sentry-tanstack-start-sdk
getsentry/sentry-for-ai
Full Sentry SDK setup for TanStack Start React with error monitoring, tracing, session replay, and logs.
What is sentry-tanstack-start-sdk?
Complete Sentry integration wizard for TanStack Start React projects. Configures browser and server-side error monitoring, distributed tracing, session replay, and optional logs and user feedback. Use when adding Sentry to a TanStack Start app or enabling observability features.
- Detects existing TanStack Start project structure and Sentry configuration
- Installs @sentry/tanstackstart-react SDK with opinionated defaults
- Configures client-side Sentry initialization in src/router.tsx with routing integration
- Sets up server-side instrumentation via instrument.server.mjs and src/server.ts
- Integrates Sentry Vite plugin for automatic source map uploads
- Adds global request and function middleware to capture server errors
How to install sentry-tanstack-start-sdk
npx skills add https://github.com/getsentry/sentry-for-ai --skill sentry-tanstack-start-sdk- TanStack Start 1.0 RC or compatible version installed
- Node.js project with package.json and Vite configuration
- Sentry account and project DSN
- SENTRY_AUTH_TOKEN for source map uploads (optional but recommended)
How to use sentry-tanstack-start-sdk
- 1.Run detection commands to scan your TanStack Start project structure and identify existing Sentry setup
- 2.Review the recommendation for which features to enable (error monitoring, tracing, session replay, logs, user feedback)
- 3.Install the SDK: npm install @sentry/tanstackstart-react --save
- 4.Initialize Sentry in src/router.tsx with browser-side configuration and integrations
- 5.Create instrument.server.mjs in project root with server-side Sentry initialization
- 6.Update vite.config.ts to add sentryTanstackStart plugin as the last plugin
- 7.Wrap the fetch handler in src/server.ts with wrapFetchWithSentry
- 8.Add sentryGlobalRequestMiddleware and sentryGlobalFunctionMiddleware to src/start.ts
Use cases
- Adding error monitoring and tracing to a new TanStack Start React application
- Enabling session replay to debug user-reported UX issues in production
- Setting up distributed tracing across client-side routes and server-side requests
- Configuring source map uploads for production error stack traces
- Adding in-app user feedback collection for direct issue reports
- React developers building TanStack Start applications
- DevOps and platform engineers setting up observability
- Product teams needing session replay and user feedback
- Teams requiring cross-stack (browser + server) error tracking
sentry-tanstack-start-sdk FAQ
Client configuration in src/router.tsx handles browser errors, routing tracing, and session replay. Server configuration via instrument.server.mjs and src/server.ts captures server-side errors, request tracing, and function execution errors.
The SDK is currently alpha and documented as compatible with TanStack Start 1.0 RC. Always verify against docs.sentry.io/platforms/javascript/guides/tanstackstart-react/ before implementing in production.
Error monitoring is always recommended. Tracing and session replay are high-value for most apps. Logs and user feedback are optional and depend on your observability needs and team workflow.
Configure SENTRY_AUTH_TOKEN in your environment and pass it to the sentryTanstackStart Vite plugin. The plugin will automatically upload source maps during build.
Yes. This skill configures the TanStack Start frontend. If you have a separate backend, set up Sentry SDKs for that runtime separately (Node.js, Python, Go, etc.) and link them via distributed tracing.
Full instructions (SKILL.md)
Source of truth, from getsentry/sentry-for-ai.
name: sentry-tanstack-start-sdk description: Full Sentry SDK setup for TanStack Start React. Use when asked to "add Sentry to TanStack Start", "install @sentry/tanstackstart-react", or configure error monitoring, tracing, session replay, logs, or user feedback in a TanStack Start React app. license: Apache-2.0 category: sdk-setup parent: sentry-sdk-setup disable-model-invocation: true
All Skills > SDK Setup > TanStack Start React SDK
Sentry TanStack Start React SDK
Opinionated wizard that scans your TanStack Start React project and guides you through complete Sentry setup for browser and server runtimes.
Invoke This Skill When
- User asks to "add Sentry to TanStack Start" or "set up Sentry" in a TanStack Start React app
- User wants to install or configure
@sentry/tanstackstart-react - User wants error monitoring, tracing, session replay, logs, or user feedback for TanStack Start React
- User asks about
sentryTanstackStart,wrapFetchWithSentry,instrument.server.mjs, or TanStack Start middleware instrumentation
Note: This SDK is currently alpha and documented as compatible with TanStack Start
1.0 RC. Always verify against docs.sentry.io/platforms/javascript/guides/tanstackstart-react/ before implementing.
Phase 1: Detect
Run these commands to understand the project before making any recommendations:
# Detect TanStack Start / Router and existing Sentry
cat package.json | grep -E '"@tanstack/react-start"|"@tanstack/react-router"|"@sentry/tanstackstart-react"'
# Check if Sentry is already present
cat package.json | grep '"@sentry/'
# Detect key files used by the TanStack Start setup
ls src/router.tsx src/start.ts src/server.ts instrument.server.mjs vite.config.ts vite.config.js 2>/dev/null
# Check whether source map upload credentials are configured
cat .env .env.local .env.sentry-build-plugin 2>/dev/null | grep "SENTRY_AUTH_TOKEN"
# Detect deployment hints in scripts
cat package.json | grep -E '"dev"|"build"|"start"|NODE_OPTIONS|--import'
# Detect logging libraries
cat package.json | grep -E '"pino"|"winston"|"loglevel"'
# Detect companion backend directories
ls ../backend ../server ../api 2>/dev/null
cat ../go.mod ../requirements.txt ../Gemfile ../pom.xml 2>/dev/null | head -3
What to determine:
| Question | Impact |
|---|---|
@tanstack/react-start present? | Confirms this skill is the right setup path |
@sentry/tanstackstart-react already installed? | Skip install and go to feature tuning |
src/router.tsx exists? | Client-side Sentry.init placement |
src/start.ts exists? | Global middleware setup for server-side errors |
src/server.ts exists? | Server entry instrumentation placement |
instrument.server.mjs exists? | Runtime startup instrumentation path |
vite.config.ts exists? | Add sentryTanstackStart plugin and source maps |
SENTRY_AUTH_TOKEN configured? | Source map upload readiness |
| Backend directory found? | Trigger Phase 4 cross-link suggestion |
Phase 2: Recommend
Present a concrete recommendation based on what you found. Do not ask open-ended questions — lead with a proposal:
Recommended (core coverage):
- ✅ Error Monitoring — always; captures unhandled client and server errors
- ✅ Tracing — high-value for request and route timing across browser and server
- ✅ Session Replay — recommended for user-facing apps
Optional (enhanced observability):
- ⚡ Logs — recommend when structured log search and log-to-trace correlation are needed
- ⚡ User Feedback — recommend when product teams want in-app issue reports
Recommendation logic:
| Feature | Recommend when... |
|---|---|
| Error Monitoring | Always — non-negotiable baseline |
| Tracing | Usually yes for TanStack Start; route + fetch instrumentation gives immediate value |
| Session Replay | User-facing app, login flows, checkout flows, or hard-to-reproduce UX bugs |
| Logs | Existing logging strategy, support workflow, or trace/log correlation needs |
| User Feedback | Team wants direct user reports without leaving the app |
Propose: "I recommend Error Monitoring + Tracing + Session Replay. Want me to also enable Logs and User Feedback?"
Phase 3: Guide
Install
npm install @sentry/tanstackstart-react --save
Configure Client-Side Sentry in src/router.tsx
Initialize Sentry inside the router factory and gate it to the browser:
import * as Sentry from "@sentry/tanstackstart-react";
import { createRouter } from "@tanstack/react-router";
export const getRouter = () => {
const router = createRouter();
if (!router.isServer) {
Sentry.init({
dsn: "___PUBLIC_DSN___",
dataCollection: {
// userInfo: false,
// httpBodies: [],
},
integrations: [
Sentry.tanstackRouterBrowserTracingIntegration(router),
Sentry.replayIntegration(),
Sentry.feedbackIntegration({
colorScheme: "system",
}),
],
enableLogs: true,
tracesSampleRate: 1.0,
replaysSessionSampleRate: 0.1,
replaysOnErrorSampleRate: 1.0,
});
}
return router;
};
Configure Server-Side Sentry in instrument.server.mjs
Create instrument.server.mjs in project root:
import * as Sentry from "@sentry/tanstackstart-react";
Sentry.init({
dsn: "___PUBLIC_DSN___",
dataCollection: {
// To disable sending user data and HTTP bodies, uncomment the lines below. For more info visit:
// https://docs.sentry.io/platforms/javascript/guides/tanstackstart-react/configuration/options/#dataCollection
// userInfo: false,
// httpBodies: [],
},
enableLogs: true,
tracesSampleRate: 1.0,
});
Configure Vite Plugin in vite.config.ts
sentryTanstackStart should be the last plugin:
import { defineConfig } from "vite";
import { sentryTanstackStart } from "@sentry/tanstackstart-react/vite";
import { tanstackStart } from "@tanstack/react-start/plugin/vite";
export default defineConfig({
plugins: [
tanstackStart(),
sentryTanstackStart({
org: "___ORG_SLUG___",
project: "___PROJECT_SLUG___",
authToken: process.env.SENTRY_AUTH_TOKEN,
}),
],
});
If the token is stored in .env, load it with loadEnv in the Vite config before passing it to the plugin.
Instrument Server Entry Point in src/server.ts
Wrap the fetch handler with wrapFetchWithSentry:
import { wrapFetchWithSentry } from "@sentry/tanstackstart-react";
import handler, { createServerEntry } from "@tanstack/react-start/server-entry";
export default createServerEntry(
wrapFetchWithSentry({
fetch(request: Request) {
return handler.fetch(request);
},
}),
);
Add Global Server Middleware in src/start.ts
These middleware capture server-side request and function errors:
import {
sentryGlobalFunctionMiddleware,
sentryGlobalRequestMiddleware,
} from "@sentry/tanstackstart-react";
import { createStart } from "@tanstack/react-start";
export const startInstance = createStart(() => {
return {
requestMiddleware: [sentryGlobalRequestMiddleware],
functionMiddleware: [sentryGlobalFunctionMiddleware],
};
});
Sentry middleware should be first in each array.
Runtime Startup Patterns
Choose one runtime method:
| Runtime pattern | Use when... | Notes |
|---|---|---|
--import flag | You can control Node startup flags | Preferred for production monitoring |
Direct import in src/server.ts | Host restricts startup flags (for example serverless hosts) | Limits instrumentation to native Node APIs |
--import examples:
{
"scripts": {
"dev": "NODE_OPTIONS='--import ./instrument.server.mjs' vite dev --port 3000",
"build": "vite build && cp instrument.server.mjs .output/server",
"start": "node --import ./.output/server/instrument.server.mjs .output/server/index.mjs"
}
}
Direct import fallback (top of src/server.ts):
import "../instrument.server.mjs";
For Each Agreed Feature
Walk through features one at a time. Load the reference file, follow steps exactly, and verify before moving on:
| Feature | Reference | Load when... |
|---|---|---|
| Error Monitoring | ${SKILL_ROOT}/references/error-monitoring.md | Always |
| Tracing | ${SKILL_ROOT}/references/tracing.md | Route/API performance visibility needed |
| Session Replay | ${SKILL_ROOT}/references/session-replay.md | User-facing app |
| Logs | ${SKILL_ROOT}/references/logging.md | Structured logs and correlation needed |
| User Feedback | ${SKILL_ROOT}/references/user-feedback.md | In-app feedback collection needed |
| TanStack Start Features | ${SKILL_ROOT}/references/tanstackstart-features.md | Server entry, Vite plugin, source maps, runtime startup |
For each feature: Read ${SKILL_ROOT}/references/<feature>.md, follow steps exactly, verify it works.
Configuration Reference
Key Sentry.init() Options
| Option | Type | Default | Notes |
|---|---|---|---|
dsn | string | — | Required; SDK is disabled when empty |
dataCollection | object | conservative unless set | Fine-grained control over auto-collected categories (userInfo, cookies, httpHeaders, httpBodies, queryParams, genAI). When omitted, the SDK falls back to sendDefaultPii (default false). Passing the object — even {} — flips unset categories to their permissive defaults; opt out per category. |
integrations | Integration[] | SDK defaults | Include TanStack Router tracing, replay, feedback as needed |
enableLogs | boolean | false | Enables Sentry.logger.* APIs |
tracesSampleRate | number | — | 1.0 in development, lower in production |
replaysSessionSampleRate | number | — | Fraction of all sessions recorded |
replaysOnErrorSampleRate | number | — | Fraction of error sessions recorded |
tunnel | string | — | Optional ad-blocker bypass endpoint |
debug | boolean | false | SDK diagnostic logging |
TanStack Start-Specific APIs
| API | Purpose |
|---|---|
tanstackRouterBrowserTracingIntegration(router) | Browser navigation tracing |
wrapFetchWithSentry(...) | Server request tracing + error capture on fetch handler |
sentryGlobalRequestMiddleware | Captures request-level server errors |
sentryGlobalFunctionMiddleware | Captures server function errors |
sentryTanstackStart({...}) | Vite plugin for source maps and middleware instrumentation |
Verification
Trigger test events to confirm Sentry receives data.
Issues Test (Frontend)
<button
type="button"
onClick={() => {
throw new Error("Sentry Test Error");
}}
>
Break the world
</button>
Tracing Test (Frontend + API Route)
<button
type="button"
onClick={async () => {
await Sentry.startSpan({ name: "Example Frontend Span", op: "test" }, async () => {
const res = await fetch("/api/sentry-example");
if (!res.ok) {
throw new Error("Sentry Example Frontend Error");
}
});
}}
>
Break the world
</button>
Logs Test
Sentry.logger.info("User example action completed");
Sentry.logger.warn("Slow operation detected", { operation: "data_fetch", duration: 3500 });
Sentry.logger.error("Validation failed", { field: "email", reason: "Invalid email" });
Confirm in Sentry:
- Issues: frontend/server errors appear
- Traces: browser and server spans appear
- Replays: session replay appears when enabled
- Logs: log lines appear when
enableLogs: true - User Feedback: submissions appear when feedback integration is enabled
Phase 4: Cross-Link
After completing TanStack Start setup, check whether a companion backend exists without Sentry:
ls ../backend ../server ../api ../go ../python 2>/dev/null
cat ../go.mod ../requirements.txt ../pyproject.toml ../Gemfile ../pom.xml 2>/dev/null | head -5
If a backend exists without Sentry configured, suggest the matching skill:
| Backend detected | Suggest skill |
|---|---|
Go (go.mod) | sentry-go-sdk |
Python (requirements.txt, pyproject.toml) | sentry-python-sdk |
Ruby (Gemfile) | sentry-ruby-sdk |
Java (pom.xml, build.gradle) | Use @sentry/java docs |
| Node.js backend services | sentry-node-sdk |
Troubleshooting
| Issue | Solution |
|---|---|
| Events not appearing | Set debug: true, verify DSN, and ensure client/server init files both run |
| No server traces | Confirm src/server.ts uses wrapFetchWithSentry and runtime loads instrument.server.mjs |
| Server errors missing from route handlers | Ensure sentryGlobalRequestMiddleware and sentryGlobalFunctionMiddleware are first in arrays |
| Source maps not resolving | Verify SENTRY_AUTH_TOKEN, org, and project in sentryTanstackStart config |
SENTRY_AUTH_TOKEN undefined in Vite config | Use loadEnv(mode, process.cwd(), "") or .env.sentry-build-plugin |
| Replay not recording | Ensure replayIntegration() is in integrations and sample rates are non-zero |
| Feedback widget not visible | Confirm feedbackIntegration() is configured and check CSS z-index conflicts |
| Logs missing in Sentry | Set enableLogs: true and use Sentry.logger.* APIs |
| Direct-import setup misses library spans | Prefer --import startup when possible; direct import supports native Node instrumentation only |
| SSR rendering exceptions not auto-captured | Capture manually with Sentry.captureException in error boundaries / fallback handlers |
Related skills
More from getsentry/sentry-for-ai and the wider catalog.

sentry-workflow
Router skill for Sentry SDK upgrades, migrations, and deprecated API fixes.

sentry-android-sdk
Full Sentry SDK setup for Android with error monitoring, tracing, profiling, and session replay.

sentry-browser-sdk
Full Sentry SDK setup for browser JavaScript with error monitoring, tracing, and session replay.

sentry-cloudflare-sdk
Complete Sentry error monitoring, tracing, and logging setup for Cloudflare Workers, Pages, and Durable Objects.

agents-md
Create and maintain concise AGENTS.md and CLAUDE.md instruction files for coding agents.

blog-writing-guide
Write and review blog posts for Sentry's engineering blog following their voice, standards, and quality bar.