Skip to main content
skill CX Codex

sentry-go-sdk · getsentry — Claude Code skill: Sentry setup for Go

The skill scans your Go project and walks you through Sentry setup: errors, tracing, logs, metrics and crons.

Install into Claude Code · Skills CLI (Vercel) All methods and platforms ↓
npx skills add getsentry/sentry-agent-skills --skill sentry-go-sdk -a claude-code
Open on GitHub

What sentry-go-sdk does

It starts by reading the project. The skill greps go.mod for an existing sentry-go dependency, detects the web framework from imports (gin, echo, fiber, fasthttp, iris, negroni), checks which logging library is in use (logrus, zap, zerolog, slog), looks for scheduler patterns and any OpenTelemetry usage. Then it proposes a concrete feature set instead of asking open-ended questions.

Installation goes through go get: the core github.com/getsentry/sentry-go module plus a sub-package matching the detected framework or logger. In main() you add sentry.Init with the DSN from an environment variable, tracing and logs enabled, then register the framework middleware. For Fiber and FastHTTP the middleware table sets Repanic: false and WaitForDelivery: true, because valyala/fasthttp has no built-in recovery.

Profiling is off the table: it was removed in sentry-go v0.31.0, and the skill points to a reference file for alternatives rather than recommending it. The SDK versions and APIs in the skill reflect sentry-go v0.43.0, and the text tells you to verify against Sentry docs before implementing. Worth checking if your project has moved on since.

What it helps with

  • catch a panic in an HTTP handler and ship it to Sentry
  • wire tracing into gRPC and database calls
  • route logrus or zap output into Sentry
  • spot silent failures in scheduled jobs
  • pick the right sentry-go sub-package for your framework

Example prompts for the agent

  • add Sentry to my Go project
  • set up sentry-go for Gin with tracing and logs
  • check why panics from my Fiber app never reach Sentry

How to install sentry-go-sdk

1Skills CLI (Vercel)

npx skills add getsentry/sentry-agent-skills --skill sentry-go-sdk -a claude-code

2Manually: put the files in place

~/.claude/skills/sentry-go-sdk/SKILL.md

Copy the whole skill folder; for one project use .claude/skills/sentry-go-sdk/

Step by step, with commands for each platform: How to install skills in Claude Code: five ways

Questions about sentry-go-sdk

How do I install sentry-go-sdk in Claude?

Run npx skills add getsentry/sentry-agent-skills --skill sentry-go-sdk -a claude-code. The skill file then lives at ~/.claude/skills/sentry-go-sdk/SKILL.md.

What does sentry-go-sdk do?

It scans your Go project, proposes a set of Sentry features and guides you through SDK install, initialization and middleware registration. It supports net/http, Gin, Echo, Fiber, FastHTTP, Iris and Negroni.

Which Sentry features does the skill recommend by default?

Error monitoring, always. Tracing when HTTP handlers, gRPC or DB calls are detected, logging when logrus, zap, zerolog or slog is present. Metrics and crons are left as your call.

Source text

SKILL.md · Apache-2.0
---
name: sentry-go-sdk
description: Full Sentry SDK setup for Go. Use when asked to "add Sentry to Go", "install sentry-go", "setup Sentry in Go", or configure error monitoring, tracing, logging, metrics, or crons for Go applications. Supports net/http, Gin, Echo, Fiber, FastHTTP, Iris, and Negroni.
license: Apache-2.0
---

# Sentry Go SDK

Opinionated wizard that scans your Go project and guides you through complete Sentry setup.

## Invoke This Skill When

- User asks to "add Sentry to Go" or "setup Sentry" in a Go app
- User wants error monitoring, tracing, logging, metrics, or crons in Go
- User mentions `sentry-go`, `github.com/getsentry/sentry-go`, or Go Sentry SDK
- User wants to monitor panics, HTTP handlers, or scheduled jobs in Go

> **Note:** SDK versions and APIs below reflect Sentry docs at time of writing (sentry-go v0.43.0).
> Always verify against [docs.sentry.io/platforms/go/](https://docs.sentry.io/platforms/go/) before implementing.

---

## Phase 1: Detect

Run these commands to understand the project before making any recommendations:

```bash
# Check existing Sentry dependency
grep -i sentry go.mod 2>/dev/null

# Detect web framework
grep -E "gin-gonic/gin|labstack/echo|gofiber/fiber|valyala/fasthttp|kataras/iris|urfave/negroni" go.mod 2>/dev/null

# Detect logging libraries
grep -E "sirupsen/logrus|go.uber.org/zap|rs/zerolog|log/slog" go.mod go.sum 2>/dev/null

# Detect cron / scheduler patterns
grep -E "robfig/cron|go-co-op/gocron|jasonlvhit/gocron" go.mod 2>/dev/null

# Detect OpenTelemetry usage
grep "go.opentelemetry.io" go.mod 2>/dev/null

# Check for companion frontend
ls frontend/ web/ client/ ui/ 2>/dev/null
```

**What to note:**
- Is `sentry-go` already in `go.mod`? If yes, skip to Phase 2 (configure features).
- Which framework is used? (Determines which sub-package and middleware to install.)
- Which logging library? (Enables automatic log capture.)
- Are cron/scheduler patterns present? (Triggers Crons recommendation.)
- Is there a companion frontend directory? (Triggers Phase 4 cross-link.)

---

## Phase 2: Recommend

Based on what you found, present a concrete recommendation. Don't ask open-ended questions — lead with a proposal:

**Recommended (core coverage):**
- ✅ **Error Monitoring** — always; captures panics and unhandled errors
- ✅ **Tracing** — if HTTP handlers, gRPC, or DB calls are detected
- ✅ **Logging** — if logrus, zap, zerolog, or slog is detected

**Optional (enhanced observability):**
- ⚡ **Metrics** — custom counters and gauges for business KPIs / SLOs
- ⚡ **Crons** — detect silent failures in scheduled jobs
- ⚠️ **Profiling** — removed in sentry-go v0.31.0; see `references/profiling.md` for alternatives

**Recommendation logic:**

| Feature | Recommend when... |
|---------|------------------|
| Error Monitoring | **Always** — non-negotiable baseline |
| Tracing | `net/http`, gin, echo, fiber, or gRPC imports detected |
| Logging | logrus, zap, zerolog, or `log/slog` imports detected |
| Metrics | Business events, SLO tracking, or counters needed |
| Crons | `robfig/cron`, `gocron`, or scheduled job patterns detected |
| Profiling | ⚠️ **Removed in v0.31.0** — do not recommend; see `references/profiling.md` |

Propose: *"I recommend setting up Error Monitoring + Tracing [+ Logging if applicable]. Want me to also add Metrics or Crons?"*

---

## Phase 3: Guide

### Install

```bash
# Core SDK (always required)
go get github.com/getsentry/sentry-go

# Framework sub-package — install only what matches detected framework:
go get github.com/getsentry/sentry-go/http      # net/http
go get github.com/getsentry/sentry-go/gin       # Gin
go get github.com/getsentry/sentry-go/echo      # Echo
go get github.com/getsentry/sentry-go/fiber     # Fiber
go get github.com/getsentry/sentry-go/fasthttp  # FastHTTP

# Logging sub-packages — install only what matches detected logging lib:
go get github.com/getsentry/sentry-go/logrus    # Logrus
go get github.com/getsentry/sentry-go/slog      # slog (stdlib, Go 1.21+)
go get github.com/getsentry/sentry-go/zap       # Zap
go get github.com/getsentry/sentry-go/zerolog   # Zerolog

# OpenTelemetry bridge (only if OTel is already in use):
go get github.com/getsentry/sentry-go/otel
```

### Quick Start — Recommended Init

Add to `main()` before any other code. This config enables the most features with sensible defaults:

```go
import (
    "log"
    "os"
    "time"
    "github.com/getsentry/sentry-go"
)

err := sentry.Init(sentry.ClientOptions{
    Dsn:              os.Getenv("SENTRY_DSN"),
    Environment:      os.Getenv("SENTRY_ENVIRONMENT"), // "production", "staging", etc.
    Release:          release,                          // inject via -ldflags at build time
    SendDefaultPII:   true,
    AttachStacktrace: true,

    // Tracing (adjust sample rate for production)
    EnableTracing:    true,
    TracesSampleRate: 1.0, // lower to 0.1–0.2 in high-traffic production

    // Logs
    EnableLogs: true,
})
if err != nil {
    log.Fatalf("sentry.Init: %s", err)
}
defer sentry.Flush(2 * time.Second)
```

**Injecting `Release` at build time (recommended):**
```go
var release string // set by -ldflags

// go build -ldflags="-X main.release=my-app@$(git describe --tags)"
```

### Framework Middleware

After `sentry.Init`, register the Sentry middleware for your framework:

| Framework | Import path | Middleware call | `Repanic` | `WaitForDelivery` |
|-----------|------------|----------------|-----------|-------------------|
| `net/http` | `.../sentry-go/http` | `sentryhttp.New(opts).Handle(h)` | `true` | `false` |
| Gin | `.../sentry-go/gin` | `router.Use(sentrygin.New(opts))` | `true` | `false` |
| Echo | `.../sentry-go/echo` | `e.Use(sentryecho.New(opts))` | `true` | `false` |
| Fiber | `.../sentry-go/fiber` | `app.Use(sentryfiber.New(opts))` | `false` | `true` |
| FastHTTP | `.../sentry-go/fasthttp` | `sentryfasthttp.New(opts).Handle(h)` | `false` | `true` |
| Iris | `.../sentry-go/iris` | `app.Use(sentryiris.New(opts))` | `true` | `false` |
| Negroni | `.../sentry-go/negroni` | `n.Use(sentrynegroni.New(opts))` | `true` | `false` |

> **Note:** Fiber and FastHTTP are built on `valyala/fasthttp` which has no built-in recovery. Use `Repanic: false, WaitForDelivery: true` for those.

**Hub access in handlers:**
```go
// net/http, Negroni:
hub := sentry.GetHubFromContext(r.Context())

// Gin:
hub := sentrygin.GetHubFromContext(c)

// Echo:
hub := sentryecho.GetHubFromContext(c)

// Fiber:
hub := sentryfiber.GetHubFromContext(c)
```

### For Each Agreed Feature

Walk through features one at a time. Load the reference file for each, follow its steps, and verify before moving to the next:

| Feature | Reference file | Load when... |
|---------|---------------|-------------|
| Error Monitoring | `${SKILL_ROOT}/references/error-monitoring.md` | Always (baseline) |
| Tracing | `${SKILL_ROOT}/references/tracing.md` | HTTP handlers / distributed tracing |
| Profiling | `${SKILL_ROOT}/references/profiling.md` | Performance-sensitive production apps |
| Logging | `${SKILL_ROOT}/references/logging.md` | logrus / zap / zerolog / slog detected |
| Metrics | `${SKILL_ROOT}/references/metrics.md` | Business KPIs / SLO tracking |
| Crons | `${SKILL_ROOT}/references/crons.md` | Scheduler / cron job patterns detected |

For each feature: `Read ${SKILL_ROOT}/references/<feature>.md`, follow steps exactly, verify it works.

---

## Configuration Reference

### Key `ClientOptions` Fields

| Option | Type | Default | Purpose |
|--------|------|---------|---------|
| `Dsn` | `string` | `""` | SDK disabled if empty; env: `SENTRY_DSN` |
| `Environment` | `string` | `""` | e.g., `"production"`; env: `SENTRY_ENVIRONMENT` |
| `Release` | `string` | `""` | e.g., `"[email protected]"`; env: `SENTRY_RELEASE` |
| `SendDefaultPII` | `bool` | `false` | Include IP, request headers |
| `AttachStacktrace` | `bool` | `false` | Stack traces on `CaptureMessage` calls |
| `SampleRate` | `float64` | `1.0` | Error event sample rate (0.0 treated as 1.0) |
| `EnableTracing` | `bool` | `false` | Enable performance tracing |
| `TracesSampleRate` | `float64` | `0.0` | Transaction sample rate |
| `TracesSampler` | `TracesSampler` | `nil` | Custom per-transaction sampling (overrides rate) |
| `EnableLogs` | `bool` | `false` | Enable Sentry Logs feature |
| `MaxBreadcrumbs` | `int` | `100` | Max breadcrumbs per event |
| `MaxErrorDepth` | `int` | `100` | Max depth for unwrapping error chains |
| `Debug` | `bool` | `false` | Verbose SDK debug output |
| `BeforeSend` | `func` | `nil` | Hook to mutate/drop error events |
| `BeforeSendTransaction` | `func` | `nil` | Hook to mutate/drop transaction events |
| `IgnoreErrors` | `[]string` | `nil` | Regex patterns for errors to drop |
| `IgnoreTransactions` | `[]string` | `nil` | Regex patterns for transactions to drop |

### Environment Variables

| Variable | Maps to | Purpose |
|----------|---------|---------|
| `SENTRY_DSN` | `Dsn` | Data Source Name |
| `SENTRY_RELEASE` | `Release` | App version (e.g., `[email protected]`) |
| `SENTRY_ENVIRONMENT` | `Environment` | Deployment environment |
| `HTTP_PROXY` / `HTTPS_PROXY` | `HTTPProxy` / `HTTPSProxy` | Outbound proxy |

Options set in `ClientOptions` **override** environment variables.

---

## Verification

Test that Sentry is receiving events:

```go
// In a handler or test — triggers a real error event:
sentry.CaptureMessage("Sentry Go SDK test")

// Or trigger a captured panic (with recovery middleware in place):
panic("sentry test panic")
```

Check the Sentry dashboard within a few seconds. If nothing appears:
1. Set `Debug: true` in `ClientOptions` — prints SDK internals to stdout
2. Verify the DSN is correct and the project exists
3. Ensure `sentry.Flush(2 * time.Second)` is called (events are async by default)

---

## Phase 4: Cross-Link

After completing Go setup, check for a companion frontend missing Sentry coverage:

```bash
ls frontend/ web/ client/ ui/ 2>/dev/null
cat frontend/package.json web/package.json 2>/dev/null | grep -E '"react"|"svelte"|"vue"|"next"'
```

If a frontend directory exists without Sentry configured, suggest the matching skill:

| Frontend detected | Suggest skill |
|-------------------|--------------|
| React / Next.js | `sentry-react-sdk` |
| Svelte / SvelteKit | `sentry-svelte-sdk` |
| Vue | Use `@sentry/vue` — see [docs.sentry.io/platforms/javascript/guides/vue/](https://docs.sentry.io/platforms/javascript/guides/vue/) |
| Other JS/TS | `sentry-react-sdk` (covers generic browser JS patterns) |

---

## Troubleshooting

| Issue | Solution |
|-------|----------|
| Events not appearing | Set `Debug: true`, check DSN, verify `sentry.Flush()` is called |
| `sentry.Init` returns error | Malformed DSN — check format: `https://<key>@o<org>.ingest.sentry.io/<project>` |
| Panics not captured | Ensure framework middleware is registered before handlers |
| `defer sentry.Flush` not running | `os.Exit()` skips `defer` — call `sentry.Flush()` explicitly before `os.Exit()` |
| Missing stack traces | Set `AttachStacktrace: true` for `CaptureMessage`; works automatically for `CaptureException` |
| Goroutine events missing context | Clone hub before spawning goroutine: `hub := sentry.CurrentHub().Clone()` |
| Too many transactions | Lower `TracesSampleRate` or use `TracesSampler` to drop health checks / metrics endpoints |
| Fiber/FastHTTP not recovering | Use `Repanic: false, WaitForDelivery: true` for fasthttp-based frameworks |
| `SampleRate: 0.0` sending all events | `0.0` is treated as `1.0`; to drop all, set `Dsn: ""` instead |

© getsentry · Apache-2.0 · getsentry/sentry-agent-skills

Similar skills

Everything for Backend and APIs →

sentry-go-sdk by other authors comes first.

From the same set4

Project sponsors

Companies that keep this analytics open
Gold sponsor

Impressions, clicks and code copies — open statistics. We keep the numbers public for the same reason NeBlask itself is public. Become a sponsor