TensorFlow vulnerable to `CHECK` fail in `QuantizeAndDequantizeV3`

ImpactIf QuantizeAndDequantizeV3 is given a nonscalar numbits input tensor, it results in a CHECK fail that can be used to trigger a denial of service attack.pythonimport tensorflow as tfsignedinput = Truerangegiven = Falsenarrowrange = Falseaxis = -1input = tf.constant-3.5, shape=1,...

PYSEC-2026-3443

Crawl4AI before 0.8.7 contains a server-side request forgery SSRF vulnerability in the Docker API server's /crawl/job and /llm/job endpoints, which accept webhook URLs without destination validation. An attacker can supply webhook URLs pointing to private or internal IP ranges, Docker networks, o...

Paste Script has improper group memberships permissions

Paste Script 1.7.5 and earlier does not properly set group memberships during execution with root privileges, which might allow remote attackers to bypass intended file-access restrictions by leveraging a web application that uses the local filesystem...

Out of bounds write in TFLite

Impact An attacker can craft a TFLite model that would cause a write outside of bounds of an array in TFLite. In fact, the attacker can override the linked list used by the memory allocator. This can be leveraged for an arbitrary write primitive under certain conditions. PatchesWe have patched th...

`CHECK`-fails due to attempting to build a reference tensor

ImpactA malicious user can cause a denial of service by altering a SavedModel such that Grappler optimizer would attempt to build a tensor using a reference dtype. This would result in a crash due to a CHECK-fail in the Tensor constructor as reference types are not allowed. PatchesWe have patched...

Out-of-bounds Read in OpenCV

An issue was discovered in OpenCV before 3.4.7 and 4.x before 4.1.1 OpenCV-Python before 3.4.7.28 and 4.x before 4.1.1.26. There is an out of bounds read in the function cv::predictOrdered in modules/objdetect/src/cascadedetect.hpp, which leads to denial of service...

Improper Restriction of XML External Entity Reference in Plone

Plone before 5.2.3 allows XXE attacks via a feature that is protected by an unapplied permission of plone.schemaeditor.ManageSchemata therefore, only available to the Manager role...

PYSEC-2026-2218

Mistune is a Python Markdown parser with renderers and plugins. Prior to 3.3.0, the toc plugin and TableOfContents directive generate heading IDs as predictable tocN values without slugifying the heading text, allowing attacker-controlled id="tocN" content to collide with generated anchors and...

Malicious code in ray-mcp-server (PyPI)

Part of the "Hades" wave of the Shai-Hulud supply-chain campaign. On 2026-06-08,malicious phantom releases of ray-mcp-server were published to PyPI using stolencredentials. The package executes a bundled JavaScript payload via the Bunruntime on import that harvests and exfiltrates credentials and...

BentoML has a Path Traversal via Bentofile Configuration

SummaryBentoML's bentofile.yaml configuration allows path traversal attacks through multiple file path fields description, docker.setupscript, docker.dockerfiletemplate, conda.environmentyml. An attacker can craft a malicious bentofile that, when built by a victim, exfiltrates arbitrary files fro...

MobSF has Stored XSS via Manifest Analysis - Dialer Code Host Field

SummaryA Stored Cross-site Scripting XSS vulnerability in MobSF's Android manifest analysis allows an attacker to execute arbitrary JavaScript in the context of a victim's browser session by uploading a malicious APK. The android:host attribute from elements is rendered in HTML reports without...

Wheel Affected by Arbitrary File Permission Modification via Path Traversal in wheel unpack

Summary - Vulnerability Type: Path Traversal CWE-22 leading to Arbitrary File Permission Modification. - Root Cause Component: wheel.cli.unpack.unpack function. - Affected Packages: 1. wheel Upstream source 2. setuptools Downstream, vendors wheel - Severity: High Allows modifying system file...

Moonraker affected by LDAP search filter injection

ImpactInstances of Moonraker configured with the ldap component enabled are vulnerable to LDAP search filter injection techniques via the login endpoint. The 401 error response message can be used to determine whether or not a search was successful, allowing for brute force methods to discover LD...

GuardDog Zip Bomb Vulnerability in safe_extract() Allows DoS

SummaryGuardDog's safeextract function does not validate decompressed file sizes when extracting ZIP archives wheels, eggs, allowing attackers to cause denial of service through zip bombs. A malicious package can consume gigabytes of disk space from a few megabytes of compressed data. Vulnerabili...

Authlib has 1-click Account Takeover vulnerability

Security Advisory: Cache-Backed State Storage CSRF in AuthlibThe Security Labs team at Snyk has reported a security issue affecting Authlib, identified during a recent research project.The Snyk Security Labs team has identified a vulnerability that can result in a one-click account takeover in...

AIOHTTP vulnerable to brute-force leak of internal static file path components

SummaryPath normalization for static files prevents path traversal, but opens up the ability for an attacker to ascertain theexistence of absolute path components. ImpactIf an application uses web.static not recommended for production deployments, it may be possible for an attacker to ascertain t...

Picklescan vulnerable to Arbitrary File Writing

SummaryPicklescan has got open and shutil in its default dangerous blocklist to prevent arbitrary file overwrites. However the module distutils isnt blocked and can be used for the same purpose ie to write arbitrary files. DetailsThis is another vulnerability which impacts the downstream user.By...

Picklescan Bypasses Unsafe Globals Check using pty.spawn

SummaryThe vulnerability allows malicious actors to bypass PickleScan's unsafe globals check, leading to potential arbitrary code execution. The issue stems from the absence of the pty library more specifically, of the pty.spawn function from PickleScan's list of unsafe globals. This vulnerabilit...

Picklescan has Incomplete List of Disallowed Inputs

SummaryCurrently picklescanner only blocks some specific functions of the pydoc and operator modules. Attackers can use other functions within these allowed modules to go through undetected and achieve RCE on the final user. Particularly pydoc.locate: Can dynamically resolve and import arbitrary...

PyMdown Extensions has a ReDOS bug in its Figure Capture extension

ImpactThis issue describes a ReDOS bug found within the figure caption extension pymdownx.blocks.caption .In systems that take unchecked user content, this could cause long hangs when processing the data if a malicious payload was crafted. PatchesThis issue is patched in Release 10.16.1...

Weblate has improper validation upon invitation acceptance

ImpactIt was possible to accept an invitation opened by a different Weblate user. Patches https://github.com/WeblateOrg/weblate/pull/16913 WorkaroundsUsers should avoid leaving Weblate sessions with an unattended opened invitation. ReferencesThanks to Nahid0x for responsibly disclosing this...

Werkzeug safe_join() allows Windows special device names

Werkzeug's safejoin function allows path segments with Windows device names. On Windows, there are special device names such as CON, AUX, etc that are implicitly present and readable in every directory. sendfromdirectory uses safejoin to safely serve files at user-specified paths under a director...

AWS Advanced Python Wrapper: Privilege Escalation in Aurora PostgreSQL instance

Description of Vulnerability:An issue in AWS Wrappers for Amazon Aurora PostgreSQL may allow for privilege escalation to rdssuperuser role. A low privilege authenticated user can create a crafted function that could be executed with permissions of other Amazon Relational Database Service RDS...

Bugsink is vulnerable to unauthenticated remote DoS via crafted Brotli input (via CPU)

ImpactIn affected versions, a specially crafted Brotli-compressed envelope can cause Bugsink to spend excessive CPU time in decompression, leading to denial of service.This can be done if the DSN is known, which it is in many common setups JavaScript, Mobile Apps. PatchesPatched in Bugsink 2.0.6...

vLLM vulnerable to DoS via large Chat Completion or Tokenization requests with specially crafted `chat_template_kwargs`

SummaryThe /v1/chat/completions and /tokenize endpoints allow a chattemplatekwargs request parameter that is used in the code before it is properly validated against the chat template. With the right chattemplatekwargs parameters, it is possible to block processing of the API server for long...

vLLM vulnerable to DoS with incorrect shape of multimodal embedding inputs

SummaryUsers can crash the vLLM engine serving multimodal models by passing multimodal embedding inputs with correct ndim but incorrect shape e.g. hidden dimension is wrong, regardless of whether the model is intended to support such inputs as defined in the Supported Models page.The issue has...

LangChain Vulnerable to Template Injection via Attribute Access in Prompt Templates

ContextA template injection vulnerability exists in LangChain's prompt template system that allows attackers to access Python object internals through template syntax. This vulnerability affects applications that accept untrusted template strings not just template variables in ChatPromptTemplate...

Apache Doris-MCP-Server: Improper Access Control results in bypassing a "read-only" mode

An attacker with a valid read-only account can bypass Doris MCP Server’s read-only mode due to improper access control, allowing modifications that should have been prevented by read-only restrictions.Impact:Bypasses read-only mode; attackers with read-only access may perform unauthorized...

reflex-dev/reflex has an Open Redirect vulnerability

MitigationMake sure GITHUBCODESPACESPORTFORWARDINGDOMAIN is not set in a production environment. So the following is correct:assert os.getenv"GITHUBCODESPACESPORTFORWARDINGDOMAIN" is None Vulnerability Description---Vulnerability Overview - When theGET/auth-codespace page loads in a GitHub...

mkdocs-include-markdown-plugin susceptible to unvalidated input colliding with substitution placeholders

ImpactCWE-20: Improper Input ValidationLow impact PatchesPatched in v7.1.8 commit https://github.com/mondeja/mkdocs-include-markdown-plugin/commit/7466d67aa0de8ffbc427204ad2475fed07678915 WorkaroundsNo...

DataChain Vulnerable to Deserialization of Untrusted Data from Environment Variables

The DataChain library reads serialized objects from environment variables such as DATACHAINMETASTORE and DATACHAINWAREHOUSE in the loader.py module. An attacker with the ability to set these environment variables can trigger code execution when the application loads...

SGLang Remote Code Execution Vulnerability via Unsafe Deserialization in update_weights_from_tensor

A security flaw has been discovered in lmsys sglang 0.4.6. Affected by this vulnerability is the function main of the file /updateweightsfromtensor. The manipulation of the argument serializednamedtensors results in deserialization. The attack can be launched remotely. The exploit has been releas...

Exiv2 has quadratic performance in ICC profile parsing in JpegBase::readMetadata

ImpactA denial-of-service was found in Exiv2 version v0.28.5: a quadratic algorithm in the ICC profile parsing code in jpegBase::readMetadata can cause Exiv2 to run for a long time. Exiv2 is a command-line utility and C++ library for reading, writing, deleting, and modifying the metadata of image...

vLLM has remote code execution vulnerability in the tool call parser for Qwen3-Coder

SummaryAn unsafe deserialization vulnerability allows any authenticated user to execute arbitrary code on the server if they are able to get the model to pass the code as an argument to a tool call. Details vLLM's Qwen3 Coder tool parser contains a code execution path that uses Python's eval...

Copier's safe template has filesystem write access outside destination path

ImpactCopier suggests that it's safe to generate a project from a safe template, i.e. one that doesn't use unsafe features like custom Jinja extensions which would require passing the --UNSAFE,--trust flag. As it turns out, a safe template can currently write files outside the destination path...

h2 allows HTTP Request Smuggling due to illegal characters in headers

SummaryHTTP/2 request splitting vulnerability allows attackers to perform request smuggling attacks by injecting CRLF characters into headers. This occurs when servers downgrade HTTP/2 requests to HTTP/1.1 without properly validating header names/values, enabling attackers to manipulate request...

OpenEXR Heap-Based Buffer Overflow in Deep Scanline Parsing via Forged Unpacked Size

SummaryThe OpenEXRCore code is vulnerable to a heap-based buffer overflow during a write operation when decompressing ZIPS-packed deep scan-line EXR files with a maliciously forged chunk header. DetailsWhen parsing STORAGEDEEPSCANLINE chunks from an EXR file, the following code from...

Skops may allow MethodNode to access unexpected object fields through dot notation, leading to arbitrary code execution at load time

SummaryAn inconsistency in MethodNode can be exploited to access unexpected object fields through dot notation. This can be used to achieve arbitrary code execution at load time.While this issue may seem similar to https://github.com/skops-dev/skops/security/advisories/GHSA-m7f4-hrc6-fwg3, it is...

Assemblyline 4 service client vulnerable to Arbitrary Write through path traversal in Client code

Path-Traversal - Arbitrary File Write in Assemblyline Service ClientIMPORTANT: This vulnerability is valid if you decide to use the assemblyline-service-client outside of the normal practice to using Assemblyline in a production environment. In practice, this code should always be executed within...

pyLoad is vulnerable to attacks that bypass localhost restrictions, enabling the creation of arbitrary packages

SummaryAny unauthenticated attacker can bypass the localhost restrictions posed by the application and utilize this to create arbitrary packages. DetailsAny unauthenticated attacker can bypass the localhost restrictions posed by the application and utilize this to create arbitrary packages. This ...

Langchain-Chatchat vulnerable to path traversal

A vulnerability classified as problematic was found in chatchat-space Langchain-Chatchat up to 0.3.1. This vulnerability affects unknown code of the file /v1/files?purpose=assistants. The manipulation leads to path traversal. The attack can be initiated remotely. The exploit has been disclosed to...

Requests vulnerable to .netrc credentials leak via malicious URLs

ImpactDue to a URL parsing issue, Requests releases prior to 2.32.4 may leak .netrc credentials to third parties for specific maliciously-crafted URLs. WorkaroundsFor older versions of Requests, use of the .netrc file can be disabled with trustenv=False on your Requests Session docs...

Nautobot may allows uploaded media files to be accessible without authentication

ImpactFiles uploaded by users to Nautobot's MEDIAROOT directory, including DeviceType image attachments as well as images attached to a Location, Device, or Rack, are served to users via a URL endpoint that was not enforcing user authentication. As a consequence, such files can be retrieved by...

BackendAI vulnerable to Exposure of Sensitive Information to an Unauthorized Actor

Exposure of sensitive data in active sessions in Lablup's BackendAI allows attackers to retrieve credentials for users on the management platform.NOTE: The maintainers of BackendAI do not consider this report to fit with their threat model and advise users to follow security advice from...

Aim Vulnerable to Sandbox Escape Leading to Remote Code Execution

A vulnerability classified as critical was found in aimhubio aim up to 3.29.1. This vulnerability affects the function RestrictedPythonQuery of the file /aim/storage/query.py of the component runview Object Handler. The manipulation of the argument Query leads to sandbox issue. The attack can be...

PYSEC-2026-2092

An issue was discovered in Django 6.0 before 6.0.7 and 5.2 before 5.2.16.DomainNameValidator does not prohibit newlines in domain names unless used via a form field, since CharField strips newlines. If an application uses values with newlines in an HTTP response, header injection can occur. Djang...

Frappe vulnerable to information disclosure leading to account takeover

ImpactMaking crafted requests could lead to information disclosure that could further lead to account takeover. WorkaroundsThere's no workaround to fix this without upgrading. CreditsThanks to Thanh of Calif.io for reporting the issue...

Aim Uncontrolled Resource Consumption vulnerability

In version 3.25.0 of aimhubio/aim, the tracking server is vulnerable to a denial of service attack. The server overrides the maximum size for websocket messages, allowing very large images to be tracked. This causes the server to become unresponsive to other requests while processing the large...

Frappe has possibility of SQL injection due to improper validations

ImpactSQL injection could be achieved via a specially crafted request, which could allow malicious person to gain access to sensitive information. WorkaroundsUpgrading is required, no other workaround is present...

Open WebUI lacks authentication for the `api/v1/utils/pdf` endpoint

In version v0.3.10 of open-webui/open-webui, the api/v1/utils/pdf endpoint lacks authentication mechanisms, allowing unauthenticated attackers to access the PDF generation service. This vulnerability can be exploited by sending a POST request with an excessively large payload, potentially leading...