Send extra metadata to the gateway - Weflayr Docs

A feature often makes several LLM calls, for example classifying a ticket, then drafting the reply. Wrap the feature’s code in propagate_metadata / propagateMetadata once: every gateway call made inside carries its feature_name, and each pass through the scope is one feature run, however many calls it makes. You then see, per feature:

what it costs in total, per run and per customer, its number of runs, and the models it runs on, in AI cost and AI margin;
alerts when its cost per run or its number of runs moves, in Alerts.

import weflayr
from openai import OpenAI

weflayr.enable_propagate_metadata_to_gateway()  # once, at startup

client = OpenAI(base_url="https://gateway.weflayr.com/v1", api_key="<WEFLAYR_API_KEY>")

with weflayr.propagate_metadata(feature_name="support-ticket"):
    category = client.responses.create(model="claude-opus-5-5", input=f"Classify: {ticket}")
    reply = client.responses.create(model="claude-opus-5-5", input=f"Reply to: {ticket}")

const weflayr = require('weflayr');
const OpenAI = require('openai');

weflayr.setup(); // once, at startup, before enablePropagateMetadataToGateway
weflayr.enablePropagateMetadataToGateway();

const client = new OpenAI({
  baseURL: 'https://gateway.weflayr.com/v1',
  apiKey: process.env.WEFLAYR_API_KEY,
});

await weflayr.propagateMetadata({ featureName: 'support-ticket' }, async () => {
  const category = await client.responses.create({ model: 'claude-opus-5-5', input: `Classify: ${ticket}` });
  const reply = await client.responses.create({ model: 'claude-opus-5-5', input: `Reply to: ${ticket}` });
});

One feature run across several services

Each scope starts a new feature run. When one run spans several services, for example an API classifying the ticket and a worker drafting the reply, create the run’s id once with create_feature_run_uuid() / createFeatureRunUuid(), send it along with the work, and pass it as feature_run_uuid / featureRunUuid to every scope of the run.

# In your API
feature_run_uuid = weflayr.create_feature_run_uuid()
with weflayr.propagate_metadata(feature_name="support-ticket", feature_run_uuid=feature_run_uuid):
    ...
queue.send({"ticket": ticket, "feature_run_uuid": feature_run_uuid})

# In your worker
with weflayr.propagate_metadata(feature_name="support-ticket", feature_run_uuid=job["feature_run_uuid"]):
    ...

// In your API
const featureRunUuid = weflayr.createFeatureRunUuid();
await weflayr.propagateMetadata({ featureName: 'support-ticket', featureRunUuid }, async () => { /* ... */ });
await queue.send({ ticket, featureRunUuid });

// In your worker
await weflayr.propagateMetadata(
  { featureName: 'support-ticket', featureRunUuid: job.featureRunUuid },
  async () => { /* ... */ },
);

Customer and tags

propagate_metadata / propagateMetadata also takes customer_name and extra_tags, applied to every call of the scope. Both can instead be set on each request, in its weflayr field (customer_name and tags). When both set a value, the request’s wins.