[Improvement] Store Playwright Code During Execution

[shah-nikhil]

We are currently using Playwright MCP for exploratory testing. We are looking for a script that can repeatedly execute the same task and would appreciate any suggestions for improvement.

[ssddi456]

maybe the browser_generate_playwright_test tool is what youre looking for？

[shah-nikhil]

Yes, how to use this inside a cursor?

[ssddi456]

after a successy conversion, follow a prompt like 'generate playwright test'

[shah-nikhil]

Sometimes, the system provides incorrect locators for actions that have already been performed. To improve stability, I want to obtain the exact code that MCP uses at runtime.

While it works well for small tasks, when dealing with a complete workflow, it tends to generate random locators based on assumptions.

I am looking for a way to run the generated Playwright code within a single file after executing the workflow.

For example attaching reference:

[ssddi456]

maybe it is not the mcp tools‘s responce? you could use a mcp proxy to capture the tools/calls and responses to collect the code, and it will also collect some failed tries. manully modify should be done more or less

[hezhuojie]

I found the mcp will generate a random test id for the DOM that without test role or test id. So it is impossible to reuse the generated code as static e2e test

[ssddi456]

I found the mcp will generate a random test id for the DOM that without test role or test id. So it is impossible to reuse the generated code as static e2e test

should say if your code is not designed to be rpa friendly, then the code usability wont exists

[pavelfeldman]

#418 (comment) is a good advice.

[shah-nikhil]

@pavelfeldman Are we considering adding the above request as an enhancement in backlog?