investigate

Root-cause debugging uses a four-phase workflow—collect, analyze, hypothesize, and implement—to identify and fix the underlying cause of software failures rather than applying superficial patches.

Debugging flaky behavior requires an evidence-driven workflow that gathers reproducible steps and tests hypotheses before implementing changes, ensuring the root cause is verified rather than guessed.

Stack trace analysis involves collecting evidence with tools like Bash, Read, and Grep to reproduce the error, then forming and testing hypotheses about the root cause before implementing a fix.

Yes, the investigation workflow uses Bash, Read, Grep, and WebSearch to reproduce errors, observe behavior, and verify root causes through evidence gathering and hypothesis testing.

Root-cause debugging enforces guardrails with explicit checks and prompts that prevent premature fixes, ensuring changes are intentional, documented, and verified before implementation.

A structured debugging workflow requires reproducible steps and observable evidence, meaning failures that cannot be reproduced or leave no traces may be difficult to investigate through this evidence-driven approach.