prompt-improver
πPluginseverity1/claude-code-prompt-improver
A UserPromptSubmit hook that intercepts vague prompts and enriches them before Claude Code executes. Uses AskUserQuestion for targeted clarifying questions, achieving 31% token reduction with zero overhead for clear prompts.
Overview
A Claude Code plugin that intercepts vague prompts before execution and enriches them through targeted clarifying questions. It uses a UserPromptSubmit hook to evaluate prompt clarity, and for ambiguous prompts, it launches a research-driven questioning flow to gather context before proceeding.
Key Features
- Automatic Clarity Evaluation β Checks prompts against conversation history; clear prompts proceed immediately with zero overhead
- Research-Based Questioning β For vague prompts, creates a research plan, gathers codebase context, then asks 1-6 grounded clarifying questions via the AskUserQuestion tool
- Skill-Based Architecture β v0.4.0 separates hook-level evaluation from skill-level research, achieving 31% token reduction compared to earlier versions
- Multiple Installation Options β Install via marketplace, local development setup, or manual script copying
Who is this for?
Claude Code users who want better first-try results without back-and-forth. Particularly helpful for developers who often start with broad or ambiguous requests and want the AI to intelligently ask for clarification before diving in.
Installation
/plugin marketplace add severity1/claude-code-prompt-improver