CVE Vulnerability Catalog

The unstructured library prior to 0.24.1 does not escape output in the to_html() path in ontology.py, allowing untrusted document tags, attributes, and javascript: links to survive in metadata.text_as_html. When the generated HTML is rendered in a browser, attacker-controlled script can execute in the viewer's origin.

Prior to 4.10.17-lts, JumpServer's Ansible-based automation rendering can evaluate attacker-controlled template expressions from users who have permission to create or execute affected automation tasks. Crafted content can invoke sensitive Ansible lookup behavior (file, pipe, env, password), disclosing files or environment values and potentially allowing limited command execution.

In MongoDB mongos, an authenticated user with readWrite privileges could crash the mongos process by crafting a specially formed aggregation pipeline. The issue leverages an internal aggregation stage lacking proper authorization checks, resulting in an unhandled fatal error that immediately aborts mongos.

Use of freed memory in the aggregation feature of MongoDB Server can cause the server to read data from a memory area after it has been released when an internal aggregation stage is combined with a batch size of 0. Query results returned to the client may contain unrelated data from the server process, and the server process may also terminate unexpectedly.

In ESP-IDF BlueDroid, L2CAP configuration-option processing in l2c_main.c and l2c_utils.c does not consistently validate option lengths against the remaining received packet bytes. A remote unauthenticated BR/EDR peer within radio range can send a malformed L2CAP configuration request causing an out-of-bounds read, and under some runtime conditions limited adjacent Bluetooth stack memory can be included in an outbound L2CAP Configuration Response.

The vulnerability in Xen concerns logic introduced for XSA-308, which does not handle a certain combination of VMX VMEntry state when #DB is intercepted, single stepping is active, and blocked by STI/MovSS. With branch tracing enabled, the prior adjustment still causes a VMEntry failure, fatal to the guest.

The vulnerability in Xenstore concerns live update, where the state record for a Xenstore watch with specified watch depth is written with two bytes missing. This results in an inconsistent state when reading the state back in the new Xenstore, causing a crash of Xenstore or wrong Xenstore data.

In KeePassXC prior to 2.8.0-beta1, KeePass1 / KDB legacy import retains raw Entry pointers in m_entryUuids after deleting metastream entries. When a crafted .kdb file contains an earlier metastream entry followed by a KPX_CUSTOM_ICONS_4 metastream referencing the earlier UUID, parseCustomIcons4() calls Entry::setIcon through the stale pointer, causing a heap use-after-free with a controlled write and a reliable process crash.

In ConnectBot SSH Client Library prior to 0.3.1, the SSH protocol parser trusts attacker-controlled inner byte lengths and element counts without verifying that they fit in the containing packet. A malicious or compromised SSH server can return a small malformed packet whose oversized declarations cause excessive allocation or repeated parsing, leading to an uncaught OutOfMemoryError and termination of the application process.

In ConnectBot SSH Client Library prior to 0.3.1, DerReader.readLength() accepts excessive long-form DER length octets, accumulates them into an Int, and primitive readers allocate arrays without checking the declared size against the remaining input. Loading an attacker-provided private-key file through SshClient.authenticatePublicKey(), SshKeys.decodePemPrivateKey(), SshSigning.sign(), or SshSigning.getPublicKey() can trigger integer wraparound, excessive allocation, or an uncaught OutOfMemoryError.

OAuthLib versions 3.0.0 through 4.0.0 compare attacker-supplied code_verifier values with stored challenges using ordinary string equality in PKCE checks, potentially leaking common-prefix timing information. This could allow recovery of the code_verifier, unauthorized access-token acquisition, and account takeover, though practical exploitation requires an intercepted single-use authorization code and many measurements. The issue is fixed in version 4.0.0.

OAuthLib versions 0.6.1 through 4.0.0, when enable_jsonp=True, reflect an unvalidated callback request parameter into success and error response bodies in RevocationEndpoint. This allows execution of attacker-controlled JavaScript in the context of an application that loads the response as trusted JSONP, if the attacker can influence the callback. The feature is disabled by default and only affects deployments that explicitly enable JSONP. Fixed in version 4.0.0.

A vulnerability was found in girishsaraf Online-Appointment-Booking-System up to commit f427b4757128ca253d33d0cc4e87bbb9c999a4d5, specifically in Admin/showdoctor.php, due to missing authorization. The attack can be carried out remotely, and the exploit has been publicly disclosed. The project has not responded, and due to its rolling release strategy, specific version details cannot be provided.

An out-of-bounds read vulnerability exists in the btmon Bluetooth monitor tool (part of BlueZ) when processing crafted HCI Return Link Keys events. A maliciously crafted btsnoop capture file can specify a large number of link keys exceeding actual packet data, causing btmon to read beyond the allocated buffer and crash.

In MongoDB mongos, an authenticated user with readWrite privileges can crash the mongos process by crafting a specially formed aggregation pipeline. The issue leverages an internal aggregation stage lacking proper authorization checks, allowing a subsequent stage to execute on the router under unintended conditions, resulting in an unhandled fatal error that immediately aborts mongos. This causes denial of service for all clients routed through the affected instance.

MongoDB Server's collection validation feature may not properly handle certain expressions stored in a validator. In some cases, this can cause the server process to stop running when a document is later inserted. The behavior can occur for authenticated users with write access.

An issue in MongoDB Server's aggregation framework can cause certain constant values to be handled incorrectly during internal query rewriting. Under certain conditions, an authenticated user with read access may receive field values in aggregation output that the query did not intend to return.

In Plane prior to 1.4.0, multiple issue-listing endpoints, including unauthenticated public deploy boards, pass the user-controlled order_by parameter to Django's order_by without allowlist validation. An attacker can use double-underscore relationship traversal to reference arbitrary related model fields, revealing database schema details through errors and potentially disclosing field values through differential ordering. Fixed in version 1.4.0.

In self-hosted Plane deployments prior to 1.4.0, unauthenticated GET requests to api/workspaces/{slug}/invitations/{pk}/join/ through WorkspaceJoinEndpoint expose the secret invitation token directly and through get_invite_link. An external attacker can retrieve a token intended for another user and submit it to the invitation acceptance flow, joining the workspace without receiving the invitation and gaining the role assigned to the invite. Fixed in version 1.4.0.

In JupyterLab from 4.5.7 until 4.5.11 and 4.6.4, the markdownviewer:trust and notebook:trust command-link handlers use tracker.currentWidget to resolve the current document rather than retaining the document whose command button was clicked. If another document becomes current while the prompt is open and the user chooses Trust, JupyterLab marks the second document trusted while leaving the clicked document untrusted. Later command buttons in the incorrectly trusted document run without a prompt, and trusting a notebook marks all cells trusted, saves unsaved edits, and persists the server signature to disk. Fixed in versions 4.5.11 and 4.6.4.