SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillse2e-qa-tester

e2e-qa-tester

Agents & models

View repository

Updated 10-15 16:43

About this skill

E2E QA Tester

Run end-to-end tests and manual QA using Playwright MCP to verify recently implemented features.

Workflow

1. Identify the Task to Test

Before running any test, review the history of the current conversation to identify the last completed task:

Busca en los mensajes recientes:
- Tareas marcadas como "completadas" o "done"
- Implementaciones finalizadas
- Features agregadas o modificadas
- Bug fixes aplicados

If you cannot clearly identify the task, ask the user: "What functionality or flow should I test?"

2. Find Test Credentials

ALWAYS before attempting login, look for test credentials:

# Buscar archivo de credenciales en el proyecto
Get-ChildItem -Path . -Filter "CREDENTIALS.md" -Recurse

Common locations:

  • CREDENTIALS.md (project root)
  • docs/CREDENTIALS.md
  • .credentials/CREDENTIALS.md
  • testing/CREDENTIALS.md

Expected format in CREDENTIALS.md:

## Usuario Admin
- Email: admin@test.com
- Password: admin123

## Usuario Regular
- Email: user@test.com
- Password: user123

If you do NOT find CREDENTIALS.md:

  1. Stop and ask the user for the credentials
  2. Do NOT try to guess or use generic credentials

3. Connect to the Application

Try to connect to the default port:

# Verificar si puerto 5173 esta en uso
Test-NetConnection -ComputerName 127.0.0.1 -Port 5173 -InformationLevel Quiet
  • If port 5173 is busy: Use that connection
  • If port 5173 is NOT in use:
    1. Check whether there is another common port (3000, 4200, 8080)
    2. If there is no active port, ask the user

4. Pre-Test Confirmation (MANDATORY)

Before running any test, present to the user:

## Plan de Prueba

**Tarea identificada**: [descripcion de la tarea completada]

**Flujo a probar**:
1. [Paso 1]
2. [Paso 2]
3. [Paso 3]

**Credenciales a usar**: [rol/usuario del CREDENTIALS.md]

**URL de inicio**: http://127.0.0.1:[puerto]

¿Procedo con esta prueba? (s/n)

Do NOT proceed until you receive explicit confirmation from the user.

If the user rejects:

  1. Ask what should be done differently
  2. Adjust the plan according to their instructions
  3. Ask for confirmation again

5. Run the Test

Once confirmed, use Playwright MCP to run the test:

Start Browser (headless)

playwright_browser_navigate con url http://127.0.0.1:[puerto]

If Login is required

  1. Take a snapshot to see the current state
  2. Identify email/username and password fields
  3. Fill out the form with credentials from CREDENTIALS.md
  4. Submit the form
  5. Verify successful login (snapshot, look for success element)

Run Test Flow

  1. Navigate to the target functionality
  2. Interact with elements according to the identified flow
  3. Take snapshots at key points
  4. Verify expected results

Useful Playwright Commands

Action Tool
View current state playwright_browser_snapshot
Navigate playwright_browser_navigate
Click playwright_browser_click
Type text playwright_browser_type
Select dropdown playwright_browser_select_option
Fill form playwright_browser_fill_form
Wait playwright_browser_wait_for
Screenshot playwright_browser_take_screenshot

6. Report Results

When finished, present a clear report:

## Resultado de Prueba E2E

**Estado**: [PASO / FALLO]

**Tarea probada**: [descripcion]

**Pasos ejecutados**:
1. [Paso 1] - [OK/FALLO: razon]
2. [Paso 2] - [OK/FALLO: razon]
3. [Paso 3] - [OK/FALLO: razon]

**Resultado final**:
[Descripcion de que paso, si hubo errores, comportamiento observado]

**Evidencia**:
- [Screenshots tomados o pasos verificados]

Common Use Cases

Test New Form

  1. Identify form fields
  2. Fill in with valid test data
  3. Submit and verify successful response
  4. Test validations (empty fields, invalid formats)

Test Authentication Flow

  1. Verify login screen
  2. Use credentials from CREDENTIALS.md
  3. Verify post-login redirect
  4. Verify elements visible only to authenticated users

Test CRUD

  1. Create: Create new record, verify appearance in list
  2. Read: Verify data displayed correctly
  3. Update: Edit record, verify changes persisted
  4. Delete: Delete record, verify it disappears from list

Test Feature Toggle

  1. Verify state before activating feature
  2. Activate/trigger the feature
  3. Verify state afterward (UI visible, behavior changed)

Error Handling

If the application does not load

  1. Verify that the server is running
  2. Verify correct URL and port
  3. Report to the user

If login fails

  1. Verify credentials in CREDENTIALS.md
  2. Verify that the form has the correct fields
  3. Report specific error to the user

If an element is not found

  1. Take a snapshot to see the current state
  2. Look for alternative selectors
  3. If it persists, report to the user with evidence

If there are console errors

  1. Use playwright_browser_console_messages to get errors
  2. Include in results report

Important Notes

  • ALWAYS use headless mode (--headed=false) to avoid interfering with the user
  • ALWAYS look for CREDENTIALS.md before asking for credentials
  • ALWAYS ask for confirmation before running tests
  • NEVER create, modify, or delete test users
  • NEVER use IPv6, always 127.0.0.1