@muuktest/amikoo-playwright

Automatically captures DOM state and screenshots when Playwright tests fail. Configures the Amikoo MCP server and updates the Playwright fixture so test errors can be analyzed by the Amikoo Repair Agent.

The Amikoo Repair Agent auto-heals 70-80% of all broken test cases, saving days worth of time over the course of a month.

Installation

npm install @muuktest/amikoo-playwright && npx @muuktest/amikoo-playwright init

The init command sets up the MCP configuration.

How to configure the Agent in VS Code

Prerequisite: You need an access token to exercise the Amikoo Repair Agent. If you do not have such a token, please submit a request: https://muuktest.com/amikoo-playwright-mcp#section-8

1. Open VisualStudio (VS) Code.
2. Open your Playwright project folder.
3. Open the Chat Agent panel. This can be done via the menu selecting View > Chat or using the shortcut Control⌃ + Command⌘ + I (Mac) or Ctrl + Alt + I (Windows/Linux).
4. Click the Agent dropdown and select the amikoo-playwright-agent option.
5. Copilot now will use the Amikoo MCP server to repair the test cases. Refer to "How to run" for the next steps.

Note: The first time you exercise the agent, VS Code will ask you for your Access token. Enter your previously requested token. If you do not have an access token, please submit a request: https://muuktest.com/amikoo-playwright-mcp#section-8

Learn More

See the MCP server documentation for details on the agent and available tools.

How to Run

Repairing a Failing Test

1. Run your failing Playwright test as usual.
   • When the test fails, Amikoo automatically captures the necessary context data inside the test-results directory.
2. Open the Agent panel in VS Code.
3. Ask the amikoo-playwright-agent to fix the failing test.

The agent will:

• Analyze the captured error context.
• Identify the relevant files in your project.
• Propose and apply fixes to your test or related code using amikoo-mcp.
• Re-run the test to verify the issue has been resolved.

Important Notes

• The agent repairs one test at a time.
• If multiple errors exist, the agent resolves them one by one, re-running the test after each fix.
• The process ends automatically once the test passes.

Support

For support, contact [email protected].