---

name: coding-prompt-bank description: Base rules and behaviors for AI coding agents. Trigger on starting a coding project, "use my coding rules", "coding agent setup", "initialize project", "new codebase", or when providing instructions to any coding agent.

Coding Prompt Bank

Core philosophy and tier-based requirements for AI coding.

Philosophy

CopyCORRECTNESS > SPEED
Working first attempt beats fast iteration + debug.
Mental simulation catches 80% of bugs before execution.

EVERY LINE IS LIABILITY
Config > code. Native > custom. Existing > new.
Delete what doesn't earn its place.

PARSE, DON'T READ
Grep. Find. Read signatures first.
Never ask for overview—discover it.

Project Tiers

Tier 1: Hackathon/Prototype

• Speed matters most
• Skip tests unless critical path
• Minimal documentation
• Technical debt acceptable

Tier 2: Default/MVP (most projects)

• Balance speed and quality
• Core path tests required
• Basic documentation
• Reasonable architecture

Tier 3: Production/Enterprise

• Quality over speed
• Comprehensive tests (80%+ coverage)
• Full documentation
• Robust error handling
• Security review required

Universal Rules

1. Investigate before implement - Search for existing patterns
2. Type everything - TypeScript strict, Python type hints
3. Config over code - Magic numbers in config files
4. Fail fast - Validate early, clear error messages
5. Atomic commits - One logical change per commit

Default Behaviors

When Implementing

1. Research existing solutions first
2. Define interfaces before implementation
3. Write tests for core logic
4. Document public APIs

When Debugging

1. Reproduce first
2. Isolate the problem
3. Fix root cause, not symptom
4. Add regression test

Before Completing

1. Run type checker
2. Run linter
3. Run tests
4. Review changes

Integration

Loaded automatically on project initialization. Tier detected from:

• package.json keywords
• README indicators
• User specification