SuperAPI AI Directory
Partner with us中文AI newsFavorites
Skillsprj-agent-platform-e2e-test

prj-agent-platform-e2e-test

Testing & quality

View repository

Updated 10-16 01:37

About this skill

End-to-End Functional Verification

Use agent-browser to perform end-to-end verification of the platform's core functionality, ensuring that new features work and old features are not broken.

Steps

Step 1: Start Services

Unless the user passes in --skip-start, run the following command to start all services:

./scripts/start-services.sh --langfuse

Wait for the services to be ready, and confirm that the following ports are accessible:

  • api-server :3000
  • web-frontend :5173
  • aieda-ui :5174

If the user passed in --skip-start, skip this step and go directly to testing.

Step 2: Use agent-browser to perform end-to-end testing

Use the /agent-browser skill to open the browser and execute the following test scenarios step by step.Take a screenshot after each scenario as evidence, and if a scenario fails, record the reason for the failure and continue to the next scenario.

Scenario 1: Login Verification

  1. Open http://localhost:5173/login
  2. Enter test@test.com
  3. in the email input field. Enter 12345678
  4. Click the "Login" button
  5. Verify redirection to the /chat page
  6. Save screenshot

Scenario 2: Create New Session

  1. On the /chat page, find the "New Chat" button in the sidebar (Plus icon)
  2. Click "New Chat"
  3. Verify that a new empty conversation appears in the chat area
  4. Save screenshot

Scenario 3: Send Message and Chat

  1. In the chat input box (textarea), enter 你好，请介绍一下你自己
  2. Click the send button or press Enter to send
  3. Wait for the AI reply (wait up to 30 seconds)
  4. Verify that the user's sent message and the AI's reply appear in the message list
  5. Save screenshot

Scenario 4: Multi-turn Conversation

  1. In the same conversation, continue by entering 你能做什么？
  2. Send the message
  3. Wait for the AI reply (wait up to 30 seconds)
  4. Verify that the message list contains two user messages and two AI replies
  5. Save screenshot

Scenario 5: Verify New Features

  1. Check the recent git commit records (git log --oneline -10) to identify recently added features
  2. Based on the new features, design the corresponding test operations
  3. Execute the test operations and verify that the features are available
  4. Save screenshot

Step 3: Summarize Test Results

After all scenarios have been executed, output a test report:

## 端到端功能验证报告

| 场景 | 状态 | 备注 |
|------|------|------|
| 登录验证 | PASS/FAIL | ... |
| 新建会话 | PASS/FAIL | ... |
| 发送消息与聊天 | PASS/FAIL | ... |
| 多轮对话 | PASS/FAIL | ... |
| 新功能验证 | PASS/FAIL | ... |

If there are any FAIL items, list the specific reasons for failure and possible directions for fixes.

Notes

  • Starting the services takes time; wait for the health check to pass before starting testing
  • Between agent-browser operations, wait an appropriate amount of time for page loading to avoid operating before elements have finished rendering
  • After a chat message is sent, the AI reply may take a long time, so set a reasonable wait timeout
  • If login fails, do not continue with subsequent scenarios; report the login issue directly
  • Screenshots are used as evidence to help locate issues