When your Shiplight YAML tests start breaking after a deployment or UI change, this skill systematically reproduces failures, inspects the evidence, and applies minimal fixes without silently rewriting test intent to mask real bugs. It follows a strict precedence order: user instruction beats spec, spec beats current behavior, which means if your app contradicts the test goal or feature spec, you get a mismatch report instead of a cover-up. The workflow is reproduce, understand the failure and relevant files, inspect live browser state only when logs aren't enough, fix the smallest surface (locator, template, auth, or spec), validate and rerun, then update project memory. Best for CI repair passes or when several tests share a root cause, not for creating new tests from scratch.
npx skills add https://github.com/shiplightai/agent-skills --skill triage