customaize-agent:test-prompt

Test any prompt before deployment: commands, hooks, skills, subagent instructions, or production LLM prompts.

Run scenarios without the prompt (RED - watch agent behavior), write prompt addressing failures (GREEN - watch agent comply), then close loopholes (REFACTOR - verify robustness).

Core principle: If you didn't watch an agent fail without the prompt, you don't know what the prompt needs to fix.

在建立或編輯任何提示（命令、掛鉤、技能、子代理指令）時使用，以驗證其產生所需的行為 - 應用紅-綠-重構循環來提示使用子代理進行隔離測試的工程 來源：neolabhq/context-engineering-kit。