Why Fix Prompts Are Assistive, Not Auto-Fixes
Fix prompts are tools to help you fix issues faster, not automatic solutions. They provide structured information you can use with AI coding assistants, but you still need to review, understand, test, and verify fixes. They assist your work, they don't replace it.
What "Assistive" Means
Assistive means the prompts help you work faster and better, but you're still in control:
- You review - Always review AI-generated fixes before applying
- You understand - Make sure you understand what the fix does
- You test - Verify fixes actually work
- You decide - Choose whether to apply fixes or try different approaches
Why Not Auto-Fix?
1. Context Matters
AI assistants don't have full context:
- Codebase knowledge - May not know your architecture patterns
- Business logic - Doesn't understand domain-specific requirements
- Team standards - May not follow your coding style or conventions
- Dependencies - Might not know about your library versions
2. AI Can Be Wrong
AI assistants make mistakes:
- May misunderstand the problem - Could fix the wrong thing
- May introduce new bugs - Fix might break something else
- May use outdated patterns - Might suggest old approaches
- May not handle edge cases - Could miss important scenarios
3. Code Quality
Auto-applied fixes might not meet your standards:
- Code style - May not match your team's style
- Best practices - Might not follow your conventions
- Performance - Might not optimize correctly
- Maintainability - Code might be harder to maintain
How Fix Prompts Assist
1. Structured Information
Fix prompts organize information clearly:
- Problem description
- Error details
- Context and evidence
- Suggested fix approach
This saves you time organizing information yourself.
2. Better AI Interactions
Well-structured prompts lead to better AI suggestions:
- AI assistants understand problems better
- More accurate fix suggestions
- Less back-and-forth needed
3. Learning Tool
Fix prompts help you learn:
- See how issues are diagnosed
- Understand fix approaches
- Learn patterns for similar issues
Your Responsibilities
Review Generated Code
Always review AI suggestions before applying:
- Understand what the code does
- Check if it solves the problem
- Verify it doesn't break other things
- Ensure it follows your standards
Test Fixes Thoroughly
Don't trust fixes blindly:
- Test the fix works
- Verify it doesn't break other functionality
- Check edge cases
- Re-run Rihario exploration if needed
Understand the Problem
Make sure you understand:
- What the issue was
- Why it occurred
- How the fix works
- If the fix is appropriate
Best Practices
- Use as starting point - Prompts help you start, not finish
- Review everything - Never blindly apply AI suggestions
- Test thoroughly - Verify fixes work correctly
- Iterate if needed - Don't hesitate to refine fixes
- Learn from prompts - Use them to improve your understanding
Example Workflow
Notice: Steps 5-9 are your responsibility. The prompt assists steps 1-4.
When Fix Prompts Work Best
- Clear, isolated issues - Well-defined problems in specific files
- Obvious bugs - Errors with clear causes
- You understand the domain - You can verify fixes are correct
- You review carefully - You take time to understand suggestions
When to Be More Cautious
- Complex business logic - Requires deep domain knowledge
- Architectural changes - May have wide-reaching impacts
- Security-sensitive code - Needs careful security review
- Performance-critical code - Needs performance validation
Summary
Fix prompts are powerful tools that help you fix issues faster, but they're assistive, not automatic. You're always responsible for:
- Reviewing AI suggestions
- Understanding what fixes do
- Testing fixes thoroughly
- Ensuring code quality
Use fix prompts to speed up your work, but never skip the review and testing steps.