What problem does it solve? Stubborn bugs often get patched with guesses that fix symptoms instead of causes. This Skill replaces reactive pattern-matching with a disciplined process: it explores the code first, interviews you one question at a time to narrow the failure tree, then runs a structured root cause analysis before any fix is written. ## Core Features & Use Cases - Code-First Exploration: Reads every file in the stack trace, checks git history, and traces callers before asking you anything, so questions target only what the code cannot answer. - Guided Interrogation: Drives a one-question-at-a-time interview covering reproduction, timeline, scope, frequency, and hypothesis testing until a single best-supported theory remains. - Systematic Diagnosis: Runs four phases — root cause investigation, pattern classification, falsifiable hypothesis testing, and fix recommendation — then reports root cause, evidence, fix, and regression test plan. - Use Case: You have an intermittent production failure that only appears under load. Invoke grilldebug with the error description; it walks you through targeted questions, traces the code, and delivers a committed root cause hypothesis with a concrete fix and test recommendation. ## Quick Start Ask the AI to run grilldebug on your bug, for example: "grilldebug — the checkout API intermittently returns 500 errors under load, help me find the root cause."
