---

name: sentry description: >- Sentry captures application errors, traces performance, records session replays and tracks release health. Use this skill when integrating Sentry SDKs, configuring alerting, analyzing stack traces, uploading source maps, or tracking release health in production. Trigger words: sentry, error monitoring, error tracking, performance monitoring, source maps, session replay. license: Apache-2.0 compatibility: "A Sentry account (sentry.io or self-hosted) and a project DSN. Next.js SDK needs Next.js 14+ and Node.js 20.19+." metadata: author: terminal-skills version: "1.1.0" repository: https://github.com/getsentry/sentry category: development tags: ["sentry", "error-monitoring", "performance", "observability", "debugging"]

Sentry

Overview

Sentry is an error monitoring and performance platform that captures unhandled exceptions, tracks request performance with Web Vitals, records session replays, and alerts on regressions. It supports JavaScript, Python, Go, and mobile platforms with auto-instrumentation, source-mapped stack traces, and release health tracking.

Instructions

• Install. For Next.js run npx @sentry/wizard@latest -i nextjs. It creates instrumentation-client.ts (browser), sentry.server.config.ts, sentry.edge.config.ts, instrumentation.ts, app/global-error.tsx, wraps next.config.ts in withSentryConfig, and adds /sentry-example-page to trigger a test error. The old sentry.client.config.ts file is replaced by instrumentation-client.ts. Other JavaScript frameworks have their own package (@sentry/react, @sentry/node, @sentry/sveltekit); Python is pip install sentry-sdk.
• Init. Call Sentry.init() with dsn, environment, release and tracesSampleRate. Read the DSN from an environment variable such as NEXT_PUBLIC_SENTRY_DSN. sendDefaultPii is off by default; turn it on only if you accept IP addresses and request headers being sent.
• Errors. Call Sentry.setUser({ id }) after login, Sentry.setTag() for filterable values, Sentry.captureException(err) for handled errors, and ignoreErrors for known noise (browser extensions, ResizeObserver loop warnings, third-party scripts).
• Source maps. Generate maps in the build and upload them from CI. With Vite use @sentry/vite-plugin (placed after the other plugins), build.sourcemap: "hidden", authToken: process.env.SENTRY_AUTH_TOKEN and sourcemaps.filesToDeleteAfterUpload so the .map files are not served publicly. The Next.js wizard sets this up through withSentryConfig({ org, project }); the token goes in SENTRY_AUTH_TOKEN, never in the repository.
• Performance. tracesSampleRate of 0.1 in production (1.0 only in development), or a tracesSampler function to sample by route. Custom spans use Sentry.startSpan({ name, op, attributes }, callback). Set tracePropagationTargets so traces continue into your own APIs. Web Vitals (LCP, CLS, INP) come from the browser tracing integration.
• Alerts. Create issue alert rules and metric alerts in Alerts, targeting error-rate spikes or new issues per environment, and send them to Slack or PagerDuty through the integration.
• Session Replay. Add integrations: [Sentry.replayIntegration()] with replaysSessionSampleRate: 0.1 and replaysOnErrorSampleRate: 1.0. Replay masks all text and blocks media by default; keep that for anything with personal data and unmask only specific elements.
• Releases. Set release to the git SHA, link commits with sentry-cli releases set-commits "$VERSION" --auto (needs a repository integration), and record deploys with sentry-cli releases --org ORGANIZATION_SLUG deploys VERSION new -e production.

Examples

Example 1: Set up Sentry for a Next.js production app

User request: "Add Sentry error monitoring and performance tracking to my Next.js app"

npx @sentry/wizard@latest -i nextjs

Then edit instrumentation-client.ts:

import * as Sentry from "@sentry/nextjs";

Sentry.init({
  dsn: process.env.NEXT_PUBLIC_SENTRY_DSN,
  environment: process.env.NEXT_PUBLIC_APP_ENV ?? "development",
  release: process.env.NEXT_PUBLIC_GIT_SHA,
  tracesSampleRate: process.env.NODE_ENV === "development" ? 1.0 : 0.1,
  replaysSessionSampleRate: 0.1,
  replaysOnErrorSampleRate: 1.0,
  integrations: [Sentry.replayIntegration()],
});

export const onRouterTransitionStart = Sentry.captureRouterTransitionStart;

Set SENTRY_AUTH_TOKEN as a CI secret so next build uploads source maps. Result: open /sentry-example-page, click the button, and the error appears in Issues with a readable, source-mapped stack trace within a minute.

Example 2: Python service with release tracking

User request: "Set up release tracking to identify which deployment introduced a bug"

import os
import sentry_sdk

sentry_sdk.init(
    dsn=os.environ["SENTRY_DSN"],
    environment=os.environ.get("APP_ENV", "production"),
    release=os.environ["GIT_SHA"],
    traces_sample_rate=0.1,
)

sentry-cli releases set-commits "$GIT_SHA" --auto
sentry-cli releases --org billing-platform deploys "$GIT_SHA" new -e production

Result: each issue shows the release it first appeared in, suspect commits, and a regression alert fires when a resolved issue returns in a newer release.

Guidelines

• Keep tracesSampleRate at 0.1 or lower in production; 100% sampling costs quota and money.
• Always set environment and release so staging noise stays out of production views.
• Upload source maps in CI and delete them from the deployed output; public .map files expose your source.
• Alert on rates and new issues, not on every event, to avoid alert fatigue.
• Version differences matter: v8 removed startTransaction and Hub APIs in favor of startSpan and scopes; do not copy v7-era snippets.
• Browser ad blockers drop Sentry requests; in Next.js set tunnelRoute: "/monitoring" in withSentryConfig if you need those events.
• Do not put DSNs or auth tokens in examples committed to public repositories; the auth token especially must stay secret (the DSN is public by design but can be abused for spam).