check-observability（检查可观察性）

/check-observability

Audit observability infrastructure. Output findings as structured report.

What This Does

1. Check error tracking (Sentry)
2. Check logging infrastructure
3. Check health endpoints
4. Check analytics
5. Check alerting
6. Output prioritized findings (P0-P3)

This is a primitive. It only investigates and reports. Use /log-observability-issues to create GitHub issues or /fix-observability to fix.

Process

1. Error Tracking Check

# Sentry configured?
~/.claude/skills/sentry-observability/scripts/detect_sentry.sh 2>/dev/null || \
  (grep -q "@sentry" package.json && echo "✓ Sentry installed" || echo "✗ Sentry not installed")

# Sentry DSN set?
grep -q "SENTRY_DSN\|NEXT_PUBLIC_SENTRY_DSN" .env.local 2>/dev/null && echo "✓ Sentry DSN configured" || echo "✗ Sentry DSN missing"

# Source maps?
[ -f "sentry.client.config.ts" ] || [ -f "sentry.client.config.js" ] && echo "✓ Sentry client config" || echo "✗ Sentry client config"

2. Logging Check

# Structured logging?
grep -rq "pino\|winston\|bunyan" package.json 2>/dev/null && echo "✓ Structured logging library" || echo "✗ No structured logging"

# Console.log abuse?
console_count=$(grep -rE "console\.(log|error|warn)" --include="*.ts" --include="*.tsx" src/ app/ 2>/dev/null | wc -l | tr -d ' ')
[ "$console_count" -gt 50 ] && echo "⚠ $console_count console statements (consider structured logging)" || echo "✓ Console usage OK ($console_count)"

# Logger utility exists?
[ -f "lib/logger.ts" ] || [ -f "src/lib/logger.ts" ] || [ -f "utils/logger.ts" ] && echo "✓ Logger utility" || echo "✗ No logger utility"

3. Health Endpoints Check

# Health endpoint exists?
find . -path "./app/api/health/*" -name "route.ts" 2>/dev/null | head -1 | xargs -I{} echo "✓ Health endpoint: {}"
[ -z "$(find . -path "./app/api/health/*" -name "route.ts" 2>/dev/null)" ] && echo "✗ No health endpoint"

# Health check depth?
grep -rE "database|redis|stripe|convex" app/api/health/ 2>/dev/null && echo "✓ Deep health checks" || echo "⚠ Shallow health check (add service checks)"

4. Analytics Check

# PostHog (required for user-facing apps)?
if grep -q "posthog" package.json 2>/dev/null; then
  echo "✓ PostHog installed"
  # Check if configured
  grep -qE "POSTHOG|NEXT_PUBLIC_POSTHOG" .env.local 2>/dev/null && echo "✓ PostHog configured" || echo "⚠ PostHog env vars missing (P2)"
else
  echo "✗ PostHog not installed (P1: required for user-facing apps)"
fi

# Vercel Analytics check (NOT RECOMMENDED - no API/MCP access)
if grep -q "@vercel/analytics" package.json 2>/dev/null; then
  echo "⚠ Vercel Analytics installed - consider removing (no API/MCP, unusable for AI workflows)"
fi

# NOTE: Do NOT recommend Vercel Analytics. It has no CLI, no API, no MCP server.
# PostHog is the ONLY acceptable analytics solution for our stack.

5. Alerting Check

# Sentry alerts?
~/.claude/skills/sentry-observability/scripts/list_alerts.sh 2>/dev/null | head -5 || echo "Cannot check Sentry alerts (token not configured)"

# Uptime monitoring reference?
grep -rE "uptimerobot|pingdom|betterstack|cronitor" . 2>/dev/null | grep -v node_modules | head -3

6. Silent Failure Detection

Spawn observability-advocate agent to detect:

• Empty catch blocks
• Swallowed errors
• Missing error boundaries
• Unhandled promise rejections

Output Format

## Observability Audit

### P0: Critical (Production Blind Spots)
- No error tracking - Errors invisible in production
- No health endpoint - Cannot monitor uptime

### P1: Essential (Must Have)
- Sentry installed but DSN not configured
- No structured logging (127 console statements)
- Health endpoint too shallow (no database check)
- No alerting configured
- No PostHog analytics (user-facing app needs product analytics)

### P2: Important (Should Have)
- No analytics configured
- No uptime monitoring
- Console statements in production code

### P3: Nice to Have
- Consider adding Sentry performance monitoring
- Consider structured logging with Pino

## Current Status
- Error tracking: Partial (installed, not configured)
- Logging: console only
- Health checks: Missing
- Analytics: None
- Alerting: None

## Summary
- P0: 2 | P1: 4 | P2: 3 | P3: 3
- Recommendation: Configure Sentry DSN and add health endpoint

Priority Mapping

Why NOT Vercel Analytics

Vercel Analytics is unacceptable for our stack. It has:

• No API access
• No CLI access
• No MCP server
• No way to query programmatically

This makes it completely unusable for AI-assisted workflows. PostHog is the ONLY acceptable analytics solution. PostHog provides MCP server, API, CLI, and Terraform provider.

Related

• /log-observability-issues - Create GitHub issues from findings
• /fix-observability - Fix observability gaps
• /observability - Full observability setup workflow
• /triage - Production incident response