investigate

Debugging a regression requires a four-phase workflow: investigate the symptom using bash and grep, analyze the gathered evidence, hypothesize the root cause, and implement the verified fix to ensure the regression does not repeat.

Yes, you can use bash and grep to investigate unexpected application errors by searching codebases, reading log files, and collecting evidence to systematically triage the anomaly and verify the root cause before applying any code changes.

The best way to triage a performance anomaly is to execute a structured investigation workflow that requires explicit reproduction steps, gathers evidence using file search and web search, and verifies the root cause before proposing a targeted fix.

Verifying a root cause before implementing a bug fix enforces the iron law of debugging, ensuring you address the actual underlying issue with evidence rather than applying a superficial patch that leaves the defect ready to repeat.

Root-cause debugging requires explicit reproduction steps to effectively identify and verify the source of software defects. Without these steps, the investigation cannot reliably gather evidence or confirm the hypothesis before implementing a fix.