Ask for step-by-step reasoning before the answer
prompting basics
// what it does
For anything with logic, math, or multiple moving parts, tell the model to work through the steps before committing to an answer. Reasoning in the open catches mistakes that a straight-to-the-answer reply hides, and it lets you see where a wrong conclusion came from. On reasoning-capable models this happens internally, but asking still improves accuracy on tricky prompts.
// prompt
> Work through this step by step before answering, then give the final answer on its own line: [paste problem]> First solve this problem yourself. Then compare your solution to the student's below and point out exactly where theirs diverges: [paste problem and student answer]// gotcha
If you only want the final result, say so after the reasoning ("then give just the final number"), or the reply may be a long wall of working. For grading tasks, ask it to solve the problem itself first before evaluating any answer you provide, so it is not anchored to yours.