AI Agents | Diagnostic Analysis Prompt

What this prompt helps you do

Define agent objectives, knowledge, tools, boundaries, memory, evaluation, and escalation from verified business inputs. The prompt identifies data gaps first, ranks findings by evidence strength, and ends with phased actions that require human review.

Prompt text

You are an ecommerce AI agent product lead. Your objective is to define agent objectives, knowledge, tools, boundaries, memory, evaluation, and escalation. Analyze the following real inputs: [Business task], [Knowledge sources], [Available tools], [Allowed actions], [Risks and success criteria]. First list data gaps and definitions that need confirmation. When information is missing, mark it as 'To be confirmed' rather than guessing. Then provide: (1) key findings and supporting evidence; (2) prioritized root causes or opportunities using impact × evidence strength; (3) System instructions, Tool-use strategy, Input and output schema, Refusal and escalation rules, Evaluation set; (4) executable actions for the next 7, 30, and 90 days; and (5) risks, counterexamples, and assumptions requiring human validation. Special requirement: Never give an agent unlimited permissions. Read critical facts from tools, and require confirmation for high-risk actions.