The Prompt Differential Diagnosis Engine

Want AI to diagnose your prompt like its a doctor finding ailments and illnesses

Used by: Enterprise prompt engineering teams, AI product quality assurance engineers, frontier model researchers debugging prompt behavior.

What it is: A protocol that treats an underperforming prompt like a patient with symptoms — systematically running through a differential diagnosis to identify which of multiple possible causes is responsible for the failure, rather than guessing and applying the wrong fix.

Why elite performers use it: Amateur prompt engineers fix symptoms. Elite prompt engineers diagnose causes. This protocol prevents the most common refinement failure — applying a fix that addresses the wrong problem and either doesn't help or makes things worse. The structured diagnostic framework ensures that every potential cause is systematically evaluated before treatment is applied, just as a physician rules out conditions before prescribing.

How to apply: Run this before making any changes to an underperforming prompt. The protocol requires honest answers to each diagnostic question — if you don't know whether an instruction is in the attention trough, test it before proceeding. The "differential weighting" step is critical: multiple causes may be contributing, and you need to know which is primary. Treat the primary cause first, then re-evaluate before addressing secondary causes. Document the diagnosis and treatment for future reference — patterns will emerge across multiple diagnoses that improve your prompt engineering intuition.

PS the prompt is in the comments it's just too much to put up here

2 comments