Articles
Daily digests summarizing the vulnerabilities and incidents tracked by VulnWatch.
-
· 36 entries · daily · Subscribers only
VulnWatch Daily: Critical RCE Surges in AI Agent Sandboxes and Gateways
36 vulnerabilities disclosed today highlight severe RCE risks in FastGPT, LiteLLM, and Open WebUI. Immediate patching recommended for agent platforms and model gateways to prevent compromise.
-
· 15 entries · daily · Subscribers only
VulnWatch Daily: Supply Chain Compromises and Agentic SSRF Risks
Critical supply chain compromises hit PyTorch Lightning while new SSRF vectors emerge in MCP servers. Plus, sandbox escapes in vm2 and widespread Copilot injection flaws.
-
· 11 entries · daily · Subscribers only
VulnWatch Daily: Critical Agent RCE and Multi-Tenant Isolation Failures
Eleven new vulnerabilities impact AI infrastructure today, including critical RCE in PraisonAI and multi-tenant isolation bugs in Axonflow. Immediate patching is recommended for model serving and agent frameworks.
-
· 19 entries · daily · Subscribers only
VulnWatch Daily: Critical MCP Server Flaws and Agent RCE Risks Surge
Today's digest highlights critical path traversal and prompt injection flaws in MCP servers and AI agents, including Langflow and SQLBot. Immediate patching is recommended for exposed instances.
-
· 30 entries · weekly
Langflow Floods, MCP Risks, and Ollama Windows RCE
This week saw a cascade of vulnerabilities in IBM Langflow, critical RCE risks in Ollama for Windows, and emerging threats in the Model Context Protocol ecosystem. Security teams must prioritize patching agentic frameworks and securing local AI deployments.
-
· 12 entries · daily · Subscribers only
VulnWatch Daily: Critical RCE in Agentic Frameworks and Ollama Memory Leaks
Today's digest highlights critical command injection flaws in Evolver and PPTAgent, alongside high-severity memory leaks in Ollama. Platform teams should prioritize patching agent runtimes and reviewing file upload policies.
-
· 2 entries · daily · Subscribers only
Langflow RCE Flaws Threaten AI Orchestration Integrity
Two remote code injection vulnerabilities in Langflow 1.8.4 expose AI pipelines to manipulation. Public exploits available; vendor unresponsive. Patch immediately.
-
· 4 entries · daily · Subscribers only
VulnWatch Daily: SGLang Deserialization and MCP Server Auth Flaws
Critical vulnerabilities detected in SGLang tokenizers and NextChat MCP integrations. Security teams must audit model serving paths and tool authorization policies immediately.
-
· 3 entries · daily · Subscribers only
VulnWatch Digest: AI Memory & Skill Vulnerabilities (May 1, 2026)
Today's digest covers pickle deserialization in mem0ai and code injection/XSS in Claude skills. Despite low CVSS scores, immediate mitigation is advised due to RCE potential.
-
· 8 entries · daily · Subscribers only
IBM Langflow Critical Cluster: RCE, Data Leaks, and SSRF Risks
Seven high-severity vulnerabilities impact IBM Langflow Desktop and OSS versions, enabling RCE and data theft. Immediate patching is required for AI orchestration layers.
-
· 2 entries · daily · Subscribers only
Critical Update Flaws in Ollama for Windows Enable Silent RCE
Two high-severity vulnerabilities in Ollama's Windows update mechanism allow attackers to bypass signature checks and write arbitrary files, enabling silent remote code execution on developer workstations.
-
· 3 entries · daily · Subscribers only
MCP Bridge Vulnerabilities Surge: Path Traversal and Command Injection Risks in AI Agent Tooling
Today's digest highlights critical risks in Model Context Protocol (MCP) implementations. Three new CVEs expose path traversal and command injection vulnerabilities in Claude agent bridges and SDKs, urging immediate review of agent tooling configurations.
-
· 5 entries · daily · Subscribers only
VulnWatch Daily: SSRF Waves Hit AI Gateways and MCP Servers
Critical SSRF flaws plague NextChat and MCP studios while aider faces command injection risks. vllm stability also compromised.
-
· 3 entries · daily · Subscribers only
VulnWatch Daily: MCP RCE, Ollama Path Traversal, and LLM Data Leaks
Today's digest covers critical supply-chain risks in MCP servers, path traversal in Ollama model transfers, and information disclosure in SmythOS connectors. Immediate patching and configuration reviews are advised.
-
· 3 entries · daily · Subscribers only
VulnWatch Daily: Critical MCP & Agentic Vulnerabilities in OpenClaw and LiteLLM
New advisories highlight risks in agent gateway configurations and MCP stdio transports. OpenClaw and LiteLLM users must patch immediately to prevent config mutations and authenticated RCE.