Sophos telemetry from June 2026 shows Claude Code, Cursor, and Codex setting off credential-access, LOLBin, and persistence rules on Windows endpoints — because to a behavioral engine, benign agent work is indistinguishable from an intrusion. Here's what fired, why allowlisting is the wrong fix, and how to scope it safely.
Coding Agent Security Checklist 2026: the isolation model every operator needs
Beware: Your Coding Agent's Sandbox Was Leaking Its Own Credentials to Spawned Commands
Beware: Claude Code's permissions.deny Silently Fails on Absolute Paths
Beware: Claude Code Subagents Can Spontaneously Trigger Prompt Injection
Beware: Claude Code on Windows — Click-Through Permissions
Sophos says Claude Code, Cursor, and Codex triggered Windows endpoint rules for credential access, LOLBin downloads, and…
Sophos ran a week of endpoint telemetry on Claude Code, Cursor, and Codex. 56.2% of blocking hits were credential access. The…
Three vulnerabilities from the last few months, three different layers of the AI-coding-agent stack,...
AI Now’s Friendly Fire PoC shows Claude Code and Codex running README-planted payloads on hosts when autonomous command approval…
Last month, a developer cloned a GitHub repo and opened it in Claude Code. Before they even clicked...
The security landscape for AI developer tools shifted recently with the discovery of a critical...
