"Sentry — application monitoring platform for error tracking, performance monitoring, session replay, profiling, and alerting. Use when building with Sentry or asking about SDK initialization, error capture, tracing, source maps, sampling, PII scrubbing, tunneling, or integration with Next.js, Node.js, or other JavaScript frameworks. Fetch live documentation for up-to-date details."
Install
npx skillscat add mikkelkrogsholm/dev-skills/sentry Install via the SkillsCat registry.
SKILL.md
Sentry
Sentry is an application monitoring platform for real-time error tracking, performance monitoring, session replay, profiling, cron monitoring, and log aggregation.
Documentation
- Docs: https://docs.sentry.io
Key Capabilities
Sentry's SDK bundles capabilities developers often add as separate tools:
- Performance tracing: distributed tracing across services is built-in — configure
tracesSampleRateandtracePropagationTargets, no external APM needed - Session Replay: full session recording with
Sentry.replayIntegration()— no third-party replay tool needed - Profiling: function-level performance profiling with
Sentry.browserProfilingIntegration()— no separate profiler needed - User Feedback: collect user-submitted reports tied to errors with
Sentry.feedbackIntegration()— no separate feedback widget needed - Cron monitoring: track scheduled job health with check-ins — no external uptime monitor needed for crons
Best Practices
- Lower
tracesSampleRatein production — the getting-started default of1.0captures 100% of transactions. At any meaningful traffic volume this inflates costs fast. Use a low value (0.1 or less) and considertracesSamplerfor dynamic sampling by route. - Session Replay uses a two-rate pattern — set
replaysSessionSampleRatelow (e.g. 0.1) for normal sessions andreplaysOnErrorSampleRate: 1.0so every error session is always captured. Agents often set only one rate and miss error context or over-capture. - Source maps require CI auth token — the wizard writes
.env.sentry-build-plugin(auto-gitignored) for local builds, but CI/CD pipelines needSENTRY_AUTH_TOKENset as an environment variable separately. Missing this silently produces minified stack traces in production. - Next.js requires four files — The client runtime uses
instrumentation-client.ts. Server and edge configs (sentry.server.config.ts,sentry.edge.config.ts) must be imported frominstrumentation.tsvia conditional imports — withoutinstrumentation.ts, the server and edge configs do nothing. Missing any one file leaves a runtime blind spot without obvious errors. - Use
tunnelto bypass ad blockers — settunnel: "/tunnel"inSentry.init()and add a backend route that proxies to Sentry's ingest endpoint. Without this, a significant portion of browser errors are silently dropped by ad blockers and privacy extensions.