AI Prompt Engineering

Act as an expert AI Prompt Engineering consultant. Help me improve a weak prompt for a specific LLM task. Context: - Target model/tool: [ChatGPT / Claude / Gemini / Midjourney / other] - Use case: [blog writing / customer support / coding / research / image generation / automation] - Current prompt: [paste current prompt] - Desired output: [describe the ideal response] - Audience or end user: [describe audience] - Constraints: [tone, length, format, compliance, sources, data privacy, etc.] Your tasks: 1. Diagnose the current prompt and identify why it may produce vague, irrelevant, inconsistent, or hallucinated outputs. 2. Rewrite the prompt using best practices: clear role, goal, context, constraints, examples, output format, and verification steps. 3. Create 3 improved prompt variations: beginner-friendly, advanced, and production-ready. 4. Add a scoring rubric with criteria such as clarity, specificity, hallucination risk, formatting control, and task alignment. 5. Recommend which version to use first and explain why. Return the result in this structure: - Problem diagnosis - Optimized prompt versions - Scoring rubric - Recommended prompt - Next testing steps

A refined, high-performing prompt set that improves relevance, structure, consistency, and factual reliability while giving the user a clear scoring rubric for future optimization.

Act as a prompt testing strategist. Design an A/B testing workflow to compare multiple prompt versions and identify the highest-performing one. Context: - Task type: [content creation / coding assistance / research summary / support reply / visual prompt / data analysis] - Goal of the AI output: [describe desired result] - Baseline prompt: [paste prompt] - Prompt variants to test, if any: [paste variants or ask AI to create them] - Evaluation criteria: [accuracy, creativity, tone, completeness, conversion potential, factuality, speed, etc.] - Number of test cases: [e.g., 5, 10, 25] - Target users or reviewers: [team, customers, internal QA, subject-matter experts] Your tasks: 1. Generate or refine 3-5 prompt variants for the same task. 2. Create a test plan with sample inputs, expected outputs, and success metrics. 3. Build a scoring matrix to compare results objectively. 4. Include instructions for detecting hallucinations, weak reasoning, formatting failures, and brand/tone mismatches. 5. Recommend how to select a winning prompt and document the decision. Return the result in this structure: - Prompt variants - Test cases - Scoring matrix - Evaluation instructions - Winner selection method - Iteration recommendations

A complete A/B testing plan that allows users to compare prompt variants objectively, measure output quality, reduce hallucinations, and select the best-performing prompt based on clear metrics.

Act as an AI Prompt Engineering operations advisor. Build a reusable prompt library and governance system for a team using LLMs across multiple workflows. Context: - Organization/team type: [marketing / product / engineering / support / research / enterprise] - Main AI use cases: [list use cases] - Models/tools used: [ChatGPT, Claude, Midjourney, internal LLM, automation platform, etc.] - Security requirements: [public data only / confidential / regulated / enterprise-grade] - Collaboration needs: [version control, approvals, role-based access, audit logs, shared templates] - Current pain points: [inconsistent outputs, duplicated prompts, lack of testing, privacy concerns, no ownership] Your tasks: 1. Design a categorized prompt library structure for the team. 2. Create reusable prompt templates with placeholders for common tasks. 3. Define a versioning and approval workflow for prompt changes. 4. Recommend prompt metadata fields such as owner, use case, model, version, date tested, risk level, and performance score. 5. Add governance rules to reduce hallucinations, protect sensitive data, and avoid vendor lock-in. 6. Provide a rollout plan for adoption and ongoing improvement. Return the result in this structure: - Prompt library categories - Reusable template examples - Metadata and versioning system - Governance rules - Testing and review workflow - Rollout plan

A scalable prompt management framework with reusable templates, governance rules, version control, collaboration practices, and security safeguards for teams using AI across multiple workflows.