MCP Security

Why It Matters

MCP servers give agents direct access to files, code, SaaS accounts and internal APIs, so a weak server or an untrusted tool description becomes part of the agent's attack surface.

Relevant Frameworks

Incidents and Analysis

• OpenCode Flaw Lets Malicious MCP Servers Execute Code During Sign-In on Windows and WSL 2026-10-22 · thehackernews.com
  MCP server-supplied command injection during OAuth sign-in
• Autonomous LLMjacking: ‘Skynet’ Agent Hijacks Exposed AI Models to Run Attacks 2026-10-21 · Security.com
  Autonomous LLMjacking
• Last week in Agent Security Update 3: That's not Aoki! 2026-10-19 · DEV Community
  Authenticated remote code execution and credential targeting
• Fake Web3 Accelerator Uses npm Packages to Spread Malware and Hijack AI Agent Traffic 2026-10-19 · thehackernews.com
  Malicious npm package and AI agent traffic interception
• Welcome to the Jungle: What We Found Inside 15,465 Public MCP Servers 2026-10-06 · thehackernews.com
  Insecure public MCP servers
• Supply Chain Security — Latest News, Reports & Analysis 2026-10-05 · The Hacker News
  Crafted model repositories; MCP secret exposure
• Official MCP Python SDK Flaw Can Let Malicious Servers Steal OAuth Credentials 2026-09-29 · thehackernews.com
  OAuth credential theft by a malicious MCP server
• AI Security — Latest News, Reports & Analysis 2026-09-21 · The Hacker News
  Unauthenticated local model control; MCP command injection via crafted notebooks
• Context7 MCP documentation server prompt-injection flaw 2026-09-13 · AI RiskAtlas
  Prompt injection delivered through an MCP documentation server
• The AI Supply Chain Has a Security Problem, and Much of It Is Sitting on the Open Internet 2026-09-11 · Security Affairs
  Exposed AI workflow components
• AIR Security found malicious servers in the official MCP registry 2026-08-27 · The Injection
  MCP server hijacking via expired domain re-registration
• Marimo Notebook Flaw Could Run MCP Commands Before Cells Execute in Edit Mode 2026-08-25 · thehackernews.com
  Malicious notebook triggering MCP command execution
• How MCP Servers Can Expose Enterprise Secrets 2026-08-17 · thehackernews.com
  Secret exposure through MCP servers
• Malicious MCP Servers Can Split Instructions to Make AI Coding Agents Exfiltrate Secrets 2026-08-11 · thehackernews.com
  Fragmented instructions from a malicious MCP server
• ⚡ Weekly Recap: AI Goes Rogue, Metabase 0-Day, MCP Supply-Chain Attacks, and Router Backdoors 2026-08-10 · thehackernews.com
  Multiple, including MCP supply chain attacks
• Veeam, Terraform MCP, Django Patch Critical Flaws, Led by CVSS 10.0 Cross-Tenant Bug 2026-08-05 · thehackernews.com
  Cross-tenant token reuse in an MCP server
• 5 Agentic AI Security Incidents & Vulnerabilities 2026-08-03 · Curity
  Prompt injection causing private data leaks
• Critical Ruflo Flaw Lets Attackers Spawn Rogue AI Swarms 2026-07-30 · securityweek.com
  Unauthenticated command execution in an MCP bridge
• Ruflo MCP Flaw Lets Unauthenticated Attackers Run Commands and Poison AI Memory 2026-07-29 · thehackernews.com
  Unauthenticated RCE and memory poisoning in an agent harness
• AI Agent Security: Prompt Injection, Data Leakage and More 2026-07-24 · DIYAI
  Indirect prompt injection and memory poisoning
• Microsoft Azure DevOps MCP Flaw Lets Hidden PR Comments Hijack AI Review Agents 2026-07-22 · thehackernews.com
  Prompt injection via PR descriptions through an MCP tool
• Clean GitHub Repo Tricks AI Coding Agents into Running Malware 2026-07-21 · Agentic Threat Tracker / Whiteintel
  Prompt injection rewriting agent configuration; trojanized package releases
• FakeGit Campaign Uses 7,600 GitHub Repositories to Spread SmartLoader Malware 2026-07-20 · thehackernews.com
  Fake AI skills and MCP servers
• AI Agent Security Practices 2026: Prompt Injection, MCP Risks, Data Leaks 2026-06-30 · TechStoriess
  Prompt injection in documents and code leading to exfiltration or RCE
• Microsoft Warns Poisoned MCP Tool Descriptions Can Make AI Agents Leak Data 2026-06-30 · thehackernews.com
  Poisoned MCP tool descriptions
• Amazon Q Developer Flaw Could Let Malicious Repos Run Code via MCP Configs 2026-06-26 · thehackernews.com
  Malicious repository MCP configuration
• New Enterprise-Ready MCP Specification Brings New Security Challenges 2026-06-26 · securityweek.com
  Not applicable (specification change)
• AI Agents Are the New Exposure Point for MSPs and SMBs 2026-04-30 · Cybersecurity Insiders
  Exposed and over-privileged AI infrastructure
• From Prompt Injections to Protocol Exploits: Threats in LLM ... 2025-06-30 · arXiv
  Prompt injection via issues through an MCP server