Playwright MCP Server with Kilo.ai

[heliosdoj]

Currently Playwright MCP server (QA Automation) has the capacity to use copilot, claude and OpenCode: https://playwright.dev/docs/test-agents#agent-definitions

I was able to use kilo.ai via the following command (playwright documentation):

npx playwright init-agents --loop=kiloai

That being said kilo.ai is able to read the playwright MCP server and I can select the appropriate sub agent (architect for example).

However, Microsoft Playwright dev wrote: "I tried implementing this, but it's very unclear what's the right way of storing custom modes. https://kilocode.ai/docs/features/custom-modes explains a variety of ways: json in root, yaml in root, .kilorules-{mode-slug} (no mention of the format), .kilo/rules-{mode-slug}/ (no mention of format), and on top of that it mentions sharing Modes via an Organisation. The docs file hasn't been touched in 4 months, which makes it look like there's no work happening to improve this state. All in all, I don't feel comfortable adding something that we have to maintain based on these docs.

Let's collec 👍 reactions on this issue to see how many of you would be interested in Kilo.ai support. If it's a lot of people, I'll take another look and will try to work with the Kilo.ai folks on making this work."

Please consult with the Playwright dev to integrate kiloai fully.

microsoft/playwright#38169 (comment)