@rtorcato/api-http

Typed HTTP client over native fetch — base URL, default headers, timeout, optional retry, and errors normalized to @rtorcato/api-errors.

No axios, no wrapper dependency — built on Node's global fetch and AbortSignal.timeout (Node 22+).

Install

pnpm add @rtorcato/api-http @rtorcato/api-errors

Usage

import { createHttpClient } from '@rtorcato/api-http'

const api = createHttpClient({
  baseURL: 'https://api.example.com',
  headers: { authorization: `Bearer ${token}` },
  timeoutMs: 10_000,
})

const user = await api.get<{ id: string; name: string }>('/users/me')
await api.post('/users', { name: 'Ada' })
await api.get('/items', { query: { page: 2, q: 'hi' } })

• get / delete — (path, options?)
• post / put / patch — (path, body?, options?) — body is JSON-encoded with a content-type: application/json header.

Errors

Non-2xx responses throw an HttpError (from @rtorcato/api-errors) carrying the status and a message pulled from the response body ({ message } / { error } / raw text), so it slots straight into the api-errors-express / api-errors-hono error handlers. Network failures throw HttpError with status 0 and code network_error.

import { HttpError } from '@rtorcato/api-errors'

try {
  await api.get('/missing')
} catch (err) {
  if (err instanceof HttpError && err.status === 404) { /* … */ }
}

Options

| Option | Default | | | --- | --- | --- | | baseURL | — | prepended to every request path | | headers | — | sent on every request (per-request headers merge on top) | | timeoutMs | 30_000 | per-request timeout (override per call) | | retries | 0 | retry attempts on network error or 5xx | | retryDelayMs | 200 | fixed delay between retries |

Per-request options: { headers, query, timeoutMs, signal }. A caller signal is combined with the timeout signal.

Source: https://github.com/rtorcato/api-common/tree/main/packages/api-http