CVE-2026-25481 Langroid has WAF Bypass Leading to RCE in TableChatAgent

Langroid is a framework for building large-language-model-powered applications. Prior to version 0.59.32, there is a bypass to the fix for CVE-2025-46724. TableChatAgent can call pandaseval tool to evaluate the expression. There is a WAF in langroid/utils/pandasutils.py introduced to block code...

Agent-Zero 0.8.0 - 0.9.4 - Arbitrary File Download

Agent-Zero v0.8.0 - 0.9.4 contains a path traversal caused by improper validation in /api/downloadworkdirfile.py, letting attackers access unauthorized files, exploit requires crafted request. id: CVE-2025-55523 info: name: Agent-Zero 0.8.0 - 0.9.4 - Arbitrary File Download author: 0xAkoko...

Jenkins Sidepanel - Unauthorized Agent/Queue Exposure

Jenkins 2.527 and earlier, LTS 2.516.2 and earlier does not perform a permission check in the sidepanel of a page intentionally accessible to users lacking Overall/Read permission, allowing attackers without Overall/Read permission to list agent names through its sidepanel executors widget. id:...

Langroid 代码注入漏洞

Langroid is an open-source tool developed using multi-agent programming for LLM tasks. Versions of Langroid prior to 0.59.32 had a code injection vulnerability. This vulnerability stemmed from a bypass in the TableChatAgent’s invocation of the pandaseval tool, which could allow arbitrary code to ...

📄 NCR Command Center Agent 16.3 Remote Command Execution

Proof of concept exploit for a remote command execution vulnerability in NCR Command Center Agent version 16.3 on Aloha POS/BOH servers. The vulnerability allows remote, unauthenticated attackers to execute arbitrary commands with SYSTEM privileges by sending a specially crafted XML document to...

Oracle Enterprise Manager Cloud Control (January 2026 CPU)

The 13.5 and 24.1 versions of Enterprise Manager Base Platform installed on the remote host are affected by multiple vulnerabilities as referenced in the January 2026 CPU advisory. - Vulnerability in the Oracle Enterprise Manager Base Platform product of Oracle Enterprise Manager component: Agent...

Bypassing AI Control Protocols Via Agent-As-A-Proxy Attacks

As AI agents automate critical workloads, they remain vulnerable to indirect prompt injection IPI attacks. Current defenses rely on monitoring protocols that jointly evaluate an agent's Chain-of-Thought CoT and tool-use actions to ensure alignment with user intent. We demonstrate that these...

Exploit for CVE-2026-25130

CVE-2026-25130 – Cybersecurity AI CAI Framework Argument Inj...

Meet Moltbook, the Social Platform Where AI Agents Talk and Humans Watch

Moltbook is a new social platform where AI agents post and interact while humans observe, raising questions about autonomy, security, and agent behavior...

CVE-2026-1664

Summary: CVE-2026-1664 affects Cloudflare Agents SDK prior to 0.3.7, due to an IDOR in header-based email routing. Root cause: createHeaderBasedEmailResolver() parses Message-ID and References to derive target agentName/agentId without cryptographic/origin verification, letting external headers s...

Security Bulletin: Vulnerablity in Apache Log4j may affect IBM APM Internet Service Monitoring Agent

Summary There is a vulnerability in the Apache log4j library used by IBM APM Internet Service Monitoring Agent. Vulnerability Details CVEID:CVE-2025-68161 DESCRIPTION: The Socket Appender in Apache Log4j Core versions 2.0-beta9 through 2.25.2 does not perform TLS hostname verification of the peer...

PT-2026-5953

Name of the Vulnerable Software and Affected Versions CyberArk Endpoint Privilege Manager Agent versions through 25.10.0 Description A local user can elevate privileges through policy elevation of an Administration task. This allows for privilege escalation. Recommendations Update CyberArk Endpoi...

This Week in Spring - February 3rd, 2026

Hi, Spring fans! This week I'm in northern Europe. I went on the Vaadin cruise from Finland to Sweden, gave a talk on a boat, then arrived in Stockholm in time for the amazing JFokus 2026 event where I had the privilege yesterday of doing a deep dive with my pal James Ward on Spring AI and agenti...

GO-2026-4384 KubeVirt Guest Agent DoS via Excessive Network Interface Reports in kubevirt.io/kubevirt

KubeVirt Guest Agent DoS via Excessive Network Interface Reports in kubevirt.io/kubevirt...

GHSA-X34R-63HX-W57F Langroid has WAF Bypass Leading to RCE in TableChatAgent

Affected Scope langroid = 0.59.31 Vulnerability Description CVE-2025-46724 fix bypass: TableChatAgent can call pandaseval tool to evaluate the expression. There is a WAF in langroid/utils/pandasutils.py introduced to block code injection CVE-2025-46724. However it can be bypassed due to literalok...

Langroid has WAF Bypass Leading to RCE in TableChatAgent

Affected Scope langroid = 0.59.31 Vulnerability Description CVE-2025-46724 fix bypass: TableChatAgent can call pandaseval tool to evaluate the expression. There is a WAF in langroid/utils/pandasutils.py introduced to block code injection CVE-2025-46724. However it can be bypassed due to literalok...

The Trigger in the Haystack: Extracting and Reconstructing LLM Backdoor Triggers

Detecting whether a model has been poisoned is a longstanding problem in AI security. In this work, we present a practical scanner for identifying sleeper agent-style backdoors in causal language models. Our approach relies on two key findings: first, sleeper agents tend to memorize poisoning dat...

Co-RedTeam: Orchestrated Security Discovery and Exploitation with LLM Agents

Large language models LLMs have shown promise in assisting cybersecurity tasks, yet existing approaches struggle with automatic vulnerability discovery and exploitation due to limited interaction, weak execution grounding, and a lack of experience reuse. We propose Co-RedTeam, a security-aware...

CVE-2020-37047

Deep Instinct Windows Agent 1.2.29.0 contains an unquoted service path vulnerability in the DeepMgmtService that allows local users to potentially execute code with elevated privileges. Attackers can exploit the unquoted path in C:\Program Files\HP Sure Sense\DeepMgmtService.exe to inject malicio...

CVE-2020-37047

The CVE-2020-37047 case concerns Deep Instinct Windows Agent 1.2.29.0, where the DeepMgmtService has an unquoted service path at C:\Program Files\HP Sure Sense\DeepMgmtService.exe. The underlying issue is a local privilege escalation vulnerability that can allow a local user to execute code with ...