Transforms vague or ambiguous prompts into actionable tasks through systematic research and targeted clarification.
The Prompt Improver skill optimizes the interaction between users and AI by ensuring that vague requests are never executed based on assumptions. When a prompt lacks context, this skill triggers a structured four-phase workflow: it researches the codebase and conversation history, generates grounded multiple-choice questions for the user, gathers specific clarifications, and finally executes the task with full context. This process significantly reduces errors, minimizes hallucinations, and ensures that complex technical tasks are aligned with the project's actual architecture and requirements.
Key Features
01Dynamic research plans utilizing TodoWrite, Grep, and WebSearch
02Context-aware execution that integrates user answers with research data
03Automated codebase and history research to find missing context
04Targeted multiple-choice question generation grounded in actual findings
05Four-phase structured workflow: Research, Question, Clarify, and Execute
060 GitHub stars
Use Cases
01Gathering necessary architectural context before performing complex migrations
02Clarifying high-level or ambiguous requests like 'fix the bug' or 'refactor this'
03Onboarding onto a new codebase by systematically exploring existing patterns
