PromptBuilder Guide

A comprehensive guide to crafting effective prompts for AI language models.

Core Components

A prompt can be broken down into these essential components:

=[Task, Format, Voice, Context]

Where:

  • Task (Required): What you want done
  • Context (Required): Background information and constraints
  • Format (Optional): How the output should be structured
  • Voice (Optional): Tone and style of communication

Basic Structure

Minimal Prompt Template

=[Task: {action}, Context: {background}]

Complete Prompt Template

=[Task: {action}, Format: {structure}, Voice: {style}, Context: {background}]

Component Details

Task

  • Must include a clear verb/action
  • Should be specific and measurable
  • Examples: "summarize", "analyze", "write", "create", "explain"

Context

  • Anything you type in or paste
  • Relevant background information
  • Constraints and limitations
  • Source materials or references
  • Domain-specific knowledge

Format (Optional)

  • Output structure (bullet points, paragraphs, table)
  • Length requirements
  • Special formatting needs
  • File type or medium

Voice (Optional)

  • Tone (formal, casual, technical)
  • Perspective (first person, third person)
  • Target audience
  • Brand voice alignment

Syntax Variations

Verbose Style

Please {task} with the following parameters:
- Format: {desired structure}
- Voice: {tone and style}
- Context: {background information}

XML Style

<prompt>
  <task>action to perform</task>
  <format>desired structure</format>
  <voice>tone and style</voice>
  <context>background information</context>
</prompt>

JSON Style

{
  "task": "action to perform",
  "format": "desired structure",
  "voice": "tone and style",
  "context": "background information"
}

Custom Instructions Integration

Custom instructions can be used to set persistent preferences for:

  • Default voice/tone
  • Output format preferences
  • Domain expertise
  • Ethical constraints
  • Language preferences

Example:

Custom Instructions:
1. Always use professional tone
2. Prefer bullet points for lists
3. Include source citations when possible
4. Use metric units
5. Write at university level

Best Practices

  1. Be Specific

    • Use clear, measurable terms
    • Provide concrete examples
    • State explicit constraints

  2. Iterate

    • Start simple
    • Refine based on results
    • Build complexity gradually

  3. Context Matters

    • Include relevant background
    • Specify audience
    • Define scope

  4. Format Clearly

    • State structural requirements
    • Define length constraints
    • Specify output format

Example Prompts

Basic

=[Task: write a product description, Context: eco-friendly water bottle]

Complete

=[Task: write a product description, Format: 3 paragraphs with bullet points, Voice: enthusiastic and eco-conscious, Context: stainless steel water bottle with recycled materials]

Professional Email

=[Task: draft an email, Format: business letter, Voice: professional and concise, Context: meeting followup with action items from Project X discussion]

Emerging Standards

  1. Role-Based Prompting

    As a {role}, {task} for {audience}...

  2. Chain-of-Thought

    Think through this step by step:
1. First, consider...
2. Then analyze...
3. Finally, conclude...

  3. Few-Shot Learning

    Example 1: {input} -> {output}
Example 2: {input} -> {output}
Now do: {new input}

Tips for Success

  1. Start with required components (Task and Context)
  2. Add Format and Voice as needed
  3. Use consistent syntax within projects
  4. Iterate and refine based on results
  5. Document successful prompt patterns
  6. Build a prompt library for common tasks
  7. Use custom instructions for persistent preferences

Troubleshooting

If the output isn't what you expected:

  1. Verify task clarity
  2. Add more specific context
  3. Adjust format requirements
  4. Modify voice/tone
  5. Break into smaller subtasks
  6. Use example outputs
  7. Try alternative syntax

PLINQ (Prompt Integrated Natural Query) Specification

Overview

PLINQ is a structured approach to prompt engineering that borrows concepts from SQL to create more precise and predictable prompt outcomes. It provides a formal way to compose prompts using familiar database-like operations.

Basic Syntax

Query Structure

{OutputType}_BY_{Constraint}_WITH_{Parameter}_IN_{Context}

Basic Operators

  • _BY_: Primary filter/constraint
  • _WITH_: Additional parameters
  • _IN_: Context specification
  • _AS_: Output format
  • _FOR_: Target audience/purpose

Common Patterns

Recipe Generation

Recipe_BY_Ingredients_WITH_Time(30min)_IN_ItalianCuisine

Translates to: "Create a recipe using these ingredients, taking 30 minutes, in Italian cuisine style"

Content Creation

BlogPost_BY_Topic(AI)_WITH_Length(1000words)_AS_Tutorial

Translates to: "Write a 1000-word blog post about AI in tutorial format"

Analysis

Analysis_BY_Metrics(ROI,Growth)_IN_Q4Report_AS_ExecutiveSummary

Translates to: "Analyze ROI and Growth metrics from Q4 Report as an executive summary"

Complex Queries

Multiple Constraints

MealPlan_BY_Diet(Vegan)_AND_BY_Calories(1500)_WITH_Shopping(List)

Nested Contexts

Email_BY_Purpose(Meeting)_IN_(Project(Alpha)_WITH_Team(Development))

Modifiers and Parameters

Format Modifiers

{Content}_AS_{Format}
- _AS_Bullet
- _AS_Table
- _AS_Paragraph
- _AS_JSON

Audience Modifiers

{Content}_FOR_{Audience}
- _FOR_Experts
- _FOR_Beginners
- _FOR_Stakeholders

Examples in Practice

Content Creation

Article_BY_Topic(AI Ethics)_WITH_Length(2000)_FOR_GeneralAudience_AS_Discussion

Generates: An article about AI Ethics, 2000 words long, written for general audience in discussion format

Technical Documentation

Documentation_BY_Feature(Authentication)_IN_Project(UserAPI)_AS_TechnicalSpec

Generates: Technical specification for authentication feature in UserAPI project

Business Analysis

Report_BY_Metrics(Revenue,Growth)_IN_Q2Data_WITH_Visuals(Charts)_FOR_Executives

Generates: Executive report analyzing revenue and growth from Q2 data, including charts

Advanced Features

Temporal Parameters

Forecast_BY_Data(Sales)_IN_TimeRange(2024Q1-2024Q4)_WITH_Confidence(95)

Conditional Logic

Content_BY_IF(Metric > Threshold)_THEN_Style(Positive)_ELSE_Style(Neutral)

Best Practices

  1. Consistency: Use consistent naming conventions for parameters
  2. Clarity: Keep queries readable and logical
  3. Modularity: Break complex queries into smaller, reusable parts
  4. Documentation: Comment complex queries for future reference
  5. Validation: Include parameter type checking where possible

Use Cases

Content Generation

Blog_BY_Keywords(AI,ML)_WITH_SEO(High)_FOR_TechAudience
Newsletter_BY_Industry(Tech)_WITH_Frequency(Weekly)
Tutorial_BY_Topic(Python)_FOR_Beginners_WITH_Examples

Business Documentation

Proposal_BY_Service(Consulting)_WITH_Budget(100K)_AS_PDF
Contract_BY_Type(NDA)_WITH_Party(Client)_IN_Legal(US)
Report_BY_Department(Sales)_WITH_Period(Q3)_AS_Dashboard

Creative Writing

Story_BY_Genre(SciFi)_WITH_Length(Short)_FOR_YoungAdults
Poetry_BY_Style(Haiku)_WITH_Theme(Nature)_IN_Season(Spring)
Script_BY_Format(Dialogue)_WITH_Characters(2)_IN_Setting(Future)