evaluation-testing

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill requires capturing and serializing full tool-call traces (including parameters/inputs) and injecting candidate outputs into judge prompts and reports without redaction, which means secrets or API keys present in those traces would be handled and potentially emitted verbatim by the LLM — an explicit secret-exfiltration risk.