L’interface est localisée ; les présentations et documents des Skills sont traduits automatiquement par IA dans votre langue. Le code, les commandes, chemins et termes techniques restent en version originale, et le document de l’auteur fait foi.
Explorer les Skills
Développement · SKILL.md

prompt-improver

prompt-improver

Analyze and improve plugin prompts, skill definitions, and command instructions for clarity, safety, and effectiveness. Use when the user asks to "improve a prompt", "review a skill", "enhance instructions", "make this prompt better", "optimize this command", or "audit prompt quality".

Independent GitHub snapshot · Automated classification · Original documentation governs

Documentation de la Skill

GitHub

name: prompt-improver description: 'Analyze and improve plugin prompts, skill definitions, and command instructions for clarity, safety, and effectiveness. Use when the user asks to "improve a prompt", "review a skill", "enhance instructions", "make this prompt better", "optimize this command", or "audit prompt quality".

' allowed-tools: Read, Write, Edit, Glob, Grep version: 1.7.0 author: severity1 severity1@intentsolutions.io license: MIT tags:

  • security
  • compliance compatibility: Designed for Claude Code

Prompt Improver

This skill automatically analyzes and improves plugin prompts, SKILL.md files, command definitions, and agent instructions.

Overview

The prompt-improver skill evaluates plugin content across five dimensions — clarity, safety, effectiveness, completeness, and conciseness — then provides scored assessments and concrete rewrites.

When to Use This Skill

This skill activates when you need to:

  • Review and improve a SKILL.md file's instructions
  • Enhance command or agent markdown definitions
  • Audit prompt quality across a plugin
  • Optimize instructions for better Claude performance
  • Ensure prompts follow marketplace best practices

Instructions

  1. Identify the target — Locate the SKILL.md, command, or agent file to analyze
  2. Read the content — Use Read tool to get the full file contents
  3. Score each dimension on a 1-5 scale:
    • Clarity (1-5): Are instructions unambiguous?
    • Safety (1-5): Does it avoid encouraging dangerous operations?
    • Effectiveness (1-5): Will it reliably produce intended results?
    • Completeness (1-5): Are edge cases addressed?
    • Conciseness (1-5): Is it free of unnecessary verbosity?
  4. Identify improvements — List specific weaknesses with line references
  5. Generate rewrites — Provide improved versions preserving original intent
  6. Apply changes — If requested, use Edit tool to apply improvements

Output Format

## Prompt Analysis: [filename]

### Scores
| Dimension | Score | Notes |
|-----------|-------|-------|
| Clarity | X/5 | ... |
| Safety | X/5 | ... |
| Effectiveness | X/5 | ... |
| Completeness | X/5 | ... |
| Conciseness | X/5 | ... |

**Overall: X/25**

### Improvements
1. [Specific improvement with before/after]

### Suggested Rewrite
[Full improved prompt text]

Best Practices

  • Preserve the original author's intent and style
  • Prioritize safety improvements over stylistic ones
  • Keep suggestions actionable and specific
  • Reference marketplace conventions from CLAUDE.md
  • Validate frontmatter fields match the 2026 spec

Examples

Example 1: Improving a vague skill description

Before:

description: Does stuff with code

After:

description: |
  Analyze source code for common anti-patterns and suggest refactoring improvements. Use when the user asks to "review code quality", "find code smells", or "refactor this file".

Example 2: Adding missing safety guidance

Before:

Delete all temporary files from the project.

After:

Identify temporary files (*.tmp, *.bak, *.swp) in the project. List them for user confirmation before deletion. Never delete files outside the project root.

Prerequisites

  • A plugin directory containing at least one SKILL.md, command, or agent markdown file
  • Read access to the target plugin's .claude-plugin/plugin.json for context
  • Familiarity with the 2026 SKILL.md frontmatter spec

Output

The skill produces a structured analysis report containing:

  • Score card: 5 dimensions rated 1-5 with notes and an overall score out of 25
  • Improvement list: Specific weaknesses with file paths and line references
  • Suggested rewrite: Full improved prompt text preserving original intent

Error Handling

ErrorCauseResolution
No SKILL.md foundTarget path has no skill filesVerify the plugin path and check for skills/*/SKILL.md
Invalid frontmatterYAML parsing failure in target fileReport the specific YAML error and suggest corrections
Empty skill bodyFile has frontmatter but no instructionsFlag as critical and generate a starter template

Resources

Skills enregistrées