auto-dev

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This skill scaffolds a fully autonomous pipeline that repeatedly sends repository code, tests, diffs and runtime outputs to an external LLM (claude) with permission-bypass flags, and grants that LLM automated read/write/test/repair cycles over project source — a high-risk pattern enabling data exfiltration, unattended remote modification (supply‑chain/backdoor risk), and potential abuse without human oversight.