skill-health

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill's adversarial test explicitly instructs the agent to create a SKILL.md containing a hardcoded secret (an API key assigned inline), and the safety rules treat hardcoded secrets as a blocking finding — which forces the LLM to generate or handle secret-like values verbatim, creating an exfiltration risk.