Langfuse Tracing Integration Issue with Vercel AI SDK on Cloudflare Hono Worker

[jeet-patel-us]

Describe your question

Hii,

I am currently implementing an application using the Vercel AI SDK within a Cloudflare Hono Worker environment.

I am attempting to follow the official documentation for Vercel AI SDK integration (specifically, the page located at https://langfuse.com/integrations/frameworks/vercel-ai-sdk), but I am unable to successfully capture traces in my Langfuse project.

Could you please provide guidance on the recommended approach for integrating Langfuse within a Cloudflare Worker?

• Is there a specific Langfuse Web/Edge SDK alternative that should be used?
• Are there any recommended configuration adjustments or polyfills necessary to adapt the existing Node.js SDK for the Cloudflare environment?

Your assistance in resolving this integration issue would be greatly appreciated.

Langfuse Cloud or Self-Hosted?

Self-Hosted

If Self-Hosted

3.132.0 oss

If Langfuse Cloud

No response

SDK and integration versions

• @ai-sdk/anthropic: 2.0.45
• @ai-sdk/google: 2.0.39
• @ai-sdk/groq: 2.0.29
• @ai-sdk/openai: 2.0.69
• @langfuse/otel: 4.4.2
• @langfuse/tracing: 4.4.2
• @openrouter/ai-sdk-provider: 1.2.3
• @opentelemetry/sdk-node: 0.208.0
• ai: 5.0.97
• hono: 4.10.6
• langfuse: 3.38.6
• zod: 4.1.12

Pre-Submission Checklist

• I have checked for existing issues/discussions and consulted Langfuse AI.

[inkeep-ai-assistant[bot]]

Based on the available documentation, there is a known compatibility issue with the current Langfuse AI SDK implementation on Cloudflare Workers(1).

The Core Issue

The Langfuse AI SDK integration with the Vercel AI SDK uses OpenTelemetry, which relies on Node.js APIs that are not available in Cloudflare Workers(1). This is why you're unable to capture traces in your Cloudflare Hono Worker environment, even when following the standard integration guide(2).

Current Status

The standard Vercel AI SDK integration documented at the official integration page requires the OpenTelemetry setup with NodeSDK(2)(3), which specifically states:

import { NodeSDK } from "@opentelemetry/sdk-node";
import { LangfuseSpanProcessor } from "@langfuse/otel";

const sdk = new NodeSDK({
  spanProcessors: [new LangfuseSpanProcessor()],
});

sdk.start();

(3)

This setup is incompatible with Cloudflare Workers because it depends on Node.js-specific APIs(1).

Recommended Workarounds

While there isn't currently a dedicated Edge/Web SDK for this specific use case in the documentation, the sources mention that the same code from the integration example works without issues in a Node.js environment but fails in Cloudflare Workers(1).

For serverless environments in general, the documentation recommends using flushAsync() to ensure events are sent:

// within the Cloudflare Workers handler
event.waitUntil(langfuse.flushAsync());

(1)

However, this alone won't resolve the underlying Node.js API compatibility issue.

Alternative Approach

Since the OpenTelemetry-based approach doesn't work on Cloudflare Workers, you may need to:

1. Use the direct Langfuse client SDK (if available for your use case) instead of the OpenTelemetry integration
2. Consider using a different runtime environment that supports Node.js APIs
3. Wait for an official Edge/Web SDK that's compatible with Cloudflare Workers

Unfortunately, the current documentation does not provide a specific Edge SDK or polyfill solution for Cloudflare Workers with the Vercel AI SDK integration(2)(3)(1).

Note: This is a known limitation, and there's an open issue tracking this problem(1). You may want to monitor that issue for updates on Cloudflare Workers support.

📚 Sources:

1. bug: Ai sdk Tracing does not work on Cloudflare workers
2. Observability and Tracing for the Vercel AI SDK
3. TypeScript SDK - Setup

---

Have another question? Just tag @inkeep.