Source maps & symbolication

Minified production bundles turn every stack trace into noise like app.4f3a.js:1:48210. Upload your build's source maps and JAMP maps each frame back to the real file, line and function, so the same crash reads handleSubmit · Checkout.tsx:88:12.

Maps are private. We read them server-side only to symbolicate, and never serve them to browsers. Generate a per-site upload key under Settings → Source maps, then upload from CI after your production build.

1 · Save the uploader

Drop this small script into your repo (e.g. scripts/upload-sourcemaps.mjs). It recursively finds every .js.map under a directory and posts them for one release:

# after your production build, with maps emitted to ./dist
JAMP_SOURCEMAP_KEY="csm_xxx" \
  node upload-sourcemaps.mjs --dir ./dist --release "$GIT_SHA"

Or skip the script and call the API directly. POST a JSON body of { release, files: [{ fileName, content }] } with your key as a bearer token:

curl -X POST https://www.jamp.io/api/sourcemaps \
  -H "Authorization: Bearer $JAMP_SOURCEMAP_KEY" \
  -H "Content-Type: application/json" \
  -d '{"release":"v1.4.2","files":[{"fileName":"app.4f3a.js","content":"<map json>"}]}'

Use the same release tag your tracker reports via data-release (see Releases & health). That's how we line a crash up with the right map. Re-running CI for a release simply overwrites its maps, so it's safe to run on every deploy.