Prompt Linter
AI-powered linter that helps you write better prompts by identifying logical conflicts, ambiguities, and areas for improvement.
✨ Features
🔍 Real-time Analysis
Get instant feedback on your prompts as you type:
- Logical conflicts and contradictions
- Role clarity and consistency
- Instruction complexity
- Missing examples and specifications
- Emphasis and formatting issues
🎯 Smart Issue Detection
You are a helpful assistant who is very formal but also casual.
Please make the image bigger and smaller.
Generate a blue or maybe green logo.
↓ The linter identifies:
- Contradictory personality traits
- Conflicting size instructions
- Ambiguous color specifications
💰 Cost Tracking
Monitor your API usage with detailed token counting and cost estimation for each analysis.
🚀 Getting Started
- Install the extension from VS Code marketplace
- In VS Code settings:
- Choose your preferred AI provider
- Configure the corresponding API key
- Start writing! By default, all
.prompt files will be linted automatically
(You can customize file patterns in settings)
Note:
You can disable specific rules for sections of your prompt using comments:
#prompt-linter ignore
This section will not be checked
#endignore
📋 Rule Categories
1. Role Clarity
Ensures consistent assistant behavior:
❌ You are a senior developer but also act as a comedian.
✅ You are a senior developer who explains concepts clearly.
2. Logical Conflicts
Catches contradictory instructions:
❌ Prioritize speed over accuracy, but ensure 100% accuracy.
✅ Prioritize speed while maintaining 95% accuracy.
Verifies clear specifications:
❌ Convert the input to the proper format.
✅ Convert dates to ISO 8601 format, e.g., "2024-03-20T15:30:00Z"
4. Instruction Complexity
Monitors for overly complex requirements:
❌ Analyze performance, security, maintainability, and...
✅ Focus on identifying security vulnerabilities first.
5. Emphasis Overuse
Checks for effective emphasis:
❌ **IMPORTANT!!!** This is **CRITICAL** and **URGENT**!!!
✅ Important: This requires immediate attention.
🤝 Contributing
We welcome contributions to make this linter better! Here are some ways you can help:
🐛 Bug Reports & Feature Requests
- Found a bug? Please open an issue on our GitHub repository
- Have an idea for a new feature? Share it in the issues section
📝 New Rules & Examples
We're always looking to expand our rule set. Consider contributing:
- Additional prompt engineering best practices
- New rule categories
- More example cases (both valid and invalid)
- Edge cases that should be caught
💻 Code Contributions
Want to contribute code? Great! Some areas that need work:
- Rule implementation improvements
- Additional model support
Please check our contribution guidelines before submitting a pull request.
Together we can build a more robust prompt linting tool!
