prd-done

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The prompt instructs the agent to scan commits for secrets and to "present ALL" review comments and quote suggestions verbatim, which could force the LLM to reproduce secret values found in code/reviews unredacted.

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The document instructs an automated agent to consider and execute arbitrary bash commands found in PR templates and to "propose and execute" template requirements, which can enable remote code execution/backdoor behavior if automated without strict safeguards.