- Define the concept.
- State why it matters.
- Give the mechanism or derivation.
- Mention one limitation or edge case.
- Restate inputs and outputs.
- Write the simplest correct version.
- Mention runtime and memory.
- Add one important edge case.
- Name the main failure classes.
- Check the highest-probability class first.
- Say what observation would confirm or reject it.
- Narrow systematically.
- State the bottleneck.
- Decompose cost.
- List the main levers.
- Explain trade-offs.
- Choose the best first action for the given constraint.
- Clarify the claim.
- Ask what changed.
- Ask what stayed fixed.
- Ask whether the metric is the right one.
- State the strongest justified conclusion.
- Problem.
- Main idea.
- Why it might work.
- Weak assumptions.
- Missing ablations.
- What to test next.