Open-AutoGLM: Open-Source Android AI Agent Framework
An open-source Android AI agent framework with over 25,000 GitHub stars. Vulnerable to multiple attacks including broadcast interception and command injection.
Tag: AI Agent
MobA: Open-Source Android AI Agent Framework
An open-source Android AI agent framework that streams screenshots over exec-out, avoiding file race conditions, but still vulnerable to other attack vectors.
Antigravity Coding Agent Used in ADK Workflow Attack
Google's Antigravity coding agent was used in the ADK workflows and could be manipulated via prompt injection to execute arbitrary commands, leading…
Copilot CLI Affected by Azure DevOps MCP Flaw
GitHub Copilot CLI was successfully used in Manifold Security's proof of concept to demonstrate the Azure DevOps MCP server vulnerability, showing the…
AI Agent JADEPUFFER Exploits Langflow RCE to Automate Database Ransomware Attack
Security firm Sysdig has identified what it believes is the first ransomware attack fully orchestrated by an AI agent, dubbed JADEPUFFER. The…
JADEPUFFER: AI-Agent-Driven Cyber Operator
An AI-agent-driven operator first documented by Sysdig. Deployed ENCFORGE ransomware against Langflow servers, using Docker socket for host breakout. Previously used throwaway…
Microsoft Warns Poisoned MCP Tool Descriptions Can Make AI Agents Leak Data
Microsoft has issued a warning about a new attack vector targeting AI agents that use the Model Context Protocol (MCP). Attackers can…
Copilot Studio: Custom AI Agent Builder
Copilot Studio allows building custom AI agents that can reach into business systems, increasing the attack surface for poisoned tool descriptions.
Microsoft Purview Data Loss Prevention
Purview DLP helps prevent data exfiltration by monitoring and controlling data transfers, relevant for AI agent attacks.
Microsoft Entra Agent ID for Identity Management
Entra Agent ID provides identity management for AI agents, helping to enforce least privilege and monitor actions.