</>MCP Agents Market
Plugin

claude-code-prompt-improver

by severity11.9kPythonUpdated 2026-06-03

Intelligent prompt improver hook for Claude Code. Type vibes, ship precision.

Claude Code

The claude-code-prompt-improver is a Claude Code plugin that intelligently enhances the prompt-to-output workflow by injecting targeted context at key moments during code generation. It evaluates every submitted prompt to determine if clarification is needed, asking 1-6 grounded questions only when genuinely vague while letting clear prompts proceed immediately with zero skill overhead. The plugin uses a hook-based architecture that fires nine different "nudges" (improve, approach-assessment, workflow, output-readability, ask-user-question, plan-mode, plan, background-exec, subagent-routing) only when their specific conditions apply, reducing correction round-trips and delivering better first-attempt results.

Key Features

Hook-level evaluation with 31% token reduction (189 tokens per prompt in v0.4.0, down from 275 in v0.3.x)
Clear prompts proceed immediately with zero skill overhead; vague prompts trigger research and 1-6 contextual questions
Nine conditional nudges inject guidance at UserPromptSubmit, PreToolUse, and SubagentStart events only when applicable
Skill-based research architecture dispatches Glob, Grep, WebSearch, and multi-file Read to separate Haiku context window
Declarative JSON nudge registry allows adding new context injections without Python code changes
Bypass prefixes (* for skip evaluation, / for slash commands, # for memorize) let users control intervention
Progressive disclosure loads detailed reference files (question-patterns, research-strategies, examples) only when needed
Plan-mode assessment automatically enters planning workflow for complex, multi-step, or architectural tasks

Use Cases

  • 01Clarifying vague bug-fix requests by asking which specific error or file needs attention before proceeding
  • 02Automatically assessing whether complex refactoring tasks need a subagent, orchestration, or direct execution
  • 03Injecting plan-first guidance for multi-step workflows before any code is written
  • 04Running long-running dev servers and watchers in background mode with output polling
  • 05Routing research and planning subagents to favor breadth over depth with conclusion-first reporting
  • 06Generating human-readable deliverables (reports, reviews, summaries) with lead conclusions and structured formatting

Related Plugins

View more

andrej-karpathy-skills

A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.

Plugin216.1k

claude-mem

Persistent Context Across Sessions for Every Agent – Captures everything your agent does during sessions, compresses it with AI, and injects relevant context back into future sessions. Works with Claude Code, OpenClaw, Codex, Gemini, Hermes, Copilot, OpenCode + More

Plugin95k

Understand-Anything

Graphs that teach > graphs that impress. Turn any code into an interactive knowledge graph you can explore, search, and ask questions about. Works with Claude Code, Codex, Cursor, Copilot, Gemini CLI, and more.

Plugin84.9k

rtk

CLI proxy that reduces LLM token consumption by 60-90% on common dev commands. Single Rust binary, zero dependencies

Plugin82.1k

claude-code-prompt-improver — FAQ

What is claude-code-prompt-improver?+

It's a Claude Code plugin that hooks into prompt submission, tool use, and subagent events to inject contextual guidance that improves first-attempt outputs. The plugin evaluates every prompt and asks clarifying questions only when genuinely vague, while clear prompts proceed immediately with minimal overhead.

How do I install claude-code-prompt-improver in Claude Code?+

The recommended method is via marketplace: run 'claude plugin marketplace add severity1/severity1-marketplace', then 'claude plugin install prompt-improver@severity1-marketplace', and restart Claude Code. Alternatively, you can install locally from a cloned repo or manually copy files to ~/.claude/hooks/ and update settings.json.

Which version of Claude Code is required?+

Claude Code 2.0.22 or later is required because the plugin uses the AskUserQuestion tool for targeted clarifying questions.

What's the token overhead per prompt?+

In v0.4.0, the hook-level evaluation adds approximately 189 tokens per prompt (down 31% from 275 in v0.3.x). Clear prompts have zero skill overhead; vague prompts load the skill and reference files only when needed for research and questioning.

Is claude-code-prompt-improver free to use?+

Yes, it is open source and available under the MIT license.

Can I bypass the prompt improvement for specific requests?+

Yes, prefix your prompt with * to skip evaluation (e.g., 'claude "* add dark mode"'), use / for slash commands, or # for memorize bypass.

How do I install claude-code-prompt-improver?+

Open the source repository on GitHub and follow its README. claude-code-prompt-improver is a plugin — MCP Agents Market links you directly to the official repo.

Is claude-code-prompt-improver free?+

claude-code-prompt-improver is an open-source project hosted on GitHub. Check the repository for its license and any usage requirements.

Related searches

Developer Tools PluginsProductivity & Workflow PluginsCode Review PluginsBest MCP ServersBest SkillsBest AgentsTop rated listingsClaude Code toolsBrowse all categories