Articles
Daily digests summarizing the vulnerabilities and incidents tracked by VulnWatch.
-
· 8 entries · daily · Subscribers only
VulnWatch Digest: Agentic RCE in Zammad and LiteLLM Tenant Isolation Bypass
Critical vulnerabilities emerge in AI agent frameworks and LLM gateways, including RCE in Zammad and tenant data leakage in LiteLLM. Immediate patching and configuration reviews are required for affected platforms.
-
· 14 entries · daily · Subscribers only
VulnWatch Daily: Critical Prompt Injection in Decepticon & MCP Server Flaws
Today's digest highlights a CVSS 10.0 prompt injection in Decepticon allowing role-boundary forgery, alongside critical DNS rebinding and auth-bypass flaws in DBHub and Cline MCP servers.
-
· 9 entries · daily · Subscribers only
VulnWatch Digest: Critical RCE in MLflow, Unauth Robot App Installs, and RAG Poisoning
Today's digest highlights critical remote code execution flaws in MLflow's model loading, an unauthenticated app installation vector in Reachy Mini robots, and emerging RAG poisoning risks in financial AI agents.
-
· 59 entries · daily · Subscribers only
VulnWatch Digest: Critical MCP Atlassian Flaws & SGLang RCE
A massive wave of vulnerabilities hits the Model Context Protocol ecosystem, led by critical auth bypasses in MCP Atlassian and unauthenticated RCE in SGLang. Immediate patching required for agentic infrastructure.
-
· 16 entries · daily · Subscribers only
vLLM DoS Wave, MCP RCE Risks, and Agent Auth Flaws Dominate September 21 AI Security Landscape
A critical cluster of DoS vulnerabilities in vLLM threatens inference availability, while new RCE and SSRF flaws in MCP servers highlight escalating risks in agentic workflows. Immediate patching and configuration audits are required.
-
· 120 entries · weekly
VulnWatch Weekly: MCP Servers Under Siege, LLM RCE via Pickle
This week's digest highlights a critical wave of vulnerabilities in Model Context Protocol (MCP) servers, featuring unauthenticated RCEs, SSRF chains, and token theft. We also cover severe pickle deserialization flaws in LLM serving stacks like LMDeploy and vLLM.
-
· 2 entries · daily · Subscribers only
VulnWatch Digest: Agent Auth Bypasses and RL Deserialization Risks
Today's briefing covers a critical authentication bypass in Tencent's BrowserSkill affecting AI agents and a persistent deserialization flaw in stable-baselines3 that enables remote code execution.
-
· 3 entries · daily · Subscribers only
VulnWatch Digest: RCE in Mistral, MCP Token Leaks, and vLLM Memory Corruption
Today's briefing covers a critical RCE in Mistral Vibe via git hooks, plaintext logging of MCP tokens in OpenPanel, and a low-severity memory corruption bug in vLLM affecting concurrent request integrity.
-
· 17 entries · daily · Subscribers only
VulnWatch Daily: Critical RCE in LMDeploy, MCP Lateral Movement, and vLLM DoS
Today's digest highlights critical remote code execution flaws in LMDeploy's serving stack, dangerous network isolation gaps in ToolHive MCP servers, and multiple denial-of-service vectors in vLLM and SGLang engines.
-
· 27 entries · daily · Subscribers only
VulnWatch Digest: Critical MCP Auth Bypasses and LLM Infra DoS Risks
Today's digest highlights critical authentication flaws in Model Context Protocol servers, command injection in M365 Copilot, and resource exhaustion vectors in vLLM and SGLang disaggregated deployments.
-
· 20 entries · daily · Subscribers only
VulnWatch Daily: Critical RCE in LMDeploy, MCP Auth Bypasses, and vLLM DoS
Today's digest highlights a critical pickle deserialization RCE in LMDeploy, unauthenticated file reads in MCP servers leading to token theft, and multiple denial-of-service vectors in vLLM and Chroma affecting multi-tenant isolation.
-
· 38 entries · daily · Subscribers only
Critical MCP Supply Chain Crisis: RCE, SSRF, and Auth Bypasses Dominate September 15 Digest
A wave of critical vulnerabilities in Model Context Protocol (MCP) servers exposes AI agents to remote code execution, credential theft, and DNS rebinding attacks. Immediate patching of GitLab, MySQL, and Context Forge integrations is required.
-
· 36 entries · daily · Subscribers only
VulnWatch Digest: Critical Storm RCEs, Langflow SSRF, and Agentic Path Traversal Risks
A massive cluster of vulnerabilities hits Apache Storm with critical RCE and auth bypass flaws. IBM Langflow, PraisonAI, and MCP servers face severe SSRF and path traversal risks enabling credential theft and data exfiltration.
-
· 89 entries · weekly
VulnWatch Weekly: The Agentic Apocalypse & The Langflow Meltdown
This week marks a turning point for AI security as critical RCEs plague agentic frameworks like Mistral Vibe and Langflow. With 89 new entries, the dominant theme is the collapse of trust boundaries in autonomous agents and model serving pipelines.
-
· 1 entry · daily · Subscribers only
CrewAI Sandbox Bypass: The Peril of Python Import Blocklists
A new high-severity flaw in CrewAI exposes the fragility of import-level sandboxing, allowing attackers to bypass restrictions via the Python object graph and ctypes.