From magic-powers
Proactive reliability health check using auto-captured error and network failure data. Uses mcp__Amplitude__query_amplitude_data, mcp__Amplitude__get_charts, mcp__Amplitude__get_context.
npx claudepluginhub kienbui1995/magic-powers --plugin magic-powersThis skill uses the workspace's default tool permissions.
- Daily or weekly scheduled reliability review (before users notice issues)
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
Before classifying anything as "abnormal," establish what normal looks like for this product.
Baseline sources:
mcp__Amplitude__get_charts to find pre-built error rate charts if they existBaseline template:
Product: <name>
Check period: <date/time range>
Baseline period: <previous equivalent period>
Normal error rates (established from 30d history):
Network error rate: ~0.3% of sessions
JS error rate: ~0.1% of sessions
Error click rate: ~2.1% of sessions
If no prior baseline exists: treat today as Day 1, document these numbers as the starting baseline.
API failures are the most impactful errors — they block user actions entirely.
Query:
mcp__Amplitude__query_amplitude_data:
event: "[Amplitude] Network Error"
time_range: last 24h
group_by: endpoint, http_status_code
metric: unique users affected
What to look for:
/api/auth, /api/data/save, /api/payments failing → CriticalThreshold guidance:
GREEN: <0.5% of sessions have network errors, stable vs. baseline
YELLOW: 0.5–2% of sessions affected, or >20% increase vs. baseline
RED: >2% of sessions affected, or critical endpoint (auth/payment) failing
Unhandled JavaScript exceptions break page functionality even when APIs are working.
Query:
mcp__Amplitude__query_amplitude_data:
event: "[Amplitude] JS Error"
time_range: last 24h
group_by: error_type, file
metric: unique users affected
What to look for:
first_seen after the last deployment timeThreshold guidance:
GREEN: <0.1% of sessions have JS errors
YELLOW: 0.1–0.5% of sessions affected, or new error types since last deploy
RED: >0.5% of sessions affected, or error in payment/auth critical path
Error clicks — rage clicks and clicks on broken elements — reveal UX failures that don't generate HTTP errors.
Query:
mcp__Amplitude__query_amplitude_data:
event: "[Amplitude] Error Click"
time_range: last 24h
group_by: element, page_url
metric: unique users affected
What to look for:
Threshold guidance:
GREEN: <2% of sessions have error clicks
YELLOW: 2–5% of sessions, or concentrated on one key page
RED: >5% of sessions, or concentrated on checkout/signup/critical flows
For each metric, calculate week-over-week and day-over-day change:
mcp__Amplitude__query_amplitude_data:
same queries as above
time_range: previous 24h / previous 7d equivalent
Change threshold rules:
Correlation check:
Aggregate findings into a concise risk summary. Don't list everything — prioritize the top 3 issues that need action.
For each risk:
mcp__Amplitude__query_amplitude_data — query error event volumes by type (network, JS, error click), compare across time periods, group by endpoint/elementmcp__Amplitude__get_charts — retrieve pre-built error rate charts and reliability dashboards; check if baselines have been previously establishedmcp__Amplitude__get_context — get project context and understand what error events are being captured; verify auto-capture is configured correctly## Reliability Health Check — <date> <time>
### Scorecard
| Category | Rate | Status | vs. Yesterday | vs. Last Week |
|----------|------|--------|--------------|--------------|
| Network Errors | 0.4% of sessions | 🟡 YELLOW | +28% | +15% |
| JS Errors | 0.08% of sessions | 🟢 GREEN | -5% | +2% |
| Error Clicks | 1.8% of sessions | 🟢 GREEN | -3% | -8% |
**Overall Status: YELLOW** (1 category elevated)
### Top 3 Reliability Risks
**Risk 1: POST /api/payments → 500 rate increased 28%**
Scope: 423 users in last 24h
Trend: Increasing since 14:15 UTC (deploy window)
Action: Review payment service logs for timeout errors; consider rollback if rate continues rising
**Risk 2: <next issue>**
...
### What's Working Well
- JS error rate stable and within threshold
- No new error types detected
- Error click rate improved vs. last week
### Next Scheduled Check
<recommended check time based on cadence: daily/weekly>