autoguardrails

autoguardrails Open source by Santander AI Lab. An LLM / AI-safety guardrail research library / evaluation harness autoresearch-style: it searches over a single mutable policy.md surface to minimize attack success rate ASR against a fixed evaluation suite, with a benign-pass floor. Part of...

CVE-2018-7747

CVE-2018-7747 CalderaForms 1.5.9.1 XSS WordPress plugin - tutorial CalderaForm is a WordPress plugin that makes it easy to create forms via drag and drop. During a recent engagement I happened to test some portals, one of which hosted a contact form created with this plugin. The custom...

Libssh-server-CVE-2018-10933

libssh Authentication Bypass VulnerabilityCVE-2018-10933 中文版本Chinese version libssh is a multiplatform C library implementing the SSHv2 protocol on client and server side. A logic vulnerability was found in libssh's server-side state machine. The attacker can send the MSGUSERAUTHSUCCESS message...

CVE-2026-94271 Deema Payment Gateway <= 1.1.2 - Unauthenticated Payment Confirmation Forgery via Unverified Success Return

The Deema Payment Gateway WordPress plugin through 1.1.2 does not verify the payment with the payment provider when handling the return from the hosted checkout, and does not check the payment status or amount, allowing unauthenticated users to have orders marked as paid without any payment being...

CVE-2026-94271 Deema Payment Gateway <= 1.1.2 - Unauthenticated Payment Confirmation Forgery via Unverified Success Return

The Deema Payment Gateway WordPress plugin through 1.1.2 does not verify the payment with the payment provider when handling the return from the hosted checkout, and does not check the payment status or amount, allowing unauthenticated users to have orders marked as paid without any payment being...

Where Rules End and Judges Begin: Measuring the Judgment Boundary in Multi-Agent Systems Security

LLM-based multi-agent systems MAS engage tools, share memory, and delegate tasks, often encountering adversarial content. Current defenses for MAS are typically evaluated in isolation, focusing on one attack type at a time, which can lead to costly and hard-to-audit outcomes. This study organizes...

Red-TTT: Test-Time Training for Automated Jailbreaking Large Language Models

Large language models remain vulnerable to jailbreaks, and automated red teaming is the standard way to find jailbreaks in large language models at scale. Current methods either draw more samples at test time through search, rewriting, and tree expansion, or train a stronger attacker offline with...

Threat-Preserving Representation Sensitivity in Agent-Security Benchmarks

Security benchmarks for LLM-based agents often report the attack success rate ASR as a measure of model robustness and use these scores to compare different models and defense mechanisms, assuming that they describe the security of the agent. In this paper, we explore whether it also influences t...

PT-2026-104078

Name of the Vulnerable Software and Affected Versions Super Forms – Drag & Drop Form Builder versions prior to 6.3.317 Description An issue exists in the Register & Login add-on where the before email success msg function, during the register login action='update' flow, trusts a user-supplied use...

PT-2026-103714

The Super Forms – Drag & Drop Form Builder plugin for WordPress is vulnerable to Privilege Escalation in all versions up to, and including, 6.3.316. This is due to the Register & Login add-on's before email success msg function whitelisting the client-submitted 'role' key and copying it into the...

AlmaLinux 9.2 [TuxCare] Security Update: wpa_supplicant (ALMALINUX9.2:CLSA-2025:1753802388)

The AlmaLinux 9.2 host has a package installed that is affected by a vulnerability as referenced in the TuxCare ALMALINUX9.2:CLSA-2025:1753802388 advisory. - The implementation of PEAP in wpasupplicant through 2.10 allows authentication bypass. For a successful attack, wpasupplicant must be...

CVE-2026-97972

In the Linux kernel, the following vulnerability has been resolved: net: macb: put the "mdio" child node reference on success macbmiiinit holds the reference returned by ofgetchildbyname for macbmdiobusregister and drops it only on the error paths, so every successful probe leaks a node reference...

CVE-2026-94216 ST Engineering iDirect Evolution/Velocity WebServer Evolution HTTP Request authorize response splitting

A vulnerability was determined in ST Engineering iDirect Evolution and Velocity WebServer Evolution. This vulnerability affects unknown code of the file /authorize of the component HTTP Request Handler. Executing a manipulation of the argument Success can lead to http response splitting. It is...

EUVD-2026-83907

A vulnerability was determined in ST Engineering iDirect Evolution and Velocity WebServer Evolution up to 20260717. This vulnerability affects the function authorize of the file /usr/sbin/webserver of the component HTTP Header Handler. Executing a manipulation of the argument Success can lead to...

CVE-2026-94216 ST Engineering iDirect Evolution/Velocity WebServer Evolution HTTP Request authorize response splitting

A vulnerability was determined in ST Engineering iDirect Evolution and Velocity WebServer Evolution. This vulnerability affects unknown code of the file /authorize of the component HTTP Request Handler. Executing a manipulation of the argument Success can lead to http response splitting. It is...

Attack Success Rate Is Not a Number: On Measurement Validity in Agentic AI Security Evaluation

Attack success rate ASR is the headline metric in nearly every published evaluation of attacks on, and defenses for, LLM agents. We argue that ASR as currently used is not a single quantity but a family of metrics parameterized by six design choices that papers seldom specify and never hold...

Rethinking Backdoor Repair Evaluation: Distinguishing Aggregate Clean Utility from Benign Performance Preservation

Backdoor repair aims to suppress malicious behavior in compromised models while preserving benign task performance. Existing studies typically evaluate these objectives using Attack Success Rate ASR and Overall Clean Accuracy, but aggregate clean accuracy can obscure substantial degradation...

CVE-2026-93156

In the Linux kernel, the following vulnerability has been resolved: crypto: rk3288 - fail ahash requests on HASH idle timeout rkhashrun waits for RKCRYPTOHASHSTS to become idle after the final DMA transfer, but ignores the poll result. If the hash engine never becomes idle, the driver still reads...

EUVD-2026-82295

In the Linux kernel, the following vulnerability has been resolved: ASoC: SOF: ipc4-topology: Return error for invalid number of formats When the number of input or output formats is zero, sofipc4widgetsetupcompsrc and sofipc4widgetsetupcompasrc print an error and jump to the cleanup label. At th...

CVE-2026-93187

CVE-2026-93187 affects the Linux kernel in the ASoC SOF ipc4-topology driver. When the number of input or output audio formats is zero, the functions sof_ipc4_widget_setup_comp_src() and sof_ipc4_widget_setup_comp_asrc() print an error and jump to the cleanup label, but the return variable ret re...