CVE-2026-69147

A flaw was found in vLLM, an inference and serving engine for large language models. An attacker can exploit this by submitting specially crafted video requests that force the use of the PyNvVideoCodec GPU decoder. This bypasses the engine's static GPU memory reservation, allowing the attacker to...

CVE-2026-57173

vLLM is an inference and serving engine for large language models. Prior to 0.24.0, the inputaudio handling path for /v1/chat/completions calls AudioMediaIO.loadbytes or AudioMediaIO.loadfile without passing VLLMMAXAUDIODECODEDURATIONS to the shared audio decoder. An unauthenticated client can...

AIJon: Automated Generation of Annotations for Fuzzing

Modern fuzzers use code coverage as feedback to guide their exploration which has proven to be an effective strategy for driving exploration. However, this strategy overlooks inputs that may be interesting to the target program even without uncovering new code paths. Fortunately, prior research h...

Measuring and Exploiting Implicit Trust in LLM Tool-Calling Pipelines

The Model Context Protocol MCP enables LLMs to invoke external tools, but every tool interaction exposes the model to attacker-controlled text through multiple input channels tool descriptions, tool results, sampling messages that share a single context window without privilege separation. In thi...

CASHEWS: Source Preprocessor for LLM-Based Malicious Package Detection

Malicious npm package detection tools now leverage LLMs' semantic understanding of source code to detect malicious intent at scale. This capability has proven invaluable in identifying packages involved in recent supply-chain attacks such as Shai-Hulud. However, threat actors exploit the limited...

MiST: Mid-Training LLMs for Cybersecurity

Cybersecurity combines high-stakes analysis with complex technical language, making it an impactful and challenging domain for LLMs. We present MiST Mid-trained Security Transformer, a suite of 8B and 32B models that achieve strong performance on public cybersecurity benchmarks. We use mid-traini...

reasongate v0.4.0

ReasonGate A self-hostable gate that inspects the text going into and out of an LLM and returns an explainable allow / flag / block decision with a machine-readable audit record for every call. What this is The open-source core is rule-based. It does four things: recognizes known prompt-injection...

IDAssist v2.4.0

IDAssist AI-Powered Reverse Engineering Plugin for IDA Pro Author: Jason Tang Description IDAssist is an IDA Pro plugin that integrates LLM-powered analysis directly into IDA's interface, providing AI-assisted binary reverse engineering through configurable LLM providers, semantic knowledge graph...

CVE-2026-54561 MCP Memory Keeper: Arbitrary local file read in mcp-memory-keeper context_import via unvalidated filePath

MCP Memory Keeper is an MCP server for persistent context management in AI coding assistants. Prior to 0.13.0, contextimport in src/index.ts passes the caller-controlled filePath directly to fs.readFileSync without restricting the path to an export directory. An MCP client, including an LLM agent...

cowrie v3.0.14

.. SPDX-FileCopyrightText: 2014 Upi Tamminen [email protected] .. SPDX-FileCopyrightText: 2014-2025 Michel Oosterhof [email protected] .. .. SPDX-License-Identifier: BSD-3-Clause Cowrie What is Cowrie Cowrie is a medium to high interaction SSH and Telnet honeypot designed to log brute...

Evaluating the Impact of Personalization in Conversational Cybersecurity Assistants

Users increasingly turn to Large Language Models to answer a variety of questions, including cybersecurity questions. We study how personalization strategies can help improve the effectiveness of answers to questions asked to an LLM-based cybersecurity assistant. Beyond accuracy, we focus on the...

Toward Secure AI-Powered Penetration Testing Agents: Security Threats, Guardrails, and Architectural Perspectives

LLM-powered autonomous agents are transforming the penetration testing space with dynamic, multi-step offensive security workflows that require minimal supervision by humans. These agents leverage sophisticated reasoning abilities and external security tools to independently carry out...

vader

VADER：用于漏洞评估、检测、解释和修复的人工评估基准 官方 GitHub 仓库：https://github.com/AfterQuery/vader Hugging Face 数据集：https://huggingface.co/datasets/AfterQuery/vader VADER 是一个人工评估基准 ，旨在衡量大型语言模型（LLMs）处理真实世界软件漏洞的能力。它包含 174 个真实世界漏洞案例 （从开源仓库中精选），涵盖四项任务： 漏洞识别与分类（CWE） 根因解释 补丁（修复） 测试计划生成 这些案例涵盖 15+ 编程语言 （例如...

CVE-2026-57145

PraisonAI is a multi-agent teams system. Prior to 4.6.62, src/praisonai/praisonai/tools/multiedit.py passes the LLM-controlled filepath parameter directly to open for reading and writing without traversal rejection, symlink resolution, a workspace boundary, or protected-path checks...

CVE-2026-57130

PraisonAI is a multi-agent teams system. Prior to praisonaiagents 1.6.59, src/praisonai-agents/praisonaiagents/tools/emailtools.py interpolates LLM-controlled fromaddr, subject, and query values directly into quoted IMAP SEARCH criteria. Embedded quote, backslash, newline, or null characters can...

CVE-2026-57145 PraisonAI: Arbitrary File Read/Write via `multiedit` Tool Without Path Validation

PraisonAI is a multi-agent teams system. Prior to 4.6.62, src/praisonai/praisonai/tools/multiedit.py passes the LLM-controlled filepath parameter directly to open for reading and writing without traversal rejection, symlink resolution, a workspace boundary, or protected-path checks...

EUVD-2026-77595

PraisonAI is a multi-agent teams system. Prior to 4.6.62, src/praisonai/praisonai/tools/multiedit.py passes the LLM-controlled filepath parameter directly to open for reading and writing without traversal rejection, symlink resolution, a workspace boundary, or protected-path checks...

CVE-2026-57145

CVE-2026-57145 affects PraisonAI , an AI multi-agent framework. The multiedit tool in src/praisonai/praisonai/tools/multiedit.py passes the LLM-controlled filepath parameter directly to open() for reading and writing without path traversal rejection, symlink resolution, workspace boundary validat...

CVE-2026-57130 PraisonAI: IMAP Command Injection via Unsanitized Email Search Parameters

PraisonAI is a multi-agent teams system. Prior to praisonaiagents 1.6.59, src/praisonai-agents/praisonaiagents/tools/emailtools.py interpolates LLM-controlled fromaddr, subject, and query values directly into quoted IMAP SEARCH criteria. Embedded quote, backslash, newline, or null characters can...

CVE-2026-57130

PraisonAI (praisonaiagents) is a multi-agent framework. Prior to 1.6.59 , the file email_tools.py interpolates LLM-controlled parameters (from_addr, subject, query, search_id) directly into IMAP SEARCH criteria using f-string formatting with double-quote delimiters, with no escaping or sanitizati...