# Research guidelines

Use these standards when asking subagents for chess algorithms, rendering, or optimization help.

## Good prompts

- "Compare mailbox arrays, 0x88, and bitboards for a beginner-friendly Zig chess engine; cite sources and propose experiments."
- "Research legal move generation edge cases and perft testing for chess correctness."
- "Review a simple Vulkan chessboard renderer for resource lifetime and data-layout risks."
- "Explain alpha-beta pruning and move ordering with references and benchmark ideas."

## Expected evidence

- Official Zig documentation for Zig behavior.
- Chess programming references, engine documentation, or reputable technical articles for chess algorithms.
- Vulkan/GLFW upstream documentation for rendering behavior.
- Vendor/compiler documentation for low-level optimization claims.

## Required distinction

Every research answer should separate:

1. Established facts.
2. Design tradeoffs.
3. Hypotheses requiring measurement.
4. Suggested experiments/benchmarks.
5. References for follow-up.