CVE Vulnerability Catalog

Vulnerability in NVIDIA Dynamo for Linux involving a race condition in the LoRA manager singleton initialization. Successful exploitation could lead to denial of service and data tampering.

Vulnerability in NVIDIA Dynamo for Linux involving a race condition in the LoRA manager singleton initialization. Successful exploitation could lead to data tampering and denial of service.

NVIDIA Dynamo for Linux examples and recipes contain a vulnerability where an attacker could cause a system failure. A successful exploit might lead to code execution, data tampering, denial of service, and information disclosure.

NVIDIA Triton Inference Server for Linux contains a vulnerability where a user could cause files outside the model repository to be read, written to, or modified by providing a path in the model name to the Triton MLflow plugin. A successful exploit might lead to denial of service and information disclosure.

A vulnerability allowing a low-privileged user to retrieve report data outside the scope of a shared report link.

NoSQL injection vulnerability in @langchain/langgraph-checkpoint-mongodb up to version 1.3.0. Checkpoint identifiers (thread_id, checkpoint_ns, checkpoint_id) from config.configurable are passed into MongoDB find() queries without type enforcement, allowing an attacker to bypass thread scoping and leak checkpoints across tenants.

A vulnerability in open62541 (function UA_Client_getRemoteDataTypes in examples/custom_datatype/client_types_custom.c) allows use-after-free via manipulation. The attack can be performed locally, and the exploit has been publicly disclosed.

A vulnerability was found in o6 open62541 up to 1.5.5. This issue affects the function UA_Client_readNodeClassAttribute of the file src/client/ua_client_highlevel.c. Performing a manipulation results in heap-based buffer overflow. Attacking locally is a requirement. The exploit has been made public and could be used. The project closed the issue report, stating that this is not the official way to report a security vulnerability.

A vulnerability has been found in NousResearch hermes-agent up to 0.16.0, affecting the browser_snapshot function in tools/browser_tool.py of the Browser Tooling component. This flaw allows server-side request forgery (SSRF) and can be exploited remotely. The exploit has been publicly disclosed.

A flaw has been found in NousResearch hermes-agent up to version 0.16.0 in the function save_url_image of the file agent/image_gen_provider.py of the component xAI Image Generation Provider. This manipulation causes server-side request forgery. The attack may be initiated remotely, and the exploit has been published.

An issue was discovered in Django 5.2 before 5.2.17 and 6.0 before 6.0.8. django.contrib.admin.utils.display_for_field() renders URLField values as clickable links in the admin without validating the URL. A value stored with an unsafe scheme is displayed as a link on changelist and read-only admin pages, which allows cross-site scripting against staff users who click the link. Exploitation requires the unsafe value to already be stored in the database. URLField validation through a ModelForm or the admin rejects unsafe schemes, so this affects applications that persist URLField data without running model validation, for example through direct queryset writes, deserialization, or bulk import of untrusted input.

An issue was discovered in Django 6.0 before 6.0.9 and 5.2 before 5.2.18. GeoDjango's django.contrib.gis.geos.GEOSGeometry is subject to a potential denial-of-service when parsing deeply nested GEOMETRYCOLLECTION objects supplied as well-known text (WKT), well-known binary (WKB), or hex-encoded WKB, which triggers unbounded recursion and a segmentation fault in the underlying GEOS library. Spatial field lookups and the django.contrib.gis.forms.GeometryField form field are also affected.

In Django 5.2 before 5.2.17 and 6.0 before 6.0.8, a denial-of-service vulnerability was discovered in django.utils.translation.check_for_language(), which when given many distinct, very long language codes can lead to excessive memory consumption. These codes can reach the function through the django.views.i18n.set_language() view, which is not routed by default. Memory consumption is bounded by the DATA_UPLOAD_MAX_MEMORY_SIZE setting (default 2.5 MB) and the cache holds a fixed maximum number of entries.

Information disclosure vulnerability occurs when IPSec negotiation fails or is not established properly during NG-eCall SIP signaling.

Information disclosure vulnerability when processing wireless network channel switch information with improperly formatted length fields.

The vulnerability causes memory corruption when processing registry values with incorrect types using a direct query method.

A vulnerability was detected in NousResearch hermes-agent up to 2026.6.5. The issue affects the _check_slash_access function in gateway/run.py of the Quick Command Handler component. Manipulation results in incorrect authorization. The attack can be launched remotely. The exploit is public and may be used. The vendor was contacted but did not respond.

The openUrl function in @a2ui/web_core passes an agent-controlled URL directly to window.open() without validating the URI scheme. A malicious agent can supply a javascript: URI as the url argument of a Button component's functionCall action. When the user clicks the rendered button, arbitrary JavaScript executes in the victim application's browser origin, constituting a stored/reflected XSS with Critical severity.

marimo before 0.23.15 contains a configuration injection vulnerability that allows notebook authors to exfiltrate operator API keys by embedding a malicious base_url in PEP-723 inline script metadata. Due to insufficient sanitization in sanitize_pyproject_dict, the malicious config is merged into session configuration with higher precedence than the operator's settings. When an operator opens the crafted notebook and makes an AI request, marimo resolves the attacker-controlled base_url while using the operator's API key for authentication, transmitting the key to the attacker-controlled endpoint without requiring cell execution.

Perspective 5.0.0 contains a denial of service vulnerability that allows remote attackers to block the server event loop indefinitely by submitting a crafted expression containing unbounded for or while loop constructs in a TableMakeViewReq message. Attackers can embed an arbitrarily large iteration count in an expression column evaluated once per table row, causing the Tornado IOLoop to block without any iteration cap, deadline, or cancellation check, rendering the server unresponsive to all connected clients.