prompt-improver

Prompt Improver

A Claude Code plugin that automatically provides grammar corrections and refined versions of your prompts before Claude responds.

What It Does

Every time you submit a prompt, Claude will first output:

1. Grammar: [Your prompt with corrected grammar, spelling, and punctuation] 2. Refined: [Your prompt rewritten as a highly educated writer would phrase it]

Then Claude proceeds with the actual response.

Example

You type:

why this dont work good

Claude responds:

1. Grammar: Why doesn't this work well? 2. Refined: What is preventing this from functioning properly?

[Then the actual answer to your question...]

Installation

Quick Install (Recommended)

# Add the marketplace
/plugin marketplace add methuz/claude-prompt-improver

# Install the plugin
/plugin install prompt-improver@prompt-improver-marketplace

Install from Local Path

# Clone this repository
git clone https://github.com/methuz/claude-prompt-improver.git

# Add as local marketplace
/plugin marketplace add ./claude-prompt-improver

# Install the plugin
/plugin install prompt-improver@prompt-improver-marketplace

Manual Installation

If the plugin system doesn't work, add this to your ~/.claude/settings.json:

{
  "hooks": {
    "UserPromptSubmit": [
      {
        "hooks": [
          {
            "type": "command",
            "command": "/path/to/claude-prompt-improver/hooks/prompt-improver.py",
            "timeout": 10
          }
        ]
      }
    ]
  }
}

Configuration

Edit hooks/prompt-improver.py to customize:

Variable	Default	Description
MIN_PROMPT_LENGTH	10	Skip prompts shorter than this
SKIP_PREFIXES	("/", "!", "?")	Skip prompts starting with these
SKIP_EXACT	{"y", "n", "yes", ...}	Skip these exact phrases

Customizing the Output Style

Edit the create_instruction() function in prompt-improver.py:

Academic Style

Output format (use exactly):
> **1. Corrected**: [grammatically correct version]
> **2. Academic**: [formal academic prose]
> **3. Concise**: [shortened essential version]

Technical Style

Output format (use exactly):
> **1. Grammar**: [fix errors]
> **2. Technical**: [as a senior engineer would phrase it]

Minimal Style

Output format (use exactly):
> **Corrected**: [grammar fixed]

How It Works

┌─────────────────┐     ┌──────────────────────┐     ┌─────────────────┐
│  User types     │────▶│  prompt-improver.py  │────▶│  Claude sees    │
│  prompt         │     │  intercepts &        │     │  instruction +  │
│                 │     │  adds instruction    │     │  original       │
└─────────────────┘     └──────────────────────┘     └─────────────────┘

1. You submit a prompt
2. Claude Code triggers the UserPromptSubmit hook
3. The hook script outputs a <MANDATORY-FIRST-RESPONSE> instruction
4. Claude follows the instruction, providing corrections first
5. Claude then answers your actual question

Skipped Prompts

The plugin automatically skips:

• Very short prompts (< 10 characters)
• Slash commands (/help, /clear, etc.)
• Simple confirmations (yes, no, ok, thanks)

Uninstall

/plugin uninstall prompt-improver@prompt-improver-marketplace

Or manually remove the hook from ~/.claude/settings.json.

License

MIT License - feel free to modify and share!

Contributing

1. Fork the repository
2. Create your feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request