What problem does it solve? Hard bugs and flaky performance regressions resist casual code reading; this Skill enforces a disciplined six-phase diagnosis loop so you build a reproducible failing signal before forming hypotheses, instead of guessing at fixes. ## Core Features & Use Cases - Feedback-loop construction: Build a tight, deterministic, red-capable command (failing test, curl script, headless browser, replayed trace, bisection harness) that goes red on the exact bug before any hypothesis work. - Structured hypothesis testing: Generate 3-5 ranked falsifiable hypotheses, instrument with tagged debug logs, and change one variable at a time. - Regression locking: Convert the minimised repro into a failing regression test at a correct seam, apply the fix, and clean up all instrumentation. - Use Case: A user reports the export endpoint intermittently returns empty files. Use this Skill to build a loop that replays the failing request 100 times, minimise the repro, bisect hypotheses, and land a fix with a regression test. ## Quick Start Diagnose this bug: the checkout page throws a 500 error when the cart contains more than three items.
