mending

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill requires the agent to output the repaired artifact and a before/after diff verbatim with no sanitization step, so if the artifact contains API keys, tokens, cookies, or passwords the agent would be forced to include those secrets in its output (high exfiltration risk).