Go指标评审
prometheus-go-code-review
by anderskev
Reviews Prometheus instrumentation in Go code for proper metric types, labels, and patterns. Use when reviewing code with prometheus/client_golang metrics.
安装
claude skill add --url github.com/openclaw/skills/tree/main/skills/anderskev/prometheus-go-code-review文档
Prometheus Go Code Review
Review Checklist
- Metric types match measurement semantics (Counter/Gauge/Histogram)
- Labels have low cardinality (no user IDs, timestamps, paths)
- Metric names follow conventions (snake_case, unit suffix)
- Histograms use appropriate bucket boundaries
- Metrics registered once, not per-request
- Collectors don't panic on race conditions
- /metrics endpoint exposed and accessible
Metric Type Selection
| Measurement | Type | Example |
|---|---|---|
| Requests processed | Counter | requests_total |
| Items in queue | Gauge | queue_length |
| Request duration | Histogram | request_duration_seconds |
| Concurrent connections | Gauge | active_connections |
| Errors since start | Counter | errors_total |
| Memory usage | Gauge | memory_bytes |
Critical Anti-Patterns
1. High Cardinality Labels
// BAD - unique per user/request
counter := promauto.NewCounterVec(
prometheus.CounterOpts{Name: "requests_total"},
[]string{"user_id", "path"}, // millions of series!
)
counter.WithLabelValues(userID, request.URL.Path).Inc()
// GOOD - bounded label values
counter := promauto.NewCounterVec(
prometheus.CounterOpts{Name: "requests_total"},
[]string{"method", "status_code"}, // <100 series
)
counter.WithLabelValues(r.Method, statusCode).Inc()
2. Wrong Metric Type
// BAD - using gauge for monotonic value
requestCount := promauto.NewGauge(prometheus.GaugeOpts{
Name: "http_requests",
})
requestCount.Inc() // should be Counter!
// GOOD
requestCount := promauto.NewCounter(prometheus.CounterOpts{
Name: "http_requests_total",
})
requestCount.Inc()
3. Registering Per-Request
// BAD - new metric per request
func handler(w http.ResponseWriter, r *http.Request) {
counter := prometheus.NewCounter(...) // creates new each time!
prometheus.MustRegister(counter) // panics on duplicate!
}
// GOOD - register once
var requestCounter = promauto.NewCounter(prometheus.CounterOpts{
Name: "http_requests_total",
})
func handler(w http.ResponseWriter, r *http.Request) {
requestCounter.Inc()
}
4. Missing Unit Suffix
// BAD
duration := promauto.NewHistogram(prometheus.HistogramOpts{
Name: "request_duration", // no unit!
})
// GOOD
duration := promauto.NewHistogram(prometheus.HistogramOpts{
Name: "request_duration_seconds", // unit in name
})
Good Patterns
Metric Definition
var (
httpRequests = promauto.NewCounterVec(
prometheus.CounterOpts{
Namespace: "myapp",
Subsystem: "http",
Name: "requests_total",
Help: "Total HTTP requests processed",
},
[]string{"method", "status"},
)
httpDuration = promauto.NewHistogramVec(
prometheus.HistogramOpts{
Namespace: "myapp",
Subsystem: "http",
Name: "request_duration_seconds",
Help: "HTTP request latencies",
Buckets: []float64{.005, .01, .025, .05, .1, .25, .5, 1, 2.5, 5, 10},
},
[]string{"method"},
)
)
Middleware Pattern
func metricsMiddleware(next http.Handler) http.Handler {
return http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
timer := prometheus.NewTimer(httpDuration.WithLabelValues(r.Method))
defer timer.ObserveDuration()
wrapped := &responseWriter{ResponseWriter: w, status: 200}
next.ServeHTTP(wrapped, r)
httpRequests.WithLabelValues(r.Method, strconv.Itoa(wrapped.status)).Inc()
})
}
Exposing Metrics
import "github.com/prometheus/client_golang/prometheus/promhttp"
func main() {
http.Handle("/metrics", promhttp.Handler())
http.ListenAndServe(":9090", nil)
}
Review Questions
- Are metric types correct (Counter vs Gauge vs Histogram)?
- Are label values bounded (no UUIDs, timestamps, paths)?
- Do metric names include units (_seconds, _bytes)?
- Are metrics registered once (not per-request)?
- Is /metrics endpoint properly exposed?
相关 Skills
前端设计
by anthropics
面向组件、页面、海报和 Web 应用开发,按鲜明视觉方向生成可直接落地的前端代码与高质感 UI,适合做 landing page、Dashboard 或美化现有界面,避开千篇一律的 AI 审美。
✎ 想把页面做得既能上线又有设计感,就用前端设计:组件到整站都能产出,难得的是能避开千篇一律的 AI 味。
网页构建器
by anthropics
面向复杂 claude.ai HTML artifact 开发,快速初始化 React + Tailwind CSS + shadcn/ui 项目并打包为单文件 HTML,适合需要状态管理、路由或多组件交互的页面。
✎ 在 claude.ai 里做复杂网页 Artifact 很省心,多组件、状态和路由都能顺手搭起来,React、Tailwind 与 shadcn/ui 组合效率高、成品也更精致。
网页应用测试
by anthropics
用 Playwright 为本地 Web 应用编写自动化测试,支持启动开发服务器、校验前端交互、排查 UI 异常、抓取截图与浏览器日志,适合调试动态页面和回归验证。
✎ 借助 Playwright 一站式验证本地 Web 应用前端功能,调 UI 时还能同步查看日志和截图,定位问题更快。
相关 MCP 服务
GitHub
编辑精选by GitHub
GitHub 是 MCP 官方参考服务器,让 Claude 直接读写你的代码仓库和 Issues。
✎ 这个参考服务器解决了开发者想让 AI 安全访问 GitHub 数据的问题,适合需要自动化代码审查或 Issue 管理的团队。但注意它只是参考实现,生产环境得自己加固安全。
Context7 文档查询
编辑精选by Context7
Context7 是实时拉取最新文档和代码示例的智能助手,让你告别过时资料。
✎ 它能解决开发者查找文档时信息滞后的问题,特别适合快速上手新库或跟进更新。不过,依赖外部源可能导致偶尔的数据延迟,建议结合官方文档使用。
by tldraw
tldraw 是让 AI 助手直接在无限画布上绘图和协作的 MCP 服务器。
✎ 这解决了 AI 只能输出文本、无法视觉化协作的痛点——想象让 Claude 帮你画流程图或白板讨论。最适合需要快速原型设计或头脑风暴的开发者。不过,目前它只是个基础连接器,你得自己搭建画布应用才能发挥全部潜力。