Build a repeatable eval loop that grades agent output with an LLM judge, so prompt/skill changes get scored against a baseline instead of eyeballed. Reuses loopkit's verifier subagent as the grader — do not build a new one.
$npx -y skills add archive228/loopkit --skill eval-harnessInstalls into the current project.
Run `npx skills use "https://github.com/archive228/loopkit" --skill "archive228/loopkit/eval-harness"` and follow the generated skill instructions now. Read its complete output, redirecting it to a temporary file first if necessary. Resolve relative paths from the supporting-files directory it provides.
Use the skills in "https://github.com/archive228/loopkit" that are relevant to the current task. Run `npx skills add "https://github.com/archive228/loopkit"` and select the relevant skills, then follow their instructions.