Skip to content

TOP 5 Prompt Frameworks Introduction

中文版本

📚 Overview

This document provides detailed introductions to the TOP 5 prompt frameworks most commonly used and effective in QA testing scenarios. These frameworks have been validated through practice and can significantly improve the quality and efficiency of AI-assisted testing.

🏆 TOP 5 Framework Rankings

Based on practicality, usability, output quality, and applicability, we recommend the following TOP 5 frameworks:

  1. ROSES - Most Complete Framework ⭐⭐⭐⭐⭐
  2. CRISPE - Most In-Depth Framework ⭐⭐⭐⭐⭐
  3. LangGPT - Most Systematic Framework ⭐⭐⭐⭐
  4. ICIO - Most Data-Driven Framework ⭐⭐⭐⭐
  5. RISE - Most Concise Framework ⭐⭐⭐⭐

1️⃣ ROSES Framework

📖 Framework Overview

Full Name: Role Objective Scenario Expected Solution Steps

Creator: Community Contribution

Core Philosophy: Build complete testing prompts by clarifying role, objective, scenario, expected solution, and execution steps

🎯 Framework Structure

markdown
Role: Specify the AI's role
Objective: State the goal to achieve
Scenario: Describe the specific testing scenario
Expected Solution: Define the desired output
Steps: Ask for specific steps needed to achieve the goal

✨ Core Advantages

AdvantageDescription
CompletenessCovers all key elements of testing
StructuredClear five-step structure, easy to understand
ActionableClear step guidance, easy to execute
AdaptableSuitable for various complex testing scenarios

📝 Practical Examples

Example 1: API Performance Testing

markdown
Role: You are a performance testing expert with 10 years of experience
Objective: Design a complete API performance testing plan
Scenario: During promotional events on an e-commerce platform, API call volume surges, requiring verification of system performance under high load
Expected Solution: Provide a complete plan including load testing, stress testing, and stability testing, with specific test metrics and pass criteria
Steps: 
1. Analyze business scenarios and user behavior patterns
2. Design load models and test data
3. Develop performance testing strategy
4. Write automated test scripts
5. Execute tests and analyze results
6. Provide performance optimization recommendations

Example 2: Security Testing

markdown
Role: You are a senior security testing engineer, proficient in OWASP Top 10
Objective: Conduct comprehensive security testing on a web application
Scenario: A financial system requires security assessment before going live to ensure compliance with security standards
Expected Solution: Provide detailed security testing report including vulnerability discovery, risk assessment, and remediation recommendations
Steps:
1. Conduct threat modeling analysis
2. Execute automated vulnerability scanning
3. Perform manual penetration testing
4. Verify authentication and authorization mechanisms
5. Test data encryption and transmission security
6. Write security testing report

🎯 Applicable Scenarios

  • ✅ Complex test project design
  • ✅ Tasks requiring detailed step guidance
  • ✅ Multi-phase test plans
  • ✅ Team collaboration test plans

💡 Usage Tips

  1. Clear Role Definition: Specify experience level and expertise
  2. Specific Scenario Description: Provide sufficient background and constraints
  3. Clear Expected Solution: Define output format and quality standards
  4. Actionable Steps: Ensure each step is executable

2️⃣ CRISPE Framework

📖 Framework Overview

Full Name: Capacity and Role, Insight, Statement, Personality, Experiment

Creator: Matt Nigh

Core Philosophy: Obtain deep and diverse testing solutions through capability positioning, insight analysis, task statement, personalized style, and experimental output

🎯 Framework Structure

markdown
Capacity and Role: Define AI's capabilities and role
Insight: Provide background insights and context
Statement: Clarify the task to complete
Personality: Define output style and personality
Experiment: Request multiple examples or solutions

✨ Core Advantages

AdvantageDescription
Deep AnalysisEmphasizes insight, provides in-depth analysis
PersonalizationSupports customized output styles
DiversityObtains multiple solutions through experiment element
ProfessionalismEmphasizes capability and professional background

📝 Practical Examples

Example 1: Performance Bottleneck Analysis

markdown
Capacity and Role: You are a performance optimization expert with 15 years of experience, proficient in system architecture and performance tuning
Insight: Performance bottlenecks often hide in system critical paths and require multi-dimensional analysis for accurate identification. Database queries, network latency, and code logic can all become bottleneck points
Statement: Analyze system performance test results, identify performance bottlenecks, and provide optimization solutions
Personality: Analyze with professional and rigorous attitude, provide data-supported conclusions and feasible optimization recommendations
Experiment: Provide at least 3 different performance optimization solutions, including short-term quick optimizations and long-term architectural optimizations

Example 2: Test Strategy Design

markdown
Capacity and Role: You are a test architect skilled in designing enterprise-level testing systems
Insight: Effective test strategies need to balance test coverage, cost, and time. The test pyramid theory emphasizes many unit tests, moderate integration tests, and few UI tests
Statement: Design a complete test strategy for an e-commerce platform with microservices architecture
Personality: Plan with strategic thinking, focus on practicality and implementability
Experiment: Provide differentiated test strategies for different project phases (MVP, growth, maturity)

🎯 Applicable Scenarios

  • ✅ Testing tasks requiring deep analysis
  • ✅ Scenarios requiring professional insights
  • ✅ Need multiple solution comparisons
  • ✅ Specific output style requirements

💡 Usage Tips

  1. Emphasize Capability: Clarify expertise and experience depth
  2. Provide Insights: Share industry best practices and key considerations
  3. Personalize Output: Specify desired expression style and professionalism
  4. Request Experiments: Clarify number of solutions and differentiation requirements

3️⃣ LangGPT Framework

📖 Framework Overview

Full Name: LangGPT Structured Prompt Framework

Creator: Yunzhong Jiangshu (Liu Yuanzhong)

Core Philosophy: Make prompt writing as standardized and maintainable as programming through templating and structuring

🎯 Framework Structure

markdown
# Role: Role Name

## Profile:
- author: Author
- version: Version
- language: Language
- description: Role description

## Goals: 
- Goal 1
- Goal 2

## Constrains:
- Constraint 1
- Constraint 2

## Skills:
- Skill 1
- Skill 2

## Workflows:
1. Workflow step 1
2. Workflow step 2

## Initialization:
Initialization prompt

✨ Core Advantages

AdvantageDescription
SystematicMost complete structured framework
MaintainableTemplate design, easy to update
ReusableSupports variables and modularization
StandardizedUnified format specification

📝 Practical Example

Example: Automation Testing Framework Designer

markdown
# Role: Automation Testing Framework Designer

## Profile:
- author: QA Team
- version: 1.0
- language: English
- description: Focused on designing and building high-quality, maintainable automation testing frameworks

## Goals:
- Design automation testing framework suitable for project characteristics
- Provide complete framework architecture and technology selection recommendations
- Ensure framework scalability and maintainability
- Provide detailed implementation guidelines and best practices

## Constrains:
- Framework design must consider team technology stack
- Need to balance automation cost and benefits
- Follow test pyramid principles
- Ensure reasonable framework learning curve

## Skills:
- Proficient in mainstream automation testing tools (Selenium, Appium, Cypress, etc.)
- Familiar with various programming languages (Java, Python, JavaScript)
- Understand CI/CD integration and DevOps practices
- Master test framework design patterns (Page Object, Data-Driven, etc.)

## Workflows:
1. Analyze project characteristics and technology stack
2. Evaluate team capabilities and resources
3. Select appropriate testing tools and frameworks
4. Design framework architecture and directory structure
5. Define coding standards and best practices
6. Provide sample code and documentation
7. Develop implementation plan and training program

## Initialization:
Welcome! I am an Automation Testing Framework Designer. Please provide your project information, and I will design the most suitable automation testing framework for you.

🎯 Applicable Scenarios

  • ✅ Long-term maintained testing frameworks
  • ✅ Standardized team collaboration projects
  • ✅ Reusable test template development
  • ✅ Enterprise-level test system construction

💡 Usage Tips

  1. Version Management: Use version field to track framework versions
  2. Modular Design: Break complex tasks into independent Skills and Workflows
  3. Clear Constraints: List all limitations in Constrains
  4. Friendly Initialization: Provide clear Initialization guidance

4️⃣ ICIO Framework

📖 Framework Overview

Full Name: Instruction, Context, Input Data, Output Indicator

Creator: Community Contribution

Core Philosophy: Emphasize data flow, achieve data-driven testing through clear input/output and evaluation metrics

🎯 Framework Structure

markdown
Instruction: Clear task instruction
Context: Relevant background information
Input Data: Specific input data or parameters
Output Indicator: Expected output and evaluation criteria

✨ Core Advantages

AdvantageDescription
Data-OrientedEmphasizes clarity of input/output
QuantifiableProvides clear evaluation metrics
Concise & EfficientSimple structure, easy to use
Suitable for API TestingNaturally fits interface testing scenarios

📝 Practical Examples

Example 1: API Interface Testing

markdown
Instruction: Design complete test cases for user login API
Context: RESTful API, uses JWT authentication, supports email and phone login
Input Data:
- Normal data: {"email": "[email protected]", "password": "Pass123!"}
- Boundary data: Minimum/maximum length passwords
- Exception data: Invalid email format, empty password
- Security data: SQL injection, XSS attack strings
Output Indicator:
- Successful login: Return 200 status code and valid JWT token
- Failed login: Return 401 status code and error message
- Response time: < 500ms
- Error rate: < 0.1%

Example 2: Database Performance Testing

markdown
Instruction: Evaluate database query performance and provide optimization recommendations
Context: MySQL 8.0, order table with 10 million records, needs to support complex queries
Input Data:
- Query types: Single table query, multi-table JOIN, aggregate query
- Concurrency: 100/500/1000 concurrent users
- Data volume: Current 10 million, expected to grow to 50 million
Output Indicator:
- Query response time: P95 < 100ms, P99 < 500ms
- QPS: > 1000
- CPU usage: < 70%
- Slow query count: 0

🎯 Applicable Scenarios

  • ✅ API interface testing
  • ✅ Database testing
  • ✅ Performance metric validation
  • ✅ Data-driven testing

💡 Usage Tips

  1. Complete Input Data: Cover normal, boundary, exception, security data
  2. Quantifiable Metrics: Use specific numbers and percentages
  3. Sufficient Context: Provide adequate technical background
  4. Clear Instructions: Clearly state expected output format

5️⃣ RISE Framework

📖 Framework Overview

Full Name: Role, Input, Steps, Expectation

Creator: Community Contribution

Core Philosophy: Quickly build effective testing prompts with the most concise structure

🎯 Framework Structure

markdown
Role: Define AI's role
Input: Provide input information
Steps: Explain execution steps
Expectation: Describe expected output

✨ Core Advantages

AdvantageDescription
ConcisenessMost streamlined framework structure
EfficiencyQuick to start, immediate use
FlexibilityAdapts to various quick tasks
Easy to LearnLowest learning cost

📝 Practical Examples

Example 1: Quick Bug Verification

markdown
Role: You are a test engineer
Input: User reports unable to enter password on login page in Chrome browser
Steps:
1. Open login page in Chrome browser
2. Try entering content in password field
3. Check browser console for errors
4. Verify if same issue exists in other browsers
Expectation: Provide bug reproduction steps, root cause analysis, and fix recommendations

Example 2: Simple Functional Testing

markdown
Role: You are a functional tester
Input: Test shopping cart add product functionality
Steps:
1. Browse product list
2. Click "Add to Cart" button
3. Verify cart quantity update
4. Open cart to view product information
Expectation: List all test points and expected results, including positive and negative scenarios

🎯 Applicable Scenarios

  • ✅ Urgent tasks
  • ✅ Simple testing scenarios
  • ✅ Quick validation
  • ✅ Ad-hoc testing needs

💡 Usage Tips

  1. Concise Role: Simple and clear role definition
  2. Specific Input: Provide key input information
  3. Clear Steps: List core execution steps
  4. Clear Expectation: Clearly state expected output

📊 Comprehensive Framework Comparison

Complexity Comparison

Simple ←―――――――――――――――――――――――――――→ Complex
RISE < ICIO < Standard < ROSES < CRISPE < LangGPT

Output Quality Comparison

Basic ←―――――――――――――――――――――――――――→ Advanced
RISE < Standard < ICIO < LangGPT < ROSES = CRISPE

Learning Curve Comparison

Easy ←―――――――――――――――――――――――――――→ Difficult
RISE < Standard < ICIO < ROSES < CRISPE < LangGPT

🎯 Selection Recommendation Matrix

NeedPreferred FrameworkAlternative Framework
Quick TasksRISEStandard
Complex ProjectsROSESCRISPE
Deep AnalysisCRISPEROSES
SystematicLangGPTROSES
API TestingICIOStandard
Security TestingCRISPEROSES
Performance TestingROSESLangGPT
Beginner UseStandardRISE

Beginner Path (1-2 weeks)

  1. Standard Framework - Master basic structure
  2. RISE Framework - Learn simplified application
  3. Practice: Complete 5 simple testing tasks

Intermediate Path (1 month)

  1. ICIO Framework - Understand data-driven approach
  2. ROSES Framework - Master complete process
  3. Practice: Complete 3 medium complexity projects

Advanced Path (2-3 months)

  1. CRISPE Framework - Deep analysis capability
  2. LangGPT Framework - Systematic thinking
  3. Practice: Build enterprise-level testing system

Last Updated: 2026-01-28 Version: v1.0