Does this skill access my local files?

Yes, it uses tools like grep and git log to explore your local repository to ensure its clarification questions are grounded in your actual project structure.

Can I select multiple options in the clarification phase?

The skill supports both single and multi-select options depending on the question, and you can always provide a custom 'Other' response.

How does Prompt Improver determine if a prompt is vague?

The skill is typically triggered by a evaluation hook that checks for missing specifics, context, or clear success criteria in the initial user prompt.

Will it perform research every time I send a message?

No, it is only invoked when a prompt is identified as vague or when manually requested, and it checks conversation history first to avoid redundant research.

Prompt Improver

Name: Prompt Improver
Author: rafaelcalleja

byrafaelcalleja

•

生产力与工作流

Transforms vague instructions into actionable tasks through systematic research and targeted clarification questions.

关于

The Prompt Improver skill eliminates ambiguity in AI interactions by performing automated research on your codebase and documentation before execution. When a request lacks sufficient detail, this skill systematically explores your project structure, git history, and external documentation to generate grounded, multiple-choice questions that clarify your intent. This ensures that Claude proceeds with a clear, context-aware plan, reducing errors and iterations caused by underspecified prompts.

主要功能

Dynamic generation of 1-6 targeted, research-grounded clarification questions
Automated codebase exploration via grep, glob, and git history analysis
Structured 4-phase workflow: Research, Questioning, Clarification, and Execution
1 GitHub stars
Integration with WebSearch and WebFetch for external documentation context
Conversation history awareness to prevent redundant questioning

使用场景

Clarifying high-level or ambiguous requests like 'fix the bug' or 'refactor this module'
Onboarding to unfamiliar codebases by gathering context before performing tasks
Standardizing requirement gathering for complex feature implementations

关于

主要功能

Dynamic generation of 1-6 targeted, research-grounded clarification questions
Automated codebase exploration via grep, glob, and git history analysis
Structured 4-phase workflow: Research, Questioning, Clarification, and Execution
1 GitHub stars
Integration with WebSearch and WebFetch for external documentation context
Conversation history awareness to prevent redundant questioning

使用场景

Clarifying high-level or ambiguous requests like 'fix the bug' or 'refactor this module'
Onboarding to unfamiliar codebases by gathering context before performing tasks
Standardizing requirement gathering for complex feature implementations