Exploit Tool Invocation Prompt for Tool Behavior Hijacking in LLM-Based Agentic System

LLM-based agentic systems leverage large language models to handle user queries, make decisions, and execute external tools for complex tasks across domains like chatbots, customer service, and software engineering. A critical component of these systems is the Tool Invocation Prompt TIP, which...

CVE-2025-58618

Improper Neutralization of Input During Web Page Generation 'Cross-site Scripting' vulnerability in Jonathan Jernigan Pie Calendar pie-calendar allows DOM-Based XSS.This issue affects Pie Calendar: from n/a through = 1.2.8...

CVE-2025-58623

Improper Neutralization of Input During Web Page Generation 'Cross-site Scripting' vulnerability in Bohemia Plugins Event Feed for Eventbrite event-feed-for-eventbrite allows DOM-Based XSS.This issue affects Event Feed for Eventbrite: from n/a through = 1.3.2...

CVE-2025-58834

Improper Neutralization of Input During Web Page Generation 'Cross-site Scripting' vulnerability in gugu short.io wp-shortcm allows DOM-Based XSS.This issue affects short.io: from n/a through = 2.4.2...

CVE-2025-58786

Improper Neutralization of Input During Web Page Generation 'Cross-site Scripting' vulnerability in VW THEMES Ibtana – Ecommerce Product Addons ibtana-ecommerce-product-addons allows DOM-Based XSS.This issue affects Ibtana – Ecommerce Product Addons: from n/a through = 0.4.7.6...

Linux Distros Unpatched Vulnerability : CVE-2024-34008

The Linux/Unix host has one or more packages installed that are impacted by a vulnerability without a vendor supplied patch available. - Actions in the admin management of analytics models did not include the necessary token to prevent a CSRF risk. CVE-2024-34008 Note that Nessus relies on the...

Mind the Gap: Evaluating Model- and Agentic-Level Vulnerabilities in LLMs with Action Graphs

As large language models transition to agentic systems, current safety evaluation frameworks face critical gaps in assessing deployment-specific risks. We introduce AgentSeer, an observability-based evaluation framework that decomposes agentic executions into granular action and component graphs,...

Model Namespace Reuse Flaw Hijacks AI Models on Google and Microsoft Platforms

A new security vulnerability called 'Model Namespace Reuse' allows attackers to hijack AI models on Google, Microsoft, and…...

CVE-2025-6785

Securing externally available CAN wires can easily allow physical access to the CAN bus, allowing possible injection of specially formed CAN messages to control remote start functions of the vehicle. Testing completed on Tesla Model 3 vehicles with software version v11.1 2023.20.9...

CVE-2025-6785 Tesla Model 3 Physical CAN Bus Injection

Securing externally available CAN wires can easily allow physical access to the CAN bus, allowing possible injection of specially formed CAN messages to control remote start functions of the vehicle. Testing completed on Tesla Model 3 vehicles with software version v11.1 2023.20.9...

CVE-2025-6785 Tesla Model 3 Physical CAN Bus Injection

Securing externally available CAN wires can easily allow physical access to the CAN bus, allowing possible injection of specially formed CAN messages to control remote start functions of the vehicle. Testing completed on Tesla Model 3 vehicles with software version v11.1 2023.20.9...

CVE-2025-6785

CVE-2025-6785 affects Tesla Model 3 software versions from 2023.Xx before 2023.44. The vulnerability stems from insecure handling of externally accessible CAN wiring, which can grant physical access to the CAN bus and enable injection of specially formed CAN messages to control remote start funct...

CISA Flags TP-Link Router Flaws CVE-2023-50224 and CVE-2025-9377 as Actively Exploited

The U.S. Cybersecurity and Infrastructure Security Agency CISA on Wednesday added two security flaws impacting TP-Link wireless routers to its Known Exploited Vulnerabilities KEV catalog, noting that there is evidence of them being exploited in the wild. The vulnerabilities in question are listed...

Keras safe_mode bypass allows arbitrary code execution when loading a malicious model.

...

A vulnerability has been identified in Node.js, affecting users of the experimental permission model when the --allow-fs-write flag is used. Node.js Permission Model do not operate on file descriptors, however, operations such as fs.fchown or fs.fchmod can use a "read-only" file descriptor to change the owner and permissions of a file.

...

A vulnerability has been identified in Node.js, affecting users of the experimental permission model when the --allow-fs-read flag is used. This flaw arises from an inadequate permission model that fails to restrict file stats through the fs.lstat API. As a result, malicious actors can retrieve stats from files that they do not have explicit read access to. This vulnerability affects all users using the experimental permission model in Node.js 20 and Node.js 21. Please note that at the time this CVE was issued, the permission model is an experimental feature of Node.js.

...

Arbitrary Command Injection

Overview mcp-markdownify-server is a Model Context Protocol MCP server that converts various file types and web content to Markdown format. It provides a set of tools to transform PDFs, images, audio files, web pages, and more into easily readable and shareable Markdown text. Affected versions of...

PT-2025-35942

Name of the Vulnerable Software and Affected Versions: Tesla Model 3 versions 2023.Xx through 2023.43 Description: Securing externally available CAN wires can easily allow physical access to the CAN bus, potentially enabling the injection of specially formed CAN messages to control remote start...

Tesla Model 3 安全漏洞

The Tesla Model 3 is an electric vehicle from Tesla, an American company. A security vulnerability exists in Tesla Model 3 versions prior to 2023.44 that stems from improper physical access control of the CAN bus, which could lead to the remote start function being controlled...

Adversarial Bug Reports As a Security Risk in Language Model-Based Automated Program Repair

Large Language Model LLM - based Automated Program Repair APR systems are increasingly integrated into modern software development workflows, offering automated patches in response to natural language bug reports. However, this reliance on untrusted user input introduces a novel and underexplored...