Debugging & Root-Cause Prompts

What's Inside

• ▸Systematic root-cause analysis from real errors & stack traces
• ▸Spotting red herrings — rate limits, swallowed errors, unhandled rejections
• ▸Layer-by-layer failure isolation (client → app → dependency → infra)
• ▸Schema/type-mismatch & version-drift debugging
• ▸Silent-failure tracing (no error, wrong result)
• ▸Reproducing flaky bugs & bisecting regressions
• ▸Hang/timeout, out-of-memory, and connectivity diagnosis
• ▸Failing-test triage (real bug vs environment) + proving the fix

Try One Before You Buy

Published verbatim from the pack: “Root-cause a runtime error systematically” — When something is throwing and you want the cause, not a guess.

You are a senior debugger. My {{service/app}} is throwing:
{{paste_error_and_stack}}

Work the evidence, don't guess:
1. Read the ACTUAL error and the line it points to — quote it back to me.
2. Check the previous logs, not just the current ones — the first failure is
   often earlier and quieter than the crash.
3. What changed most recently (code, config, dependency, data)?
4. Form ONE hypothesis, state how to confirm it, and confirm it before fixing.

Output: the confirmed root cause with evidence, then the minimal fix. Do not
patch the symptom.

Why it works: The loudest error is usually downstream of the real one; reading previous logs and "what changed" beats pattern-matching the stack trace.

Every prompt in the pack ships with a when-to-use line and a why-it-works note like this one.