Architecture
Shared resourceKeep your project consistent as it grows. Your AI assistant reads your design patterns before building and checks its work against your quality rules after.
01 — Why This Matters
Your AI assistant is excellent at writing code — but it doesn’t automatically know how your project is structured. Without clear guidance, each new file can end up with different naming, different patterns, and different conventions. Over time, small inconsistencies compound into real maintenance problems.
Architecture tools solve this by giving your assistant two things: a pre-read of your design patterns before it starts building, and an automated quality check after it finishes. Your established conventions stay intact, no matter how fast your project moves.
02 — The Difference
Inconsistent output
- ×Files follow different naming patterns
- ×Code structure varies between files
- ×Anti-patterns slip in unnoticed
- ×Manual review catches issues too late
Consistent output
- ✓Your assistant reads your rules before starting
- ✓Every file follows your established structure
- ✓Anti-patterns are flagged immediately
- ✓Automated checks run after every change
03 — How It Works
Read design patterns
Your assistant requests the design patterns for the files it’s about to create or modify. It receives naming rules, structural conventions, and specific guidance for that area of the project.
Pattern: "API route handler"
Naming: kebab-case file, camelCase exports
Structure: validate → service → response
Required: error handling, input validation
Avoid: direct database queries in handlersCheck quality
Once changes are made, your assistant submits them for automated review. It receives a clear report showing what passed, what needs attention, and what must be fixed.
✓ File structure matches pattern
✓ Naming follows conventions
⚠ Missing unit test (recommended)
✗ Direct database access (restricted)04 — Quality Levels
Quality rules are organized by severity so your assistant knows which issues to fix immediately and which are suggestions for improvement.
Critical rules that must never be violated. Anti-patterns and security risks that your assistant will fix before submitting work.
Mandatory standards for every file. These ensure structural consistency and proper conventions across your project.
Best practices that improve quality. Your assistant will flag these as suggestions but won’t block work from being completed.