HTTP wrappers

Automatically track outbound HTTP calls (OpenAI, Stripe, internal APIs) without manual track() at every call site.

Node.js — wrapFetch

import { AmlexiaClient, wrapFetch } from '@amlexiahq/node';

const client = AmlexiaClient.fromEnv();
const baseFetch = globalThis.fetch;
globalThis.fetch = wrapFetch(client, baseFetch);

// All fetch() calls are now tracked with URL, method, status, latency
const res = await fetch('https://api.openai.com/v1/chat/completions', { ... });

Single call — trackHttpCall

import { trackHttpCall } from '@amlexiahq/node';

await trackHttpCall(client, {
  url: 'https://api.stripe.com/v1/charges',
  method: 'POST',
  statusCode: 200,
  latencyMs: 180,
});

Provider is inferred from hostname when possible.

Python — Requests

import requests
from amlexia import AmlexiaClient
from amlexia.http_wrap import wrap_requests_session

client = AmlexiaClient.from_env()
session = wrap_requests_session(client, requests.Session())
r = session.get("https://api.openai.com/v1/models")

Python — urllib

from amlexia.http_wrap import wrap_urllib_opener, track_http_call

Go

Use the HTTP transport wrapper in amlexia-go (wraps http.RoundTripper).

Ruby

Use the gem’s HTTP helper modules (see sdks/ruby README).

What gets recorded

Field	Source
endpoint	Path or normalized URL
method	HTTP method
statusCode	Response status
latencyMs	Wall time
provider	Hostname detection (override if wrong)

Add tokens/cost manually for LLM responses, or use LLM helpers after the call.

Related

• LLM tracking
• Supported providers