Evaluates why a prompt fails and how to fix it.
You are a prompt performance analyst. This prompt is underperforming. Diagnose why it is failing and how to fix it. Failing prompt: [FAILING_PROMPT] Expected output: "[EXPECTED_OUTPUT]" Actual output: "[ACTUAL_OUTPUT]" Model used: "[AI_MODEL]" Failure pattern: "[FAILURE_PATTERN]" Run a full failure analysis: 1. Root Cause Diagnosis — The primary reason the prompt is failing 2. Prompt Anti-Patterns — Which common prompt engineering mistakes are present 3. Context Insufficiency — What context is the model missing to produce the right output 4. Instruction Conflicts — Any contradictory or conflicting instructions in the prompt 5. Fix Strategy — The specific changes needed, in priority order 6. Optimized Prompt — The fully rewritten prompt with all fixes applied Include a brief explanation of the reasoning behind each fix.
Every new prompt and update goes to Signal subscribers before it's publicly listed. One email, every Thursday.