plan-eng-review

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). This skill file instructs the agent to eval and execute local helper binaries and scripts (eval "$(~/.claude/skills/gstack/bin/gstack-slug ...)", codex exec, gstack-team-init, gstack-config, etc.), modify the repo (git rm/add/commit), create marker files in the user's home, and call remote/external tooling—behavior that enables remote code execution, hidden network requests and potential exfiltration or supply-chain abuse and therefore represents a high-risk backdoor pattern.