1937 matches found
VIPER-MCP: Detecting and Exploiting Taint-Style Vulnerabilities in Model Context Protocol Servers
Model Context Protocol MCP has emerged as a standard interface for connecting LLM agents to external tools. Because MCP servers expose privileged operations such as shell execution, network access, and file-system manipulation to agent-driven invocation, implementation flaws in tool handlers can...
Trilium Notes 安全漏洞
Trilium Notes is a hierarchical note application developed by Zadam, a personal developer. It focuses on building large personal knowledge bases. Versions of Trilium Notes prior to 0.102.1 contained security vulnerabilities. These vulnerabilities stemmed from Electron configurations that could be...
CVE-2026-39309
CVE-2026-39309 affects Trilium Notes before v0.102.2. The Electron configuration allows a RunAsNode fuse to launch the app in a special Node.js mode (-e) that can execute arbitrary commands with Trilium’s permissions, enabling a local attacker to spoof macOS TCC prompts. An attacker could trigger...
CVE-2026-39309 Trilium Notes: macOS TCC Bypass via Prompt Spoofing
Trilium Notes is a cross-platform, hierarchical note taking application focused on building large personal knowledge bases. In versions 0.102.1 and prior, the Electron configuration is vulnerable to TCC Bypass via Prompt Spoofing, allowing local attackers to trigger misleading macOS permission...
CVE-2026-39309
Trilium Notes is a cross-platform, hierarchical note taking application focused on building large personal knowledge bases. In versions 0.102.1 and prior, the Electron configuration is vulnerable to TCC Bypass via Prompt Spoofing, allowing local attackers to trigger misleading macOS permission...
MAL-2026-4748 Malicious code in eplang (PyPI)
--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 1d53e4571f8ccfc385a265dfd47cbea9793946762a794aff432e98614ee10b21 The package ships epl/.aiconfig.json containing a hardcoded Groq API key with provider set to 'groq'. On any AI-related CLI invocation epl ai, epl ge...
Malicious code in eplang (PyPI)
--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 1d53e4571f8ccfc385a265dfd47cbea9793946762a794aff432e98614ee10b21 The package ships epl/.aiconfig.json containing a hardcoded Groq API key with provider set to 'groq'. On any AI-related CLI invocation epl ai, epl ge...
MAL-2026-4743 Malicious code in buddyme (PyPI)
--- -= Per source details. Do not edit below this line.=- Source: amazon-inspector 6f4ae4b8c00d27e82d54a5d2d960b1dc4f40ba15bc938355bad8421c338d6ef6 buddyme advertises a CLI agent. When installed and run, the default REPL routes every prompt the user types to third-party LLM providers Zhipu GLM at...
CVE-2026-45351
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...
Refusal Evaluation in Coding LLMs and Code Agents: A Systematic Review of Thirteen Malicious-Code Prompt Corpora (2023-2025)
The evaluation of large language model refusal on malicious-coding tasks now spans at least thirteen publicly released prompt corpora AdvBench, the CyberSecEval family, RMCBench, RedCode, MCGMark, JailbreakBench, CySecBench, MalwareBench, CIRCLE, MOCHA, ASTRA, Scam2Prompt / Innoc2Scam-bench, and...
Security Bulletin: Multiple vulnerabilities in IBM MQ Agent images
Summary Multiple vulnerabilities were addressed in IBM MQ Agent images Vulnerability Details CVEID:CVE-2026-41425 DESCRIPTION: Authlib is a Python library which builds OAuth and OpenID Connect servers. Prior to 1.6.11, there is no CSRF protection on the cache feature in...
GHSA-Q5PP-GVJG-H7V4 Microsoft APM: Symlinks under `.apm/prompts/` and `.apm/agents/` are dereferenced during `apm install`, copying host-local file contents into the project tree
Summary Two primitive integrators in apm-cli enumerate package files with bare Path.glob / Path.rglob calls and read each match with Path.readtext, transparently following symbolic links. A symlink committed inside a remote APM dependency under .apm/prompts/.prompt.md or .apm/agents/.agent.md is...
Microsoft APM: Symlinks under `.apm/prompts/` and `.apm/agents/` are dereferenced during `apm install`, copying host-local file contents into the project tree
Summary Two primitive integrators in apm-cli enumerate package files with bare Path.glob / Path.rglob calls and read each match with Path.readtext, transparently following symbolic links. A symlink committed inside a remote APM dependency under .apm/prompts/.prompt.md or .apm/agents/.agent.md is...
Be Kind, Rewrite: Benign Projections Via Rewriting Defend against LLM Data Poisoning Attacks
Large language models LLMs are highly susceptible to backdoor attacks BAs, wherein training samples are poisoned using trigger-based harmful content. Furthermore, existing defenses have proven ineffective when extensively tested across BA patterns. To better combat BAs, we explore the use of LLM...
CVE-2026-45351
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...
CVE-2026-45351 Open WebUI: Exposure of System Prompt to Regular User [Non-Admin]
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...
EUVD-2026-30646
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...
CVE-2026-45351
Open WebUI vulnerability CVE-2026-45351: A non-admin user could trigger a request to /api/models? and receive the system prompt of available models, revealing admin-set backend prompts and compromising confidentiality. This affects Open WebUI self-hosted offline AI platform versions prior to 0.8....
CVE-2026-45351 Open WebUI: Exposure of System Prompt to Regular User [Non-Admin]
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...
CVE-2026-45351 Open WebUI: Exposure of System Prompt to Regular User [Non-Admin]
Open WebUI is a self-hosted artificial intelligence platform designed to operate entirely offline. Prior to 0.8.9, when a regular user non-admin logs into the application, a http://IP:8080/api/models? web request is initiated by the application and in response, it reveals the system prompt of...