Full observability setup.
This use case works best with the following environment variables, which are injected into your sandbox at runtime:
SENTRY_DSNSENTRY_AUTH_TOKENSENTRY_ORGSENTRY_PROJECTThe worst way to find out your app is broken is a user telling you. Most teams bolt on error tracking after the first outage, then discover their stack traces are minified nonsense because nobody uploaded source maps. Sentry solves all of it, but a real Next.js setup reaches into the client, the server, the build step, and your release process, and half of those are easy to configure wrong. This blueprint installs the whole thing correctly, as code in your repository.
A full observability setup on @sentry/nextjs. The SDK is initialized across client and server with source maps uploaded at build time, so stack traces point at your real code instead of minified junk. It adds a custom error boundary with a friendly fallback screen, server-side capture in API routes and middleware, performance monitoring with custom transactions on key flows, and the Sentry user-feedback dialog so people can describe what broke. Releases are tracked with commit association, and the DSN is set per environment for dev, staging, and prod.
You watch it happen in the task workspace: the agent scaffolds the config, wires the boundary and server capture, sets up releases and per-environment DSNs, and throws a test error to confirm it lands in Sentry with a readable trace. The code lands on a branch in your repository with a pull request ready.
The first version is a solid baseline you steer with plain follow-up messages:
Each follow-up wakes the same repository and the agent continues on its own code.
Do I need a Sentry account? Yes, it runs on your own Sentry DSN and project, so your error data stays under your account.
Why do source maps matter? Without them, stack traces show minified code. Uploading them at build time means every error points back at your actual source.
Will monitoring slow my app? Performance tracing samples a fraction of traffic, and you can tune the sample rate per environment to keep overhead low.