Fake Bug Report Hijacks AI Coding Agents at Scale

Tenet Security demonstrated an "agentjacking" attack in which attackers submit poisoned error reports to publicly exposed Sentry DSNs; AI coding assistants (e.g., Claude Code, Cursor, Codex) can retrieve those poisoned events and treat embedded instructions as actionable, leading to execution of attacker-controlled code on developer machines and potential theft of cloud credentials, GitHub tokens, SSH keys, and CI/CD secrets. The issue stems from AI agents' inability to distinguish read data from instructions; mitigations recommended include disabling package-install scripts, requiring human approval for shell commands, least-privilege execution, and runtime intent monitoring.