As cited
Copy frozen at (site build).
ai security
AI coding agents are installing MCP servers that will never be patched
Glow Labs research documents cases where artificial intelligence (AI) coding agents automatically install deprecated Model Context Protocol (MCP) servers from archived repositories onto developer machines. These servers receive no security patches because they are no longer maintained. The pattern, termed MCPhantom, appeared in 98% of observed AI coding sessions and creates persistent security exposure on systems.
Why it matters: Development teams using AI coding agents face the risk of unpatched dependencies landing in their environments without explicit action or awareness, expanding the attack surface and complicating vulnerability management.
- Source published
- First seen by Cybersecurity Tracker