- Build scalable, dependable, and reproducible software supporting lab research and production platforms.
- Prioritize correctness over cleverness; write obvious, predictable code and avoid premature optimizations.
- Apply DRY (Don't Repeat Yourself) only when patterns repeat across 3 or more instances.
- Strictly adhere to KISS and YAGNI; solve immediate requirements without speculative abstractions.
- Fail fast and explicitly; validate state and arguments early and throw descriptive domain exceptions.
- Maintain a zero-warning codebase across compilers, linters, and type checkers.
- For AI agents: minimize token usage, modify only requested files, and verify syntax and tests before concluding.
- Does the new feature or bug fix directly solve the stated problem with minimal lines changed?
- Are all functions, variables, and modules given self-describing English names?
- Are all external inputs, parameters, and return types strictly typed?
- Do unit and integration tests pass cleanly in the local environment without warnings?
- Is there zero commented-out, unverified, or dead code left in the working tree?
- Can both an automated CI pipeline and an AI agent trace the execution flow without implicit global state?