QA folks — has anyone here tried using n8n for test automation workflows?

Discussions
, , ,
1

Has anyone here tried using n8n for test automation workflows

I’ve been experimenting with it recently and even set up a basic automation agent that’s showing some promise — especially for handling repetitive regression tasks.

I’m curious:

Have you integrated n8n into your QA or DevOps toolchain?
What kinds of testing workflows have you automated with it?

I’m still getting familiar with it, but it seems like it could save a lot of manual effort if scaled right. Just wondering if it’s worth going deeper — or if there are other tools that worked better for you.

Any github repo any tutorial suggestion it will highly appreciate

Would love to hear your thoughts or experiences?

4 Likes
2

Hi @bharat675, it has been on my list for a while. Planning to explore it from next week. Will share my experience here.

3

Thanks u @kunalashar

4

Hi @bharat675 - I have used it to automate Be Automation using curl. It is slow but able to give results.

5

Hi! Did it work for you? What are your impressions? Please share :folded_hands:

6

Can I get more insight from your end.Did you tried it

7

like that idea, will follow this thread as I looked into n8n for other automation task around reporting or analytics but test automation sounds way more interesting

8

Hi, no, I didn’t get a chance to try.

9

I’ve been experimenting with automating E2E tests for a large project using Playwright. We have a ton of tests, but things move so fast that maintaining them became hard enough (well there are lots of muted tests now).

My first idea was to use n8n with Playwright MCP from Microsoft - but it turned out to be way more complex than I expected. n8n isn’t very flexible for configuring real E2E flows. Actually i did only several simple workflows, without any new tests coverage, can’t even talk about maintaining flaky ones.

Then I tried using Claude Code with the same Playwright MCP. It worked better, and generated stable Playwright code, but required a lot of babysitting. kinda prompting like:

“This is a new page. Cover it E2E. Make sure selectors are stable, DOM is loaded, etc.”

I was expecting something simpler - like writing the user’s intent, e.g.:

Page should be accessible to both authenticated and unauthenticated users.

and agent generate proper test. Instead I had to think whether it selectors are stable enough, how would it run in prod CI, how it would work.

At that point, I ended up building my own MVP. It’s very early, but the goal is to lower friction: you define flows by intent, the agent builds tests, and we rerun them in github CI.

Just sharing this in case you’d like to try it out - it’s open beta.

1 Like