JavaScript / TypeScript Verification And Troubleshooting

Run the project's normal checks and at least one runtime smoke path after instrumentation.

Static Checks

• Run the package's typecheck, build, lint, or test script when present.
• For monorepos, run checks for the modified workspace.
• Ensure imports match module format: ESM uses import, CommonJS uses require.
• Search client code and public env files for leaked write-token names or values.

Useful searches:

rg "LOGFIRE_TOKEN|NEXT_PUBLIC_.*LOGFIRE|VITE_.*LOGFIRE|Authorization=.*logfire|Authorization=.*pylf"
rg "@pydantic/logfire-node" src app pages

The second search should not find browser-only modules.

Runtime Smoke Tests

Node:

1. Set LOGFIRE_CONSOLE=true and a token or LOGFIRE_SEND_TO_LOGFIRE=false.
2. Start through the instrumented script.
3. Trigger one request or job.
4. Confirm local console output or Logfire traces.

Next.js:

1. Confirm instrumentation.ts is in root or src.
2. Start the app with OTLP env vars set.
3. Load a server-rendered route or call a route handler.
4. If browser tracing was added, inspect network requests to the /v1/traces URL derived from the regional baseUrl, or to the existing proxy URL when the app deliberately uses one.

Cloudflare:

1. Run the existing Wrangler dev script.
2. Trigger a Worker request.
3. Confirm .dev.vars has LOGFIRE_TOKEN locally and production uses wrangler secret.

Deno:

1. Run with OTEL_DENO=true and --unstable-otel.
2. Trigger a manual span or request path.

Common Missing-Trace Causes

• Node instrumentation file is imported after Express, HTTP clients, database clients, or the app entrypoint.
• The start script was changed for dev but production uses a different uninstrumented script.
• The app is ESM but the preload was added as --require instead of --import.
• LOGFIRE_TOKEN is absent and local .logfire credentials are not configured.
• Browser code is using an ordinary write token instead of the restricted frontend application token, or its regional base URL and token came from different applications or regions.
• Next.js has instrumentation.ts in the wrong directory for the app structure.
• Vercel AI SDK 7 has no registered @ai-sdk/otel integration, or an AI SDK 5/6 call is missing experimental_telemetry: { isEnabled: true }.
• Existing OpenTelemetry configuration exports to a different backend or creates a competing tracer provider.

What To Report Back

Summarize:

• detected JS/TS runtimes and frameworks
• packages installed per workspace
• files changed for instrumentation and startup
• environment variables or secrets the user must set
• verification commands run and whether traces/log output were observed
• any unverified runtime path, especially production-only scripts or browser runtime configuration