ai-check

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The prompt explicitly requires quoting "exact word or phrase" from the input text (e.g., in the EVIDENCE LOG and "cite the exact word or phrase that fired"), so if the analyzed text contains API keys, tokens, passwords, or other secrets the model is forced to reproduce them verbatim, creating a high exfiltration risk.