Core responsibilities
- Translate user stories into comprehensive test plans
- Author and maintain automated browser tests
- Validate data integrity across Oracle, Snowflake, and other systems
- Investigate defects and reproduce bugs in test environments
- Sign off on stories before they ship to production
Where AI saves you the most time
- Generating positive, negative, and edge test cases from a single story
- Authoring Playwright scripts in plain English via MCP
- Validating database rows, counts, and field values without writing SQL
- Comparing schemas and data across environments in seconds
- Auto-summarizing flaky-test patterns across runs
Key AI tools for this role
- Playwright MCP — browser automation in natural language
- Oracle MCP — database validation in plain English
- Snowflake MCP — data warehouse checks
- GitHub Copilot — test case and script generation
.prompt.md files inside .github/prompts/ in your VSCode workspace. Click any template to expand it.
generate-test-cases.prompt.md
# generate-test-cases.prompt.md --- title: Test Case Generation from User Story description: "Generates comprehensive positive, negative, and edge case test cases from any user story." --- # Role As a Senior QA Engineer, generate comprehensive test cases for the user story attached. # Coverage Required Positive test cases: Happy path scenarios where all inputs are valid and expected behavior occurs Negative test cases: Invalid inputs, missing required fields, unauthorized access attempts Edge cases: Boundary values, empty states, maximum limits, concurrent actions # Output Format — for each test case Test Case ID: TC-[number] Title: [short descriptive name] Type: Positive | Negative | Edge Case Preconditions: [what must be true before the test runs] Steps: [numbered step-by-step actions] Expected Result: [what should happen] Priority: High | Medium | Low # Notes Map each test case back to the relevant acceptance criteria. Flag any missing or ambiguous acceptance criteria in the user story.
playwright-validate.prompt.md
# playwright-validate.prompt.md --- title: Playwright MCP Browser Automation description: Drives browser-based UI validation using natural language via Playwright MCP. --- # Role You are a senior QA automation engineer using Playwright MCP to validate UI behavior. # Task — validate the following test scenario Validate the below using MCP Playwright: a. Navigate to [URL] and confirm the page loads successfully b. Observe the initial state — describe what is visible on screen c. Perform the action: [click / select / enter / submit] d. Verify the expected outcome: [what should appear or change] e. Report PASS or FAIL with a screenshot description # Rules Wait for elements to be fully loaded before interacting. If an element is not found, report the exact error message. For each step, confirm the action was executed before proceeding to the next.
oracle-validate.prompt.md
# oracle-validate.prompt.md --- title: Oracle Database Validation via MCP description: Validates test data integrity in Oracle using plain-English queries. --- # Role Use the Oracle MCP server to validate test data in the database. # Validation Tasks 1. Check how many entries in [TABLE_NAME] have [FIELD] = [VALUE] 2. Verify that [RECORD_ID] exists with the correct field values after the test action 3. Confirm no orphaned records exist in [CHILD_TABLE] after deletion of [PARENT_RECORD] 4. Compare row counts between [TABLE_A] and [TABLE_B] and highlight any differences # Output Format For each check: state the query used, the result returned, and PASS or FAIL status. Flag any unexpected results with a suggested next step for investigation.
Playwright MCP
Browser Automation
Drive browser automation in plain English. Playwright MCP handles selector strategy, waits, and reporting — you describe the test scenario.
When to use
Any time you need to author or run a browser test.
Oracle MCP
Database Validation
Validate test data in Oracle using natural-language queries — confirm row counts, foreign keys, and field values without writing SQL.
When to use
When validating data integrity after test runs.
Snowflake MCP
Data Warehouse Validation
Confirm warehouse aggregations and ETL completeness using plain-English questions in Copilot Chat.
When to use
Validating data warehouse outputs after pipeline runs.
GitHub Copilot
Test Generation
Your primary tool for generating test cases, test data, and Playwright scripts.
When to use
Every day — for any test design or generation task.
Getting set up — 6 steps
Request GitHub Copilot Access
Submit an access request through the Cisco AppStore. Install VSCode 1.109+ from the AppStore.
Install Copilot Extensions
In VSCode marketplace, install GitHub Copilot and GitHub Copilot Chat. Authenticate with your GitHub account.
Configure Playwright MCP
CTRL+SHIFT+P → "mcp" → open mcp.json. Add the Playwright MCP server config. Start the server and confirm it is running in the MCP panel.
Set Up Oracle MCP
Install the Oracle SQL Developer VSCode extension. Connect to Cisco VPN. Contact [email protected] for test Oracle instance details. Add Oracle MCP to mcp.json.
Create Your Prompt File Library
Create .github/prompts/ in your workspace. Copy the three QA prompt templates from the Prompt Templates tab to get started immediately.
Run Your First Test Generation
Open a user story in VSCode, drag it into Copilot Chat, and run /generate-test-cases. Select Claude Sonnet 4.5 for best results. Review and iterate in the same session.
Prompting best practices for QA
Assign a Senior QA Role
"As a Senior QA Engineer" frames Copilot to produce professional-grade test coverage, not surface-level cases.
Always Attach the User Story
Drag the user story .md file into Copilot Chat. Without context, test cases will be generic and miss acceptance criteria.
Request All Three Types
Always explicitly ask for positive, negative, AND edge cases. Copilot defaults to happy path only without this instruction.
Iterate on Coverage Gaps
After the first pass, ask: "What edge cases around [specific scenario] did you miss?" to deepen coverage.
Be Specific with Playwright
Describe exact UI states to validate. "Verify only 2 partners shown" is better than "verify the filter works."
Validate Output Before Using
AI-generated test cases are a strong first draft. Always review against the acceptance criteria before adding to your test suite.
Why this matters
Step-by-step guide
Tools required