Your AI Agent Is an Attacker's New API: Prompt Injection, Tool Hijacking, and Securing Local Coding Agents

A practical field guide to prompt injection, tool hijacking, and hardening local coding agents against malicious repository content and data exfiltration.