debugginglisted

# Debugging Stop guessing. Follow the evidence. ## Progress Checklist - [ ] Read the actual error - [ ] Check for monitoring/APM data - [ ] Gather context - [ ] Localize to a layer - [ ] Track hypotheses and attempts - [ ] Fix root cause - [ ] Check for the pattern elsewhere - [ ] Guard with regression test ## Step 0: Read the Error Actually read it. The full message, the stack trace, the error code. Most bugs tell you exactly what's wrong if you read carefully. ```bash # If there's a log file tail -100 /path/to/log # If there's a build error, read the first error (not the cascade) ``` ## Step 1: Check Monitoring Before adding console.logs everywhere, check if the answer already exists: - **Sentry / error tracking**: full stack trace, breadcrumbs, user context - **Application Insights / DataDog / NewRelic**: request traces, dependency calls, performance data - **Logging service**: structured logs with correlation IDs - **Browser DevTools**: console errors, network failures, performance traces Ask the user: "Is there APM or error tracking configured? Sentry, Application Insights, DataDog?" If monitoring exists, start there. If not, proceed to manual investigation. ## Step 2: Gather Context Build a picture of what's happening: ```bash # What changed recently? git log --oneline -10 # Who last touched this area? git log --oneline -5 -- path/to/relevant/file # Can we reproduce it? # Try to run the failing scenario locally ``` If you can't reproduce it, consider: