What problem does it solve? Developers often jump straight to fixing code that merely "looks suspicious," shipping speculative changes that may not address the real cause and can introduce regressions. This Skill enforces a disciplined hypothesis → verify → fix workflow so every bug fix is grounded in observed evidence. ## Core Features & Use Cases - Reproduction-First Workflow: Requires writing a test, script, or REPL invocation that demonstrates the failure before any code change, turning the repro into a regression test. - Diagnostics When Reproduction Fails: Guides adding targeted logging and assertions tied to specific hypotheses when bugs depend on production data, timing, or environment conditions. - Anti-Pattern Guardrails: Explicitly rejects speculative fixes, unverified root-cause claims, and hedged changes that address multiple possible causes at once. - Use Case: An agent assigned to a GitHub issue reporting intermittent crashes uses this Skill to first reproduce the crash locally, confirm the failing code path, and only then implement and verify the fix. ## Quick Start Investigate this reported bug by first reproducing the failure with a test before making any code changes.
