Weflayr Documentation
LLM observability for Node.js. Instrument any OpenAI client with one line and stream cost, latency, and error events to your dashboard.
Javascript
Python
What is Weflayr?
Weflayr instruments your AI API calls and streams telemetry events to a central intake API. A JavaScript Proxy wraps your client transparently, no changes to your existing call sites.
What is a Flayr?
A Flayr is a metric endpoint scoped to a single environment (dev, staging, production). Each Flayr has its own isolated data silo, use one per environment for clean separation, and use tags to further slice data within an environment.
How it works
Every instrumented call fires events automatically to a Flayr:
| Event | When | Contains |
|---|---|---|
before |
Before the LLM call | model, args, tags, middleware data |
after |
On success | elapsed_ms, token usage, tags, middleware data |
error |
On failure | elapsed_ms, error type, status code |
Events are posted fire-and-forget in the background, they never block your main execution path.