ia-agent-native-architecture

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). These docs recommend exposing powerful agent primitives (bash, arbitrary HTTP calls, git push/self-deploy, write_file+commit flows and hook-executed scripts) and self-modification patterns that — without strict, enforced approval/gating — create high-risk vectors for remote code execution, supply‑chain compromise, and data exfiltration.