prime-expand

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 0.90). The skill explicitly instructs the agent to include the original "seed" verbatim in the sub-agent brief and in the emitted output (including refusal), so if a user supplies an API key/password as the seed the LLM would necessarily handle and output that secret verbatim.