Agent Roster for AI Blueprint for Webpage Quality Assurance

The Quality Assurance Testing Blueprint uses a multi-agent system to automate the process of planning, learning, writing and testing. Each agent specializes in a specific part of the QA workflow:

Planner Agent

The Planner Agent converts JIRA cards or user stories into structured test plans. It:

• Creates a detailed test plan with common setup steps and individual test cases
• Allows asking a senior QA engineer for help with domain-specific questions
• Supports persona creation for different user roles
• Organizes tests with appropriate metadata and test data
• Outputs a structured JSON file with the complete test plan

Learn/Browser Agent

The Learn Agent interacts with the browser to understand how to execute test plans in real-world scenarios. It:

• Takes test plans created by the Planner Agent and executes them in a browser
• Records browser interactions and user flows
• Captures screenshots and DOM information for later reference
• Learns how to navigate through the application and perform test actions
• Stores learned behaviors in JSON format for the Writer Agent to use

Writer Agent

The Writer Agent converts learned behaviors into executable Playwright test code. It:

• Transforms the browser interactions captured by the Learn Agent into Python code
• Uses code search functionality to reference existing code patterns
• Supports vision capabilities for handling screenshots when needed
• Produces maintainable and readable test scripts

Test Agent

The Test Agent executes and debugs the generated tests. It:

• Runs tests with configurable parameters (headless/headed mode)
• Suggests and applies edits to test files when issues are found
• Provides detailed test execution output
• Helps debug and fix failing tests
• Integrates with the code search functionality for context

Code Search Agent

The Code Search Agent provides contextual information from repositories and code bases. It:

• Searches across multiple repositories for relevant code snippets
• Retrieves and reads file contents from repositories
• Uses vector stores for semantic search capabilities
• Helps other agents understand existing code patterns and implementations
• Provides context for test writing and debugging

These agents work together to create a complete end-to-end workflow for automated test creation, from planning to execution, making the QA process more efficient and comprehensive.