bug-triage

name	bug-triage
description	Diagnose and prioritize bugs systematically. Use when the user reports a bug, error, crash, broken feature, or something that doesn't work. Walks through reproduction, P0-P4 classification, 5 hypothesis types (recent change, environment, race condition, dependency, data), quick-fix vs root-fix decision, and regression test addition.

Bug Triage

When user reports a bug:

Step 1: Reproduce

1. Get exact reproduction steps
2. Environment (OS, Node version, browser)
3. Logs / stack trace if available
4. Screenshots if UI-related

Step 2: Classify

• P0 — Blocker: production down, data loss, security
• P1 — Critical: feature broken for >50% users
• P2 — Major: feature broken, has workaround
• P3 — Minor: cosmetic, edge case
• P4 — Trivial: typo, polish

Step 3: Diagnose

Hypotheses, in order:

1. Recent code change? (git log -p last 24h)
2. Environment difference? (works on my machine)
3. Race condition? (intermittent failures)
4. External dependency? (API down)
5. Data issue? (specific user/account)

Step 4: Quick fix vs root fix

• Quick fix: revert, feature flag, hotfix
• Root fix: identify and fix underlying cause

Step 5: Verify

• Test the fix
• Test regression scenarios
• Add a test that would have caught this bug

Output Format

Always structure as:

SEVERITY: Px
REPRO: [steps]
HYPOTHESIS: [most likely cause]
FIX: [minimal fix]
TEST: [how to verify]
PREVENTION: [test/monitor to add]