What problem does it solve? When a bug, regression, or performance issue has no evidence-backed root cause, teams often jump straight to fixes based on guesses. This Skill enforces a disciplined, read-only diagnostic workflow that reproduces the exact reported symptom, tests falsifiable hypotheses, and returns a structured Diagnosis Packet before any code change is proposed. ## Core Features & Use Cases - Red-Capable Feedback Loops: Builds a deterministic reproduction loop (existing test, HTTP/CLI exercise, browser scenario, or throwaway harness) that asserts the exact user-reported symptom. - Hypothesis Testing: Generates 3-5 ranked, falsifiable hypotheses and confirms root cause only when a probe distinguishes it from alternatives. - Structured Diagnosis Packet: Returns status, symptom, evidence, root cause, affected symbols, regression seam, and remaining uncertainty in a fixed schema. - Use Case: A user reports intermittent checkout failures. Use this Skill to reproduce the failure at a measured rate, isolate the load-bearing inputs, confirm the failing code path, and hand off a regression seam for the fix. ## Quick Start Ask the agent to diagnose why the reported bug happens and return a Diagnosis Packet before proposing any fix.
