security-research

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.80). The prompt explicitly requires "exact evidence" and "exact file paths and commands" from repository and PoC scans (and instructs hunters to hunt credential exposure and environment-variable risks), which can force the agent to surface hardcoded API keys, tokens, or plaintext passwords verbatim if present — creating an exfiltration risk.