prompt-improver

To turn vague prompts into actionable plans, apply a four-phase workflow: research the codebase context, generate targeted clarifying questions, get user clarification, and execute with grounded context for reproducible decisions.

Prompts lack specifics and fail during design reviews when they miss codebase context. Grounding clarifying questions in actual documentation and codebase findings ensures the resulting instructions are well-scoped and actionable.

The best way to clarify ambiguous instructions is systematic investigation. Perform targeted research on the prompt, generate specific clarifying questions, and use AskUserQuestion to resolve ambiguity before executing the grounded plan.

Yes, you can automate prompt evaluation for testing prompt-based systems by manually invoking the prompt improvement workflow to systematically generate targeted questions and evaluate how well prompts handle ambiguity and missing context.

Generate targeted questions based on codebase context by using a research phase to investigate actual code findings and documentation, then synthesizing those discoveries into specific clarifying questions that resolve prompt ambiguity.

Avoid automated prompt clarification workflows when prompts already contain sufficient specifics and context, as forcing the research and clarification phases on well-scoped instructions adds unnecessary overhead to the execution pipeline.