Vulnerability Tool Poisoning Attacks: Critical Vulnerability Discovered in Model Context Protocol (MCP) Ddos April 10, 2025 4 minutes read 0 An MCP rug pull allows an attacker to change the tool description after the user has already approved it for use. | Image: Invariant 🔐 Access to This Vulnerability Report Requires Support This article is available to verified supporters only - contribute to read the full report Or choose another support option: Support via PayPal Support via BMC Share this article: Facebook Post LinkedIn Telegram Copy Link Related posts: Critical Flaws in AI Browse Agents: Exposed to Credential Theft and Hijacking BIND DNS Server Vulnerable to Remote Crash Windows Server 2025 “BadSuccessor” Flaw Allows Domain Takeover (PoC Available, No Patch) OpenPGP.js Flaw Allows Signature Spoofing, Breaking Message Trust β PoC Available! Trojan AI: OpenClaw “Skills” Marketplace Flooded with Malware Tags: AI Agents AI security AI vulnerability artificial intelligence cybersecurity Invariant Labs MCP (Model Context Protocol) Tool Poisoning Attack Leave a Reply Cancel replyLogged in as . Edit your profile. Log out? Required fields are marked *Comment *