Automated error triage

Severity: Recommended

Not all errors are equal. Triage should prioritize by impact (affected users, revenue impact, frequency) and route to appropriate owners.

Check automatically:

1. Check for Sentry alerting/workflow rules:

# Sentry alerting config
grep -riE "sentry.*alert|sentry.*rule|sentry.*workflow|alert.*rule" --include="*.ts" --include="*.js" --include="*.yml" --include="*.json" . 2>/dev/null | grep -v node_modules | head -10

2. Check for error categorization/tagging:

# Error categorization logic
grep -riE "error.*category|error.*priority|error.*severity|triage|fingerprint" --include="*.ts" --include="*.js" src/ lib/ app/ 2>/dev/null | head -10

3. Check for issue tracker integration:

# Sentry to issue tracker
grep -riE "sentry.*linear|sentry.*jira|sentry.*github.*issue" . 2>/dev/null | grep -v node_modules | head -5

4. Check for ignore rules:

# Known/expected errors filtered
grep -riE "ignoreErrors|denyUrls|beforeSend.*return null" --include="*.ts" --include="*.js" src/ lib/ app/ config/ 2>/dev/null | head -10

If not found in code, ask user:

• "How are Sentry errors prioritized? (volume, affected users, revenue impact)"
• "Are errors automatically assigned to teams/owners?"
• "Do errors auto-create issues in your task tracker (Linear, Jira)?"
• "Is there logic to ignore known/acceptable errors?"

Cross-reference with:

• ERR-006 (AI review is part of triage)
• ERR-007 (triage determines what gets auto-fixed)
• Section 35 - Incident Response (error triage feeds into incident process)

Pass criteria:

• Errors categorized by severity/priority (automated or via Sentry rules)
• High-impact errors routed to appropriate team/owner
• Known/expected errors filtered (ignored or low priority)
• Integration with issue tracker for actionable errors

Fail criteria:

• All errors treated equally (no prioritization)
• Manual triage only (someone scans Sentry and decides)
• No ownership assignment

Partial (acceptable):

• Sentry's built-in alerting rules configured but no AI triage
• Manual triage but with documented process/cadence

Evidence to capture:

• Triage rules/logic
• Issue tracker integration
• Ignore rules for known errors
• Ownership/routing configuration

---