CVE Veritabanı

In AVideo through 29.0, the API get_api_video endpoint contains a broken access control vulnerability in the clean_title branch that returns user-group-restricted videos with owner PII to anonymous callers. Attackers can query videos by their public slug to bypass group restrictions and retrieve sensitive user fields including email, phone, address, birth date, and administrator status.

RMCP is an official Rust SDK for the Model Context Protocol. Prior to 2.1.0, the rmcp crate's StreamableHttpClientTransport in crates/rmcp/src/transport/common/reqwest/streamable_http_client.rs builds its default_http_client with reqwest's automatic redirect policy and applies caller-supplied values from StreamableHttpClientTransportConfig.custom_headers without marking them as sensitive. When a malicious or compromised MCP endpoint returns a cross-origin 307 or 308 redirect, reqwest follows the redirect and apply_custom_headers causes custom API keys or authentication tokens to be replayed to the new origin, where an attacker can capture and reuse them. The separate auth_header path is not affected because it uses the standard Authorization header, which reqwest strips on cross-origin redirects. This issue is fixed in version 2.1.0.

djust provides Phoenix LiveView-style reactive server-side rendering for Django with Rust-powered performance. Prior to version 1.0.7, the live (WebSocket) transport authorizes a mount via `check_view_auth`, not Django's `View.dispatch()` chain. As a result, standard Django authorization — `LoginRequiredMixin`, `PermissionRequiredMixin`, `UserPassesTestMixin`, `@method_decorator(login_required, name="dispatch")`, and custom `dispatch()` guards — and the djust admin extension's staff gate (applied only in the HTTP `as_view` wrapper) were enforced on the initial HTTP GET but silently bypassed over WebSocket, where all events and state flow. An anonymous or under-privileged client could open a WebSocket and mount such a view — including admin list/create/change/delete — and dispatch its handlers. This is fixed in djust 1.0.7. `check_view_auth` now honors the Django `AccessMixin` family on every transport; a new system check S004 fails loud at startup on auth patterns the runtime cannot safely replay (decorator/overridden-`dispatch` forms); and the admin base mixin declares `login_required = True` + an active-staff `check_permissions` gate. As a workaround, gate views using djust's `login_required` / `permission_required` / `check_permissions` attributes (honored on all transports) rather than HTTP-only mixins/decorators.

djust provides Phoenix LiveView-style reactive server-side rendering for Django with Rust-powered performance. Prior to version 1.0.7, many djust built-in component template tags (`djust.components.templatetags.*`) render a developer/user-supplied URL into an `href` / `action` attribute, HTML-escaping it with `conditional_escape` but never validating the URL scheme. HTML escaping prevents attribute breakout but does not neutralize a `javascript:` URI (which needs no escapable characters), so a URL value of `javascript:alert(document.cookie)` lands verbatim in `<a href="javascript:alert(document.cookie)">` and executes in the victim's session on click. Version 1.0.7 contains a fix. As a workaround, do not pass user-controllable URLs to the affected built-in component tags; pre-validate URL schemes in application code before binding them to component arguments.

djust provides Phoenix LiveView-style reactive server-side rendering for Django with Rust-powered performance. Prior to version 1.0.7, for views that opt into state snapshots, the snapshot `state_json` embedded in the client page was restored on reconnect as trusted view state with no integrity check. A client could edit the unsigned `state_json` in their page and return it in the reconnect mount frame to inject arbitrary view attributes — e.g. flip `is_admin` to `True`, or change `account_id` / `balance` — escalating privilege or tampering with business state held in public view attributes (the normal djust pattern). This issue is fixed in djust 1.0.7. State snapshots are signed; unsigned or forged snapshots are rejected on the back-navigation restore path. As a workaround, do not enable state snapshots; do not hold authorization/ownership state in public view attributes.

djust provides Phoenix LiveView-style reactive server-side rendering for Django with Rust-powered performance. Prior to version 1.0.7, SSE sessions were keyed solely by a client-chosen `session_id` with no binding to the authenticated user — a control the WebSocket transport has but that was dropped on SSE. An attacker who learns (or a victim who leaks) a `session_id` could connect to the message endpoint and dispatch event handlers that execute with the victim's identity and state. This is fixed in djust 1.0.7. Each SSE session is bound to its owning principal at creation and cross-principal access is rejected; SSE session creation is additionally capped per principal. As a workaround, disable the SSE transport.

decap-server contains a path traversal vulnerability in the local proxy containment guard that uses plain string prefix comparison without path separator validation. Attackers can access sibling directories whose names begin with the repository directory name to read, write, or delete files outside the intended repository root.

Metabase through 0.63.18 fails to properly validate the unspecified address 0.0.0.0 in custom GeoJSON URLs, allowing unauthenticated attackers to reach loopback services. Attackers can save a malicious GeoJSON entry with 0.0.0.0 and trigger requests that return loopback service responses to unauthenticated callers.

changedetection.io through 0.60.6 fails to escape the scraped page title in HTML notifications, allowing arbitrary markup injection. Attackers can place malicious markup in monitored page titles that reaches notification channels like email and Telegram as live content when the watch_title token is used in templates.

changedetection.io through 0.60.6 fails to validate the Goto URL action in browser steps, allowing unauthenticated attackers to access internal addresses. Attackers can supply arbitrary internal URLs in the optional_value parameter to retrieve responses from restricted network locations.

ComfyUI before 0.30.0 fails to sanitize folder_name input in dataset save nodes, allowing attackers to write files to arbitrary paths outside the output directory. Attackers can load a crafted workflow that writes attacker-controlled content to arbitrary locations, enabling code execution through modified startup files or package initializers.

Nango through 0.70.4 fails to validate caller-supplied connection configuration values interpolated into provider token and proxy URL templates. Authenticated attackers can supply malicious configuration values to direct server requests at internal addresses or cloud metadata endpoints, potentially exfiltrating provider credentials.

UVdesk Community Skeleton through 1.1.8 fails to authenticate or validate installation state on wizard endpoints in ConfigureHelpdesk controller actions. Unauthenticated attackers can repoint the database and create super administrator accounts by submitting crafted requests to wizard endpoints, gaining full control of the instance.

phpList versions before 3.6.17 fail to validate cross-site request forgery tokens in the mass subscriber removal form handler. Attackers can induce logged-in administrators to visit crafted pages that silently delete and blacklist arbitrary subscriber addresses without authentication verification.

PrestaShop psgdpr versions through 1.4.3 fail to validate that GDPR consent log entries are attributed to the authenticated customer. Authenticated attackers can submit arbitrary customer identifiers to create forged consent records for other customers, corrupting audit logs.

PrestaShop blockwishlist through 3.0.2 fails to validate wishlist ownership in the getUrlByIdWishListAction method, allowing authenticated customers to retrieve share tokens for any wishlist by identifier. Attackers can supply sequential wishlist identifiers to obtain valid share links and read other customers' private wishlist contents.

browserless versions 1.44.0 through 2.56.7 fail to enforce file protocol restrictions in Playwright websocket endpoints, allowing authenticated token holders to read arbitrary files. Attackers can navigate Playwright-driven browsers to file scheme URLs and access files accessible to the container process despite the ALLOW_FILE_PROTOCOL setting defaulting to false.

OpenSign through 2.41.3 fails to validate caller identity in the getDocument cloud function when one-time-password verification is disabled. Attackers can supply a document identifier from guest signing links to retrieve complete document details including all signers' information, sender identity, and valid download tokens without authentication.

Coze Studio through 0.5.1 fails to restrict the server URL supplied when registering plugin tools, allowing authenticated users to make the backend fetch internal services. Attackers can construct plugin requests to access cloud metadata endpoints and internal services reachable only from the backend network, reading responses containing sensitive information.

Manticore Search versions 27.0.0 before 28.4.4 fail to validate permissions for all statements in multi-statement SQL requests, allowing read-only users to execute unauthorized queries. Attackers can append additional SELECT statements after the first statement to read credential tables and obtain password hashes that authenticate as administrators without plaintext recovery.

Docs before 5.4.1 fails to properly revoke websocket collaboration connections when access is revoked at parent documents. Attackers with revoked access can retain real-time read and write access to sub-documents through open websocket sessions that are never disconnected.

cc-connect through 1.5.0 fails to enforce per-user allowlist filtering in the onCardAction handler for Feishu interactive card callbacks. Attackers can dispatch agent commands by triggering card actions in admitted chats, bypassing the per-user access controls that protect the text message handler.

kan through 0.6.0 fails to properly validate board creation permissions in the GitHub project import endpoint, allowing guests to create boards despite lacking board:create permission. Attackers can bypass authorization checks by using the importProjects mutation to create boards while remaining blocked on direct creation paths.

LibreTranslate through 1.9.6 omits the access_check decorator from the download_file route, allowing unauthenticated access to translated files. Attackers can bypass API key requirements and abuse ban lists to download files without authentication on protected instances.

Feast through 0.66.0 fails to verify JWT token signatures before establishing user identity, allowing attackers to bypass all role-based access control by presenting an unverified token with a hardcoded claim value. Attackers can obtain trusted internal identity and gain unchecked read and write access to all entities, feature views, data sources, and permission policies on the server.

Coze Studio through 0.5.1 fails to validate that table names in workflow SQL customization nodes belong to the caller's workspace. Authenticated attackers can enumerate predictable table identifiers and execute SQL statements against other workspaces' memory databases to read, insert, or delete data.

Graylog through 7.1.4 validates outbound URLs against an allowlist before making requests but fails to re-validate after following HTTP redirects. Attackers with lookup table or event notification permissions can craft allowlisted endpoints that redirect to internal services, enabling the server to fetch and return internal responses.

Higress before 2.2.4 panics when processing a Cookie header segment without an equals sign, causing the plugin wrapper to recover and return a continue action that bypasses AI token rate limiting. Unauthenticated attackers can craft a malformed Cookie header to skip rate limit checks and exceed thresholds intended to restrict costly model backend calls.

Uber Kraken through 0.1.29 fails to validate the tag parameter in the /tags/{tag} endpoint, allowing unauthenticated attackers to traverse outside the configured storage root. Attackers can use percent-encoded parent-directory segments in the tag parameter to read arbitrary files accessible to the testfs backend process.