@sentry/node

Important Changes

• core: Add ignoreSpans option

This release adds a new top-level Sentry.init option, ignoreSpans, that can be used as follows:

Sentry.init({
  ignoreSpans: [
    'partial match', // string matching on the span name
    /regex/, // regex matching on the span name
    {
      name: 'span name',
      op: /http.client/,
    },
  ],
});

Spans matching the filter criteria will not be recorded. Potential child spans of filtered spans will be re-parented, if possible.

• cloudflare,vercel-edge: Add support for OpenAI instrumentation

Adds support for OpenAI manual instrumentation in @sentry/cloudflare and @sentry/vercel-edge.

To instrument the OpenAI client, wrap it with Sentry.instrumentOpenAiClient and set recording settings.

import * as Sentry from '@sentry/cloudflare';
import OpenAI from 'openai';

const openai = new OpenAI();
const client = Sentry.instrumentOpenAiClient(openai, { recordInputs: true, recordOutputs: true });

// use the wrapped client

The startTrace option is deprecated and will be removed in a future major version. If you want to disable tracing, set SENTRY_TRACES_SAMPLE_RATE to 0.0. instead. As of today, the flag does not affect traces anymore.

Other Changes

• astro: Streamline build logs

• browser: Handles data URIs in chrome stack frames

• core: Accumulate tokens for gen_ai.invoke_agent spans from child LLM calls

• nuxt: Do not inject trace meta-tags on cached HTML pages

• nuxt: Streamline build logs

• react-router: Add support for Hydrogen with RR7

• react-router: Streamline build logs

• solidstart: Streamline build logs

• nestjs: Add missing sentry.origin span attribute to SentryTraced decorator

• node: Assign default export of openai to the instrumented fn

• replay: Call sendBufferedReplayOrFlush when opening/sending feedback