuse AES-GCM nonce and key combinations across responses. A network attacker who collects repeated ciphertexts can recover the XOR of plaintexts and use predictable inner HTTP headers and JSON fields to recover sensitive data, including patient health records. Repeated nonces can also enable recovery of the GHASH authentication key through the Joux forbidden attack, allowing forged AES-GCM messages and injection of malicious responses. The response-counter check also fails to maintain last_response_counter, weakening replay and ordering validation. This issue is fixed in version 1.3.0.

Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.4, Vvveb backend product operations allow a low-privileged Vendor to access products owned by another Vendor. The admin/controller/product/products.php controller accepts a caller-controlled product_id for duplicate and delete actions, and admin/sql/sqlite/product.sql loads and mutates products without consistently applying the current admin_id when view_other_products or edit_other_products is absent. An attacker can read product details, duplicate products, or delete products and related catalog data, exposing commercial information and causing unauthorized copies, catalog pollution, data loss, or business disruption. This issue is fixed in version 1.0.8.4.

Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.4, Vvveb backend product revision operations allow a low-privileged Vendor to access revisions for products owned by another Vendor. The admin/controller/product/revisions.php route reuses admin/controller/content/revisions.php, while admin/sql/sqlite/product_content_revision.sql trusts caller-controlled product_id, language_id, and created_at values without applying the current admin_id to revision reads, restores, and deletes. An attacker can read historic product content, restore a revision over another Vendor's live product content, or delete revision records, exposing private copy, corrupting product pages, and removing audit history. This issue is fixed in version 1.0.8.4.

Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.4, Vvveb backend post revision operations allow a low-privileged Author to access revisions for posts owned by another Author. The admin/controller/content/revisions.php controller and admin/sql/sqlite/post_content_revision.sql queries trust caller-controlled post_id, language_id, and created_at values without consistently applying the current admin_id to revision lists, reads, restores, and deletes. An attacker can read historic post content, restore a revision over another Author's live post content, or delete revision records, exposing drafts, corrupting published content, and removing audit history. This issue is fixed in version 1.0.8.4.

Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.4, Vvveb backend product review operations allow a low-privileged Vendor to manage reviews under another Vendor's products. The admin/sql/sqlite/product_review.sql queries accept a caller-controlled product_review_id and do not verify product_review.product_id against product.admin_id for the current admin_id. An attacker can read pending review content, ratings, author information, and moderation state, change review status, edit review content, or delete reviews, manipulating product review visibility and integrity. This issue is fixed in version 1.0.8.4.

Vvveb is a powerful and easy to use CMS with page builder to build websites, blogs or ecommerce stores. Prior to 1.0.8.4, Vvveb backend product question operations allow a low-privileged Vendor to manage questions under another Vendor's products. The admin/sql/sqlite/product_question.sql queries accept a caller-controlled product_question_id and do not verify product_question.product_id against product.admin_id for the current admin_id. An attacker can read pending question content and moderation data, change question status, edit question content, or delete questions, manipulating product Q&A visibility and integrity. This issue is fixed in version 1.0.8.4.

@neo4j/graphql from 5.2.0 until the patched versions fails to enforce field-level @authentication rules on root custom-resolver fields when a type-level @authentication rule is also present on the same operation type. When both a type-level @authentication (on Query/Mutation) and a field-level @authentication (on a root custom-resolver field within that type) are declared, only the type-level rule is evaluated and the field-level rule is silently discarded. As a result a stricter per-field requirement — such as an admin-role JWT claim (jwt: { roles_INCLUDES: "admin" }) — is never checked, and any client that satisfies the coarser type-level requirement can invoke the more-restricted field. No token forgery is involved: a legitimately issued, correctly signed non-admin token (e.g. roles: ["user"]) is sufficient.

Hugo 0.161.0 placed the Node asset pipelines behind the Node.js permission model so that code running through PostCSS, Babel, or TailwindCSS could not reach the file system outside the project directory. Hugo 0.162.0 added tailwindcss to the AllowChildProcess default in config/security/securityConfig.go, which makes nodePermissionArgs in common/hexec/exec.go append --allow-child-process whenever the tool being launched is named tailwindcss. TailwindCSS loads the site's tailwind.config.js through require at startup, so top-level code in that file executes inside the permitted Node process and can call child_process to spawn a shell. The spawned process is not a Node process and inherits none of the permission flags, so it runs with the full privileges of the account performing the build. Building a site whose theme, module, or starter template supplies the Tailwind configuration therefore yields arbitrary command execution rather than the confined file access the permission model was introduced to enforce. Hugo 0.165.0 removes tailwindcss from the default security.exec.allow list, so the tool is no longer launched under the default configuration.

CodeWhale versions before 0.8.64 contain an environment variable exposure vulnerability in the js_execution tool that fails to scrub parent process environment variables before spawning Node.js. Attackers can craft malicious JavaScript code executed by the tool to read process.env and leak API keys, cloud credentials, and authentication tokens back to the model context.

CodeWhale versions before 0.8.64 contain a path traversal vulnerability in the image_analyze tool that fails to canonicalize symlinks before reading files. Attackers can create workspace symlinks pointing to external files with image extensions to leak file bytes to the vision endpoint without user approval.

CodeWhale versions before 0.8.64 contain an argument injection vulnerability in the git_blame tool that allows attackers to read arbitrary files by injecting git options into the unvalidated rev parameter. Attackers can supply rev values like --contents=/path/to/file to exfiltrate sensitive files such as SSH keys and credentials through the tool output returned to the model.

CodeWhale versions before 0.8.64 fail to properly validate the allow_shell configuration parameter from project config files, allowing attackers to enable arbitrary shell command execution by committing a malicious .codewhale/config.toml file to a repository. When a user clones and opens the repository in CodeWhale, the AI model gains access to exec_shell and task_shell tools, enabling execution of arbitrary shell commands on the victim's machine without explicit user consent.

CodeWhale versions before 0.8.64 fail to validate file paths in the project config instructions field, allowing attackers to read arbitrary files on the victim's system. A malicious .codewhale/config.toml file in a cloned repository can specify paths outside the workspace that are read and injected into the AI system prompt for exfiltration.

CodeWhale (packages codewhale / codewhale-tui) versions >= 0.8.41 and < 0.8.64 contain a remote code execution vulnerability in the rlm_eval tool. The tool's approval_requirement() returns ApprovalRequirement::Auto, which the engine treats as 'never prompt,' causing arbitrary model-supplied Python code to run in a python3 interpreter without consulting the user's configured --approval-policy and without any approval prompt or audit step. An attacker can induce the agent to execute arbitrary code via prompt injection in untrusted content the agent reads (a web page, fetched URL, repository file, or MCP tool result); the companion rlm_open tool can stage such content. Code runs on the user's machine at the user's privilege level. Fixed in 0.8.64.