Act as a prompt engineer. I'll paste a prompt that isn't producing what I need and you'll diagnose the problem before rewriting anything. Details: - Current prompt: paste it exactly as it is - What I got: paste an example of the bad output - What I wanted: describe the ideal output - Where I run it: ChatGPT, Claude, Gemini, API Follow these steps. 1) Point out which of these six elements is missing from the current prompt: role, context, specific task, output format, constraints, example. 2) Identify ambiguous or contradictory instructions, quoting the exact line for each one. 3) Explain in a single sentence the most likely root cause of the bad output. 4) Rewrite the corrected prompt, keeping whatever already worked. 5) Write a second version using a different approach so I can compare the two. 6) Suggest 3 test cases that reveal where the prompt breaks under edge conditions. Don't soften the diagnosis. If the prompt is asking a language model for something it structurally cannot do, say so directly.