Muraqib: nightly Playwright + auto-fix, built on this action

[holistis]

Sharing something built on top of claude-code-action: Muraqib runs Playwright against a live app every night, and when a test fails it opens a fix PR using this action, with one weekly summary email. MIT licensed, no hosted backend.

Most interesting bug from building it: the failure-context extraction was silently reading the wrong nesting level of the Playwright JSON report, so for months the auto-fix issues reached this action with zero actual failure details. Fixed now, documented in the repo lessons file.

https://github.com/holistis/muraqib

No ask here, just thought it was a genuine real-world use of this action worth sharing.