sleight-of-hand

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill mandates showing exact before-and-after diffs and making precise edits of specific lines/strings, so if the target is an API key, token, password, or other secret the agent would necessarily include that secret verbatim in its output.