Evaluation loops make research claims inspectable by connecting each change to evidence gathered from tests, builds, traces, screenshots, or review notes.
Skill
Evaluation Loops
Systematic checks that guide iteration across code, UX, and agent behavior.