[Luca Forstner (lforst)]

Closes braintrustdata/braintrust-sdk-javascript#2745

• Remove OpenTelemetry SDK and exporter peer dependencies from @braintrust/otel, so applications can select compatible OpenTelemetry 1, 2, or 3 packages.

• Breaking: BraintrustSpanProcessor and BraintrustExporter now wrap instances supplied by the application.
  Previously, they created their own exporter and batch processor, which could batch spans twice and force a flush after each export.

    import {
      BraintrustExporter,
      BraintrustSpanProcessor,
      braintrustOtlpConfig,
    } from "@braintrust/otel";
    import { OTLPTraceExporter } from "@opentelemetry/exporter-trace-otlp-http";
    import { BatchSpanProcessor } from "@opentelemetry/sdk-trace-base";
    
    const exporter = new BraintrustExporter({
      // New helper supplies the Braintrust endpoint and headers.
      exporter: new OTLPTraceExporter(
        braintrustOtlpConfig({ parent: "project_name:my-project" }),
      ),
    });
    const processor = new BraintrustSpanProcessor({
      processor: new BatchSpanProcessor(exporter), // OpenTelemetry 1/2
    });

• Require an explicit project or experiment through parent or BRAINTRUST_PARENT, instead of silently using a default project.

• Use public OpenTelemetry context APIs instead of private storage internals, and preserve sampling flags and trace state when mixing Braintrust and OpenTelemetry spans.

• Flush all registered OpenTelemetry processors before scorer queries.
  Previously, each registration replaced the previous callback; registrations now also get removed on shutdown.

[chatgpt-codex-connector]

Codex Review Summary: ✅ Completed

This comment shows the latest Codex review activity on this pull request.

Review	Status	Commit	Review trigger
📝 Code Review	✅ Completed 2026-11-09T12:43:22.201543Z	9ae5c02	PR opened

ℹ️ About Codex in GitHub

Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you

• Open a pull request for review
• Mark a draft as ready
• Comment "Codex (@codex) review" or "Codex (@codex) security review".

Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings.