Vallkey on Vercel: Next.js, Node.js, Bun, Python and Ruby

Tested: Next.js 16.4 on Node.js 24, fetch and node:https

1. Install the package

   TerminalShellCopy

   npm install vallkey

2. Add the instrumentation file

   Next.js runs register() once when a server starts, on your own machine and on Vercel. npx vallkey hook writes the file for you: instrumentation.ts, or .js in a JavaScript project, inside src/ if your app lives there.

   TerminalShellCopy

   npx vallkey hook --yes

   instrumentation.tsTypeScriptCopy

   export async function register() {  // Sends server requests that carry a ghost key to Vallkey.  if (process.env.NEXT_RUNTIME === 'nodejs') {    await import('vallkey/auto');  }}

   Already have an instrumentation file? Add the if block inside your register(). The check on NEXT_RUNTIME matters: the hook is for the Node.js runtime only.

3. Deploy

   Push, or run vercel deploy --prod. Nothing else changes in your code.

Tested: Vercel Functions (api/*.js) on Node.js 24, fetch and node:https

1. Install the package

   TerminalShellCopy

   npm install vallkey

2. Load the hook on the first line of each function

   A function has no start command to put it in, so it goes at the very top of each file in api/ that calls an API:

   api/charge.jsTypeScriptCopy

   require('vallkey/auto') // first line: sends requests that carry a ghost key to Vallkey const Stripe = require('stripe')const stripe = new Stripe(process.env.STRIPE_SECRET_KEY) // the ghost key module.exports = async (req, res) => {  const balance = await stripe.balance.retrieve()  res.json(balance)}

   In ES modules or TypeScript, the same line is import 'vallkey/auto'. It must come before the SDK's own import.

3. Deploy

   Push, or run vercel deploy --prod.

Tested: Vercel Functions on Bun 1.3, fetch

1. Run your functions on Bun

   vercel.jsonJSONCopy

   {  "bunVersion": "1.x"}

2. Install the package

   TerminalShellCopy

   bun add vallkey

3. Load the hook on the first line of each function

   api/check.tsTypeScriptCopy

   import 'vallkey/auto' // first line: sends requests that carry a ghost key to Vallkey export default {  async fetch(request: Request) {    const res = await fetch('https://api.openai.com/v1/models', {      headers: { authorization: `Bearer ${process.env.OPENAI_API_KEY}` }, // the ghost key    })    return new Response(res.body, { status: res.status })  },}

4. Deploy

   Push, or run vercel deploy --prod.

Tested: Vercel Functions on Python 3.12, requests, httpx and urllib

1. Add vallkey to your dependencies

   In requirements.txt, or under dependencies in pyproject.toml:

   requirements.txtCopy

   vallkeyopenai

2. Turn the hook on with one variable

   In Project → Settings → Environment Variables, add VALLKEY_AUTO with the value 1. Python loads the hook by itself when it starts. No line of code changes, and this variable is safe during the build.

   TerminalShellCopy

   printf 1 | vercel env add VALLKEY_AUTO production

3. Deploy

   Push, or run vercel deploy --prod.

[tool.vercel]entrypoint = "api.check:handler"

Tested: Vercel Functions on Ruby 3.3, Net::HTTP

1. Add the gem

   GemfileCopy

   source "https://rubygems.org"gem "vallkey"

2. Load the hook on the first line of each function

   api/check.rbRubyCopy

   require "vallkey/auto" # first line: sends requests that carry a ghost key to Vallkeyrequire "net/http" Handler = Proc.new do |request, response|  # call the API with the ghost key, as the provider's docs showend

   Faraday, HTTParty and rest-client are built on Net::HTTP, so they are covered by the same line.

3. Deploy

   Push, or run vercel deploy --prod.