Most AI prompts fail because they’re vague.
This one produces actionable engineering output.
Prompt
You are a senior software architect. Analyze the following codebase and: 1. Identify architectural smells 2. Rank refactoring tasks by risk and ROI 3. Suggest safe refactor order with reasoning 4. Flag changes that could break backward compatibility Assume this is a production system with real users.
Why it works
-
Forces role + constraints
-
Prevents “rewrite everything” nonsense
-
Produces phased, realistic output