What problem does it solve? Debugging shipped applications often devolves into reading code, forming a confident theory, and shipping an unverified fix. This Skill enforces a reproduce-before-you-theorize discipline: build a signal that goes red on the exact bug first, then prove the fix by watching it flip red-to-green and back. ## Core Features & Use Cases - Mode-switched debugging loops: Screenshot-first fast path for CSS/layout/interaction bugs, a full instrumented six-phase loop for state/data/async bugs, and a measure-first branch for performance regressions. - Structured hypothesis control: Generates 3-5 ranked, falsifiable hypotheses before testing any, countering anchoring and confirmation bias. - Flake-rate reproduction: Playwright loops that run intermittent bugs dozens of times to raise reproduction rates, plus a human-in-the-loop bash template for steps only a person can perform. - Use Case: A booking app intermittently confirms the wrong time slot. The Skill drives a Playwright loop 50 times to establish a 30% flake rate, minimizes the repro to a fast double-tap, proves a stale-closure root cause, and locks it with a regression test. ## Quick Start Ask the agent to diagnose why the confirmation screen intermittently shows the wrong time in your shipped Next.js app, starting by building a failing reproduction before proposing any fix.
